Ask HN: If the models are so good, why doesn't OpenAI just start 100 companies?
Title is a bit of an exaggeration but the question is more: if the frontier models are so good, why even sell them as your primary money maker?
Couldn't OpenAI or Anthropic make a lot more money by just starting companies themselves that leverage the models and building their own products?
Is selling tokens really the best strategy?
Especially with commodification and prices declining wouldn't that indicate there is more money to be made on the application-side of the models rather than the utility side
The sort of obvious answer is the same reason McKinsey hasn't taken over the Fortune 500 despite consulting and having deep knowledge of the operations of most of those companies.
AI is brains. Companies need arms and legs and relationships.
The reality is that employees are also mostly not entrepreneurs, same as consultants.
There are trillions of company ideas in those tokens, just like there are trillions of shots in the fans head watching the basketball players.
Takes years, patience, capital and the right mindset to get to the point to even take the shot.
I think this is true. I also suspect there’s aspects of human intelligence not yet replicated, but it’s not clear what those are yet.
It's the relationships. Most sales deals that bring in the money are still done face to face.
I don't really know if you could do multimillion dollar deals on a web form.
Exactly. Factories will be built, but you will still need workers, project engineers, and actual architects. Automation is still not there yet, despite the elaborate harnesses of frontier models.
Given the primary goal and promise of all these AI vendors is to replace workers I think its hilarious that "you still need workers" is an argument. Wouldn't we have an abundance of workers soonish as all the replaced people could be retrained? Or is the truth that these models can't actually replace intellectual work?
you forget that these replaced workers need to be working to be able to consume the produced goods. if they are all out of a job no one will buy anything and these companies wont get off the ground.
I am well aware, people need work, therefor my retrain argument. Unless companies also do not want to train workers, which is a huge issue already.
Retrain for what?
Not asking smart questions about the corporate excuses, that's what. Back to the bread line with you!
Well the argument was "we still need workers" so surely there's something to train people for!
The argument was "people still need work to have an income" not "we still need workers"
Is it surprising there’s a gap between their vision and reality? Obviously if all that comes to pass they’re hoping the societal implications get figured out to keep the pitchforks at bay.
FWIW no I don’t believe models can replace intellectual work because intelligence is not purpose, and no other entity can tell us what we want. While AI has made building cheap and cutting through the noise even harder, it hasn’t made deciding what to be build any easier, and personally I felt that was already the bottleneck a decade ago (think bullshit jobs, App Store saturation, etc).
Of course that’s just one perspective and I could be totally wrong. But given the situation it makes perfect sense for the big labs to stay focused on building the most valuable shovels in the world and leave the business risk and finding the money for the cash furnaces to others.
> no other entity can tell us what we want
That's the marketing department's job. At some point, there will be an attempt at modeling that, too.
We live in a time and place where the only way you can truly profit is to cut costs. Selling more of the thing that you make or do to more people is harder because of population trends. Everything that can be modeled in AI and done by it, will be.
If I open a restaurant that doesn't serve anything because "robot chefs and waiters will be here soon" then in a few months I won't have a restaurant or any money.
If every semi-plausible promise of the AI vendors could be believed, every job which converts conversations into binaries could be replaced.
That's a large chunk of the software development industry, but practically nothing else.
So they promise robots, but have remarkable difficulty with even the most limited domain autonomous robots: Amazon claims to have deployed over a million robots, but is still continuously hiring in their warehouses with requirements like these:
Be 18 years or older
Understand and follow instructions in the required language
Lift up to 49 pounds
Push utility carts up to 60 pounds
Stand, walk, push, pull, squat, bend, and reach during shifts
Use carts, dollies, hand trucks, and other gear to move items around
Go up and down stairs (where applicable)
Maybe Amazon shouldn't have built anti-Dalek defenses into their warehouses.
Most of those robots were designed and built years before "real" AI was a thing. Big companies move like freight trains. Hard to start, and nearly impossible to stop once they get going.
Give it 5 years and see how strong those anti-Dalek defenses are.
> Maybe Amazon shouldn't have built anti-Dalek defenses into their warehouses.
https://www.gatech.edu/news/2026/04/01/why-iran-targeted-ama...
https://arstechnica.com/gadgets/2026/09/iran-strikes-on-amaz...
A lack of anti drone defenses turned out to be an actual problem - also interesting use case for your idea of durability to be judged against.
We got interstates the day after we invented the ICE automobile, right?
Yall really dont see the issue about replacing workers step by step? We'll reach the point of a "saturated" job market which will cause people to fall into welfare programs. Those are saturated soon as well, so what are people going to do, if they have no food and no help? Doesn't take a genius to see the dystopic near future.
Thinking our governments will solve the issues the AI vendors created ain't going to work.
The only option is to invest in robots with machine guns, to provide security services to the ultrarich.
I'd like to imagine that a factory from 1880s looked a lot different than today, with IoT and robot arms and whatnot. Intellectual work will still be needed, always will. Just think about "computers" of the Space Race era. Their whole job was to sit down with a calculator and calculate the engineering approximations for an orbit that takes a rocket to the moon. Then they installed IBM monoliths and suddenly you needed a few computer scientists only, plus mechanics for overall maintenance of the machine and utilities. And noone ever stopped teaching electronics at all. In fact, that kick-started the Moore's law era.
So we wait 150 years for AI to get good?
Say you're Matthew Boulton in 1775 who partnered with James Watt when he started popularizing his imrpoved steam engine. Could you have predicted that in 2026 we will be discussing about him here, on the Internet, (probably) on a smartphone? Putting down a number is the wrong discussion imho.
Best response I've ever seen to this. Well said.
In addition to this, most of the VC funds investing in OpenAI do have industry vertical plays.
The Firm is confederated.
The pyramid needs a top!
Pretty weak take here tbh. If the promise of AI bore any fruit market forces would immediately shift customers to cheapest offering purely because the prices would be so much cheaper.
Same reason why Apple hasn't Sherlocked any applications sold for their platform!
Oh wait.
McKinsey is just happy to rent out shovels.
I have never heard anyone claim that McKinsey is a purveyor of quality shovels. My impression is that they're corporate parasites who contribute little of value.
McKinsey and the others are good at selling solutions to a C-Suite that keep those jobs, because they're seen as the client, whole stripping out the workers who actually add value. The consulting does it's job, secures the decisions that protect the bosses, and in doing so often ruin the companies they're there to "help".
Or build presentations for companies on how they could implement an automated digging system to increase efficiency.
The real answer is that modern businesses depend very strongly on who owns the portal to a business (gatekeeping), and who owns brands, etc.
Or the pesky fact that McKinsey has a long history of given terrible advice to companies that, if and when implemented, caused them much pain. So there’s little evidence McKinsey folks would actually be good at building and running businesses in most sectors.
The former Starbucks CEO is a classic example. His infamous interview with Cramer on CNBC is the textbook of what it looks like when some McKinsey consultant gets actually put in charge. It was a total shitshow.
Likely succeeded at pontificating to CEOs via PPT, but actually being CEO? Not so much.
> So there’s little evidence McKinsey folks would actually be good at building and running businesses in most sectors.
Except there is! On average, McKinsey-lead businesses perform better: https://archive.is/HLA3s
The article cites no real evidence that stock price changes of a small group of companies had anything to do with where the CEO worked previously and the article’s suggested next read in the Economist article is “How McKinsey Lost It’s Edge” which goes against the very point you’re trying to make.
It opens with the statement “As it nears 100, is the world’s most illustrious consultancy past its prime?” and cited poor performance associated with its advisors. So your “citation” isn’t really helping the point you’re trying to make.
> The former Starbucks CEO is a classic example. His infamous interview with Cramer on CNBC is the textbook of what it looks like when some McKinsey consultant gets actually put in charge. It was a total shitshow.
Do you have a link to the specific interview you mention? Google seemingly doesn't know of an infamous one.
https://youtu.be/zaPJWgbliQ8?is=iqmcmYNEIdR2JLYt
Absolute car crash interview that eventually led the board firing him.
Maybe companies should hire McKinsey to know what NOT to do
Less wrong is a step towards more right.
Another case in point: Nike, which is now frantically trying to dig itself out of a massive hole created because they hired a bean-counting former CEO of Bain to be the CEO and made absolutely terrible decisions that demonstrated his lack of understanding of Nike's business and culture. (he got a golden parachute)
No, the model relies on human brains (at least initially some would argue). The idea is that a professional, say a physician, will bring their existing knowledge AND reputation. It's a partnership model, where the physician adds a secret sauce that distinguishes it from the rest of the otherwise commodity agents that have almost no distinguishing factor from baseline, the humans are the seed in the prng.
Also the reputation is not minor, the human can take out loans, or take a withdrawal out from their social capital by telling people that they are AI enabled and selling AI to other (presumably) non-AI people.
It's not all just GPUs and Tokens, it can't be, there needs to be some mixing with other resources in order to bake a value cake. While pure token cakes might be very yummy (for some consumers who like pure slop), by mixing GPUs with human brains, a wider space of yummy hybrid cakes can be explored, which even more consumers would enjoy (those who like partial slop).
Even drug manufacturers need retailers who use their personal connections to gain trust, and who put their criminal record on the line.
> AI is brains. Companies need arms and legs and relationships.
Do they? Haven't Google and others been proving that wrong for almost 30 years?
Someone needs to build and maintain the data centers.
Someone needs to go out and sell advertising and services. And when those things are sold, someone has to keep the subscriptions going and growing.
Business is not just person to machine or machine to machine, a lot of it, even those of you that are in pure tech jobs has someone in the field selling to clients and keeping clients engaged.
Google has an army of salespeople.
>AI is brains
Is they though?
dont question computer god!!
If AI is brains then investors are zombies
so… ok, you still can’t eat art right?
I don't think it's similar, McKinsey make money by offering non-sense presentation and helping managers justify their decisions
Actually, that sounds quite similar.
yup that what claude does too
True but at least Claude does the work for me too. It's going to build me whatever I want despite how correct of a choice it is.
Came for this comment
I think that’s the underlying criticism. If the enterprise is risky, speculative, or just straight up bullshit, then selling the shovels is the smartest move.
They would most certainly save money by building said shovels out of paperclips!
And more important, AI companies are selling shovels, not gold.
Others will buy the shovels to dig for gold, most will fail, but the shovels are already sold, and part of the profit will be used to make more shovels until the gold fever settles or dies.
They would never shove the gold machine at you at every opportunity if it really created gold.
McKinsey does not offer actual strategical advice, they offer scapegoats and insurance that when shits does hit the fan, the client will not be the one sent to prison.
The glossy magazines read by C-suite is the slick marketing part, not the actual product.
I think their integration work and consulting is second to none. Companies don’t always know how to stack rank growth opps or M&A
This is a tell you haven't interfaced with consulting firms much before.
When you say [insert firm] is [good/bad] at [thing], what you're actually saying is:
"The team of kids at the firm's Minneapolis office did a good job on this specific project."
There is no special sauce or "process" that could consistently deliver that outcome given consulting firms are just hourly billing body shops with high turnover that rarely hire true subject matter experts (they don't pay well enough). They have no equity in the advice they hand out, they don't suffer consequences if they're wrong absent doing something illegal, so nothing is ever learned other than how to extract more projects from the next lazy corporate bureaucracy.
It's wholly dependent on the kids who were staffed on the project and how insecure/eager they were for approval at the time.
Average tenure at all professional services firms (doesn't matter if its consulting, advertising or accounting) isn't very long, and there's almost zero knowledge sharing between individual offices of the same firm since they are competing for billable hours. So generally it's a crapshoot every time.
What you can consistently be sure of when hiring McKinsey, is that you are offloading responsibility for decision-making to a 3rd party fictional mirage of 'competence' that everyone pretends is real, so you can claim its not your fault if something goes wrong or is unpopular...and keep your cushy big company executive job.
Companies are almost always better off bringing in one experienced consultant that actually knows how to stand up the project and help hire the team to make it happen.
We used to have a very experienced systems engineer that we'd contract whenever a chief engineer had an idea that gained traction with directors. The systems engineer would help layout the structure and timeline of the project, basically identifying how much money and the types of people we'd need. He'd stick around long enough to get the project to the final design review and then go do something else when we started building stuff. By that point, the PM is deeply involved and the team is cohesive and productive. Maybe it's just a coincidence but every big project we've tried to stand up without his help would struggle tremendously in the early stages.
It sounds like you're agreeing with my point; good people do good work, the brand name on the invoice billing you for their time is irrelevant.
That said, I would be highly skeptical of any 'very experienced systems engineer' who took a job at a glorified post-collegiate body shop like McKinsey. At least in the US, given all the opportunities available, it means something has gone horribly wrong (EU/ex-US might be different given it's a dead zone for tech).
I dont think your comment reflects what happens in board rooms and internally at large companies. I am paying for the senior partner to have conversations with me on how others before me have done this. The packaging of a PowerPoint is to be able to efficiently share that data with others in the firm.
Is it smart for individuals to have conversations with others about their problems? This is that for CEOs and senior executives.
> consulting firms [...] rarely hire true subject matter experts (they don't pay well enough).
This may hold in the USA, but clearly not in Germany. If, as a subject matter expert, you do have the choice and can tolerate the high stress and high workload, big consulting firms typically pay much better.
The reasons why true subject matter experts nevertheless often work somewhere else are different:
- As I already mentioned: the consulting environment is very stressful and the workload is huge. This environment is rather not suitable if you have or plan to have a family.
- Consulting companies typically want to hire employees who are smooth salesmen first and subject matter experts second. Many real experts on some topic often rather tend to be more opinionated and "rough around the edges". So they very often don't get hired.
- Very often the daily job working at a consulting company is very different from the topics one is an expert in. Many subject matter experts prefer an environment where they can apply their skills better.
- For many subject matter experts, the topic where they really excel in is very narrow and specialized and very different from the topics where a consulting companies can easily sell "experts" to the customer. So, many subject matter experts don't get hired because their topic of expertise is not easy to sell to customers and/or is too niche.
I’ve seen their proposals for things like new product lines and acquisitions. Are you saying those things don’t exist? Did they use to exist and they just stopped that line of business?
They definitely exist and I'm actually quite convinced they are valuable... but it's a bit like calling Google a technology company because they sell millions Pixel phones and billions in GCP while >90% of their revenue is still advertisement (please don't quote me on the numbers they are surely incorrect, it's to illustrate the idea).
I'm making a provocative shortcut.
Yes, Google's money comes from advertising. But the company that invented, among many other things, map reduce, bigtable, and spanner, seems pretty much a technology company to me.
Bigtable these days handles seven-billion requests per second over several exabytes of data. All developed in house.
The others are similarly impressive.
https://en.wikipedia.org/wiki/MapReduce https://en.wikipedia.org/wiki/Bigtable https://en.wikipedia.org/wiki/Spanner_(database)
I'm not questioning what they invented, I'm highlighting their business model, how they actually make money.
Total or Shell use a lot of state of the art technology to mine, they are still oil companies.
Google is an information technology company. They build information technology, and use it to make money. This is in contrast to an ad agency that uses other people's software to sell ads. That company doesn't create technology.
If an oil company creates new technology to extract oil, then they are (obviously) a technology company. If an oil company outsources all technology development, they are not.
Seems like a pointless semantic argument. Is SpaceX a telecom company, since that’s the bulk of their revenue? Does it matter? Do we have to decide every company is in one and only one category?
Those are simply what the clients wanted to do themselves but they needed someone else to be accountable for their stupid ideas, or to resolve an internal conflict. A good consultant company doesn’t actually provide consultancy, but they’re good at figuring out what the client wants and propose the exact same thing as the solution, so the client can take it to their boss and say “look, …”
McKinsey has multiple consulting verticals, including mechanical engineering, design, employing real professionals for actual work in addition to their traditional business consulting and advising.
I think this comment was just a popularist repetition of discourse from social media. In other words it's a "tiktok ahh take".
I know a lot of people who work or have worked at Accenture. Is that close enough to McKinsey and far enough from social media? Because I have come to a similar conclusion after many long conversations about how the sausage is made.
GP is more cynical than I would have put it, but I think it's directionally correct; it's an easy way for the C-suite to demonstrate "due diligence" on any decision.
Accenture and McKinsey are very different. You can't draw any conclusions about one from knowing the other.
No they're not.
McKinsey would like you to think otherwise of course. They paint a brand image of only doing high-end C-suite strategy work, but ultimately thats a vanishingly small part of the business. Their bread is buttered (just like Accenture) on labor heavy implementation gigs.
Fundamentally, the business of billing out human hours is the same, as are the the incentives. Doesn't matter if its a strategy consulting firm in NYC or an advertising agency in Nepal.
They will all inflate scope, inflate titles/rates, hire mostly cheap new grads, and slowroll-and-expand any engagement.
And they will all take ANY work you give them, regardless of in-house expertise (which they rarely have, given the goal is to pay employees as little as possible to capture margin).
The advertising agency in Nepal will claim to offer strategy consulting to your C-suite, just as McKinsey NYC will happily churn out a powerpoint deck about advertising in Nepal if you're willing to pay them.
Hence why professional services firms all trade at low multiples. It's an undifferentiated commodity business of re-selling human labor in project-based chunks.
Palantir is a professional services firm, at trades very high,
Their valuation comes from the Saas side of the business.
In fact, the bear case on Palantir is always that it's actually a consulting firm masquerading as a Saas firm.
I think it's fair to say that a large portion of their valuation comes from their political ties lending an air of invincibility. If Palantir came under scrutiny it's guaranteed that neither JD Vance nor anyone else Thiel could mobilize would remain silent and let it happen.
They ARE a consulting firm masquerading as SaaS.
They happen to have a SaaS product, but you cannot use it without paying up front for huge amounts of pro services. The SaaS product doesn’t work on its own. This isn’t necessarily bad, but it’s not magic software either.
They’re not consultants, they’re forward deployed engineers!!
It's the same. Source: know people on both sides, hiring the engagement or providing it. Mickey, BCG, Bain, Accenture etc...
I've never heard of McKinsey (I'm not in the US though) but we have a lot of Accenture consultants running around at work. They're generally a waste of space, just like the Deloitte ones we have too.
From my limited experience of working with McKinsey people, I would describe it as a Fortune 500 C-Suite incubator. The hire smart people, they are very well connected and the goal is to rise the ranks and get a CEO position in a multinational. I was told by a very senior ex-McKinsey consultant that their main role is to get people promoted (regional CEO to central board for example). The projects are essentially a vehicle for an individual to move up. Their network is phenomenal. I wasn't very impressed with what i saw productivity wise, but they are polished and the point is not to impress the lower downs.
Accenture are mostly a bunch of low level talentless grunts. There are a lot of them. If you're hiring Accenture then you are in trouble in more ways than one.
McKinsey and Accenture are not the same. Brand-wise that’s Ferrari vs. Lexus.
Lexus has the quality of Toyota behind it, so the analogy doesn’t quite hold.
What has the veneer of sophistication but is actually garbage quality…
Ferrari vs Alfa Romeo, perhaps?
In fairness, this take long predates social media. But you're right that it's exactly the sort of midwit take that gets repeated, whether on social media or by word of mouth.
Does that make it incorrect?
The midwit take in this day and age is assuming you need an adversarial relationship with labor. The problem with McKinsey et al is their point of view.
Personal experience: they get fifty bodies on a two-person job. The work is fine, but not good enough to learn from. Which actually makes sense if you think of their company as a machine that can adapt to whatever problem.
It's the old handyman vs specialized contractor tradeoff (well really it's a tale as old as time; specialists are much more expensive so you want to minimize their use (see healthcare, legal professionals, etc, etc.)) A plumber is a better plumber than your handyman, a cabinetmaker is a better carpenter than your handyman; your handyman changes 1/3 per hour and is available much more frequently. You make the decision who you want to call based on a multitude of factors.
And the added wrinkle is that you need some expertise yourself to make the decision on who to call. That's why lots of people call electricians to change a wall switch or plumbers to install a new faucet. They end up paying 3x what they needed to because they didn't know a handyman can do those jobs very well, and the electricians and plumbers really don't even want those jobs.
I call an electrician to change a wall switch since an unlicensed handyman making a mistake that starts an electrical fire voids my home insurance.
Plumbing is fractionally less regulated where I am and I am allowed to change my own taps washers, but a whole tap assembly falls under similar issues as a power switch.
You can also start a fire by letting your pan get too hot, but nonetheless you make your own eggs.
Because that doesn't void your home insurance, as GP wrote.
Who do you think knows the most about a business? The experts at the leading companies in that business? Or McKinsey? Assuming the sane answer is the experts at the company, why is McKinsey being hired?
How big are those verticals? When it comes to management consulting, they enjoy the Big 3 fame, but you don't hear much of them outside of that.
As someone who knows folks that have brought McKinsey in to banks and other companies. This is pretty much the reasoning, not quite for illegality, but say you want to derail another exec's project or put a seal of approval on yours? You bring them in to say what you want said
Why does the other exec fall for it?
They likely understand full well what's happening, but it raises the cost of failure
"So the third party consultant warned against it, you did it anyway, and it caused these problems?"
They offer both
> AI is brains. Companies need arms and legs and relationships.
This is silly, these companies are owned and run by some of the most connected and powerful people on the planet, who also own and run dozens of other companies that produce things.
> The reality is that employees are also mostly not entrepreneurs, same as consultants.
I don't even know what this means. They choose to hire who they choose to hire, and they can continue to hire as long as they wish. It's like saying that the reason I'm not a marathon runner is because I only own Oxfords and flip-flops, not running shoes.
I have absolutely no idea how this could be the "sort of obvious answer." I'm surprised so many people thought it was worth a response. There's no answer here. Some of these sentences don't even make sense.
thanks pessimizer.
2 things: 1. Yes they are. Yet here we are not using WorldCoin despite its obvious utility. But backdooring every major software library in the world is pretty useful for governments. Too big to fail. But not an economic weapon, more like a cyber one.
2. Employees and consultants don't have what it takes to start a business, that's why they become employees.
Just take a second. I promise it makes sense. Many see the simple point I'm communicating.
The part you're not stating is that "what it takes" is (1) a safety net which allows you to take chances without risk, (2) access to capital to get your idea off the ground, and (3) connections to get your foot in the door.
Employees have everything else they need -- intelligence and the ability to put in the hard work. The only reason they don't have the capital and connections is because the existing "entrepreneurs" gatekeep it to minimize their competition.
I don’t think paying customers are gatekept by entrepreneurs.
Brilliant comment. Companies (that are run by executives) hire employees, not entrepreneurs because they want servants, not competition.
The landed elite will continue to defend the only moat they have -- capital and connections. They know they're not the smartest or hardest working people in the room, and if they start to recognize their employees as equals they'll quickly lose their position within their company and within society.
For AI companies to cannibalize the rest of the market, wealthy investors/executives first need to admit that they don't have much to offer society. It'll take a long time for them to allow this to happen, lest their egos (and portfolios) take a hit.
Hmm wasn't Sam Altman one of the early proponents of single-person unicorns? He certainly believes it can be done
There have been stories like Medvi which did $400m in revenue last year with only a two person team and a bunch of contractors, but it is such a specific niche around arbitraging medical prescriptions, I don't think it could be done in-house at OpenAI without it being a huge distraction.
There are IT consulting companies out there with several hundred thousand hand and feet at your disposal
"the same reason McKinsey hasn't taken over the Fortune 500" That's because what mckinsey does is produce jobs for ivy league grads, business majors, and frat members, that's it.
OpenAi is also spinning up and acquiring companies. DeployCo, Orchestra Bio, io Products Inc.
Most of what mckinsey actually does is justify companies in their most toxic actions. It isn't the CEO/Board deciding to lay off 30% of their employees, Mckinsey recommended it. Cutting corners and benefits wasn't the C level's ideas, it was mckinsey recommended alignment with industry standards for synergy.
These elite consulting firms are all about being the fall guys for things that would cause employees to start talking about unionization. They also exist as a jobs program for the kids of very wealthy people.
AI isn’t brains. AI is statistically correlated simulation of brains.
Those are very much not the same.
>> The sort of obvious answer is the same reason McKinsey hasn't taken over the Fortune 500
Sorry, this is not even close to being an answer let alone an obvious answer. The main reason is because 'reality has a surprising amount of detail' (read the excellent blog post if you have). There is a huge difference between doing some analysis, framework, and outlining recommendations on a slide vs. generating business value in the real world. The real world is messy, a complex and dynamic adaptive system, with tons of details where small decisions add up.
Put a feature team in a strong company / startup and put them up against a top consultant on a real world scenario to ship something and generate real world value/uusage and you will get the answer why.
This is exactly what OP was conveying, but in a more proverbial manner (ie. Companies need arms, legs and relationships).
I am not sure you're disagreeing with the parent comment. Both your comments boil down to the same conclusion, and your "well actually, reality is so complex, more at 10" reply is just a rehash of the comment's main premise, ignoring the dramatic opening in the reply.
You say they aren't even close to being an answer, and then give basically the same answer...
came back to write this response ;-)
The whole point of so-called AGI is to replace all of those things and it's what they've been touting since the start.
It's a serious bit of hypocrisy posed by OpenAI here, and it's telling that the AI is much more limited that the narrative of 'AGI' would have us believe.
OAI and Anthropic have massive budgets they should be able to expand, were AGI actually AGI.
They need electricity & compute which are limited by the fact that we are living on a finite planet.
They learned the lesson that if you’re in a gold rush, you don’t ever want to be the ones who are doing the digging. You want to be the ones selling shovels.
That’s especially true if it’s an artificial gold rush that you created yourself.
Agree with everything on your post. I have to say as a frontier lab employee, it hurts a little to be compared with McKinsey, even as an analogy. :-)
You're absolutely right! It's not personal. Frontier Lab employees are incredible people - at least the dozens I know are! Undersigned - A Member of the Permanent Underclass.
The even more obvious (and correct) answer is: the models are in fact nowhere near as good as they claim.
1. They are starting to do it. Take a look at OpenAI's product portfolio, there's a galaxy of attempts at application.
2. Selling tokens via long running agents is the only thing that's provably working with respect to revenue growth.
3. They need revenue yesterday. Increases in borrowing rates means access to capital isn't what it was in the 10s. Their runway is much shorter than you would imagine.
4. Some of the decision makers at these companies believe AI will exhibit Jevons Paradox and that any decrease in price will result in induced demand.
5. The frontier models are still subject to the limitations of AI. They're not good at everything. Better to have someone else learn that on their dime rather than yours.
Is 4 correct or incorrect? What data do we have so far in this regard?
AI price-performance is improving rapidly. AI usage, whether in tokens used, in cash spent, or any of the above performance-adjusted, is curving upwards too.
Causality is uncertain, but the "Jevons paradox" interpretation is certainly not contradicted by the available data. Anecdotes of per-employee AI spend rising, and multi-agent freerunning AI tasks becoming more common are also plentiful.
Personally, I think that the "Jevons paradox" interpretation is basically correct. Humanity doesn't have a limited demand on intelligence - but humans are expensive, so it's supply-constrained. The cheaper intelligence gets, the more of it ends up used. As AI gets cheaper and more capable of substituting for human intelligence, usage will rise.
Unknown. AI usage went up exponentially this year but a lot of that was driven by access to frontier and near-frontier models and a shift to agents. Does that constitute a decrease in prices because $ per unit of intelligence (whatever that could mean) has gone down? Maybe.
One issue is that token usage and revenue seem to be decoupling in real time. For Jevons paradox to be true, prices need to be going down but total spend on AI has to be going up. That's technically true right now but total usage is going way up while total spend is going up slightly. Another problem is that we may be witnessing the end of "tokenmaxxing" with Anthropic's revenue growth decelerating and OpenRouter's token volume plateauing. Two points of data do not make evidence but the possibility that AI usage is an S curve rather than exponential is live.
OpenRouter's token volume plateauing
Where's the plateau: https://openrouter.ai/data/newsletter/the-token-state-of-pla...
>3. They need revenue yesterday. Increases in borrowing rates means access to capital isn't what it was in the 10s
This is a huge problem, why is it like this
> why is it like this
In part it is like this because they've already borrowed so much.
They have borrowed almost nothing.
Investment isn't debt.
Isn't investment like a form of debt, or well obligation to make returns?
If VC provide you funds, there is an expectation of returns, otherwise the board would vote you out.
Yeah it's not a traditional loan from a bank, but it's not a 0% interest loan either.
> Isn't investment like a form of debt, or well obligation to make returns?
No.
> If VC provide you funds, there is an expectation of returns, otherwise the board would vote you out.
A person might get replaced as CEO but the company has no financial obligation to the investors.
A loan means the company does.
Legally they are dramatically different.
The duopoly AI labs haven't technically borrowed much, but a metric ass-ton of money has been borrowed by various other entities in service of them as infrastructure proxies.
That's what is driving up the cost of debt and teetering some of these companies into junk status.
you can say coreweave
Investment isn’t debt, but investment and debt demand revenue in functionally identical ways, and OpenAI has taken an enormous amount of investment, so they are subject to broadly the same pressures as a hypothetical company that had borrowed lots, no?
> investment and debt demand revenue in functionally identical ways
No they don't.
A company has legal obligation to pay loans at the agreed rates.
There is legal obligation that defines rates of return beyond "acting in shareholder interest" (and OpenAI is a public benefit corporation so even this is reduced).
Exorbitant Cloud costs and spending 22 million to solve a 1 million dollar maths problem isn’t investment though.
Fake costs. The numbers reported are what they would have charged other companies for the tokens.
That’s 22 million of capacity foregone for a marketing stunt. Or they had more capacity than demand which seems odd for a product that’s allegedly in constant demand etc.
There just isn’t a sane way to look at this as “investment”
So all those hundreds of billions of dollars investors ploughed into OpenAI was just a gift was it? No expectation of a return of any kind? What if that investment money was originally borrowed by the investors?
If the money was borrowed it would have to be paid at an agreed rate.
Because it's an investment it is risk capital. They won't be happy if they don't get a return but their return is part ownership in the company, not a legal contract giving an income stream.
But OpenAI won’t have to pay them back regardless of how unhappy they are, because it’s not debt.
The difference between investment and debt is that debt has a senior claim and fixed returns. You can't be forced to sell from investments.
they've borrowed so much they're directly competing with national governments in capital markets.
OpenAI has hardly borrowed anything. They have a $4B revolving credit line but have hardly touched it. They haven't issued any bonds.
Investment isn't debt.
> Investment isn't debt.
...yet.
SoftBank is very leveraged with …… debt.
One of the biggest AI investors.
With investors like Softbank, who needs competition?
SoftBank has a terrible track record with respect to the success rate of their investments.
This is objectively not true. Alibaba, DoorDash, ARM, Slack, etc. were all SoftBank deals. Alibaba alone is considered one of the highest ROI's in the history of investing.
Does SoftBank have some absolute blunders on large bets? Totally. Do they make some super weird billion dollar bets? Absolutely. Can you scrutinize them for making some terribly and laughably large bets go to complete zero returns? Sure. But this would be like saying Michael Jordan had a "bad track record" for scoring points because he had games playing for the Wizards where he had like 5 or 10 pts in a game.
Michael Jordan has skill.
When you can throw billions around for investments then on balance you will make a lot of money because you can take a lot of risk. But it does not say anything about any particular investment and if you look at the ratio of good bets to bad bets then SoftBank is terrible. That does not mean they won't make money or that they won't be able to keep doing this. Capital - especially at this level - begets more capital.
> When you can throw billions around for investments then on balance you will make a lot of money because you can take a lot of risk.
How does this explain a $20M bet on Alibaba that netted $60B?
> if you look at the ratio of good bets to bad bets then SoftBank is terrible.
So does a16z, Sequioa, etc...
This is venture investing 101 - they have power law returns. Don't let your survivorship bias cloud your vision.
> How does this explain a $20M bet on Alibaba that netted $60B?
It's a 26-year-old investment. The advisors who suggested it are long gone from SoftBank, and the stake in Alibaba was sold to provide liquidity for bagholder investments like WeWork.
SoftBank was lucky, and now their luck is gone. Son had an investment in Nvidia that would have been worth $100+ billion, but sold it in 2019 for only $3.6 billion. Uber is struggling, Arm is a laughingstock of a business, their stake in OpenAI is comically overleveraged, and Softbank Vision Fund is one of the most heavily ridiculed and mocked international VC funds anywhere in the world.
We have to move past a decades-old bet, SoftBank's current state is not reminiscent of their prior investments.
> The advisors who suggested it are long gone from SoftBank
You mean Masayoshi Son, the founder and CEO of SoftBank? lol wut.
https://economictimes.indiatimes.com/news/international/us/s...
> Softbank Vision Fund is one of the most heavily ridiculed and mocked international VC funds anywhere in the world.
Y'all need to stop reading TechCrunch headlines.
The rates environment directly dictates what capital is available to OpenAI and at what terms. When potential investors can get x% virtually guaranteed, then they are less willing to give cash for equity to OpenAI.
They owe people hundreds of billions, possibly even over a trillion. They've just been very clever in how they've structured it, so on paper they have low debt.
I think there’s some crossed wires here with the wider data centre build outs. OpenAI have their own flavours of sketchy behaviour but I’m not aware of them solely being on the hook for those amounts short of some weird accounting of Microsoft cloud costs onto them or something.
Don’t they have a fully committed preorder for like $500 billion in compute from Oracle starting in 2028? And Oracle is funding that buildout with junk debt?
Oracle is the one with the debt. OpenAI has an order/contract (promise) to someone else.
Presumably there are penalties if OpenAI were to cancel that order (I am confident that Oracle's army of lawyers made sure to include that). That's the unmovable rock on OpenAI's books, not the data center.
> junk debt
Not a useful term, here.
AIUI, the building is funded by making the people who order it part owners in the new data center company. Not sure and I don’t care enough to find out; clarification welcome.
https://augment.market/pulse/openai-reportedly-0-debt-665b-i...
"On paper low debt" actually means.... low debt.
but the investment is increasingly _fueled_ by debt, regardless if OpenAI is the direct borrower. just look at the bond market. you think we'd be seeing these rates if corporate bonds weren't directly competing with government-issued bonds?
That’s mostly not the Frontier Labs. They’re definitely creating a lot of the demand but the people taking out those loans are Neoclouds and Big Tech (hence why Facebook + Google et al suddenly all have terrible cash flows after years of having more money than can be spent).
AnthropenAI are the ones driving a lot of that infrastructure spending when they’re such big consumers of compute but they’re not the ones feeling the brunt of the infra spending. Although those costs will bubble up to them indirectly eventually.
Indirectly? They have contracted the capacity with the likes of Oracle, this is not 'debt' in accounting sense, but it is a financial commitment that will require either taking on debt, or getting more investors.
But those clouds are taking out debts because they have 'hard' spending commitments from OpenAI and Anthropic, which start the day the clusters are energized (supposedly).
So in effect the debt is backed by OpenAI/Ant's future revenue stream.
The 10s were totally fake due to ZIRP.
3 is biggest point imo.
Openai haven't yet amortised their current investments into the space and nor have they stopped making new ones.
I'm sure they'd be totally capable of bringing 1k new biz online but that's even more debt, and even more uncertainty.
To make point 3 happen fast, their army of Forward Deployed Engineers roam inside existing companies and build for the companies using the companies subscriptions.
Is this a thing? Where can we learn more?
I believe the point is that every software engineer using AI is one of those “Forward Deployed Engineers”.
Although the point would be equally valid if they were OpenAI consultants. (Imagine paying twice for the help: once for the consultant and once for the tokens.)
Forward deployed engineers is a role at these companies actually openAI and misanthropic. to fix the shit their products cause at their clients.
Yes it is a thing. You can read about it in their job portal or when you’re going through the interview process. It’s an actual role
this point is not bad. i imagine this is the most viable option for them since those engineers build experience and knowledge. but they only build softwates.
Some variation(s) of 5 is carrying a lot of weight. Need a person with an Idea. Need someone to refine their idea. Need someone to spend the time filtering through the slop to find the good bits. Much better to sell people on the idea of their dream being created while you sell them tokens to do that.
That all said, A person trusting the big AI companies with their golden goose idea seed leaves a lot of concerning openings down the road if that idea gets "leaked" out to others. Similar concern of mine with if the LLM outputs that made your idea happen to be reused by the big company if someone else asks similar enough questions or otherwise coaxes the AI to reveal what it did for you.
All this AI generation is gonna be a lead to a wild time for IP law.
This framing is a bit off; or from an entrepreneurial perspective which isn’t applicable to the automation of the overall economy. How economists model automation is task replacement. Work consists of tasks some core some ancillary. Ai is clearly more suited to text and command line tools making coding an ideal task for ai. Other text oriented tasks follow; math, law, etc. implementing systems to move tasks toward ai and eventually robots will take time. Similar to how it took time to build out roads and gas stations and cars to utilize oil and gasoline.
I think it was from an entrepreneur perspective on purpose. Can something almost entirely built on AI compete with a human business? AI isn't going to do car repairs but it might handle the things you listed.
On 1) though, they're an inch deep and a mile wide.
Well, when they solve all these unsolved math problems they are not 1 inch deep in math....
Wouldn't you start that way, too?
> Their runway is much shorter than you would imagine.
If they don't make it, there will be plenty of time to mine through their dataset for business ideas to sell to other people as they liquidate. Think of the Spirit Airlines dataset that was just sold.
The question formed wrong . They killed entire industries. I feel like they are at inflection point where they hired fde , but for now they try to solve fancy problems to do not visually scare their customers .
Yet picture is very sad , industries affected to the point that majority of the companies asking about their existence:
1) software dev tools all IDE , developer analytics tools eg jellyfish etc . Dead . On the same note : warp.dev that raised $80m a year ago as an ai terminal… can barely explain why do they exist now.
Visual workflow editors . Older ones and even n8n. There were literally 1000 companies that were doing small deployments. Why do you need it now ? A year ago half of the Internet was booming about these tools .
Any data visualization tools . Gone .
Data integration tools all kind of crm to crm to erp to something else . Solved problem
And you can go on and on . I would probably not exaggerate if I’d say that millions of businesses are heavily affected by, from tiny one to adobe .
give a list of industries killed please. that is a huge claim without any evidence provided (by you).
They're using "industry" in a very loose sense, they mean certain subsets of software development and B2B software tools.
Alright, do tell: Which subsets of the software development industry and which specific B2B tools, were "Killed" by generative AI?
Notion Mail, Mechanical Turk, Fiverr, Relay, a bunch of Salesforce, Upwork, Gitlab (~15% layoffs), Atlassian (lots of layoffs already - more product cuts coming), Stack Overflow, ...
And that's not counting AI companies "cannibalizing" their own products, like Sora video generation
Notion mail specifically wasn't killed because it got replaced by AI in its niche, moreso that the company is shifting their resources.
Wow I had no idea Mechanical Turk closed. I assumed it would be quite valuable for doing a lot of data labelling type of work that percolates into AI models.
The labs have their users as the perfect training data source now. Much better than mechanial turk.
Has Postman been killed?
You know you can just use ChatGPT to do this for you, right?
Here's one example for you:
Paid homework answers and study assistance
https://www.highereddive.com/news/chegg-layoffs-strategic-al...
> Chegg’s Q2 2026 revenue fell 51% year over year, to $51.8 million. Its disclosures identify AI competition among its business risks.
No offense but this is a bit of a weak example; Chegg got hit pretty early. No one since Chegg?
Stock photos got hit real hard by generative AI. Getty’s in chapter 11 and Shutterstock isn’t doing much better.
The translation market is also wrecked right now. Nobody needs a translator to Polish, the LLM is really good at that.
Just have a look at https://killedbyai.xyz/?
That site is AI slop with a bunch of questionable nonsense. For example, they suggest that Devin, Cognition's product, shut down. I'm not even sure where they got that information.
stackoverflow.
Is that an industry?
Lol, "software dev tools all IDE , developer analytics tools eg jellyfish etc . Dead"
This is the most insane perspective, it's like you have now idea what is happening across the industry. In you fantasy world here, are you saying devs are just yeeting slop into production, no debugging, no profiling, no code review, just prompt to prod? Because that's not happening.
That's definitely happening lol
Maybe I should rephrase, it's not happening at any org that will be in business beyond the next couple years. Customers don't care about your software factory, they care about a product that works reliably.
> Customers don't care about your software factory, they care about a product that works reliably.
You don't think AI is capable of putting out CRUD apps into the ether that work reliably? Even my local GLM5.3-Flash is capable of doing that. These products make up a non negligible amount of the industry. Additionally reducing headcount is about to hit in the next year or so. Where are juniors gonna find jobs? Etc etc...
Rails cli could do this a decade ago, there’s no major difference here other than the number if lines dumped.
> there’s no major difference here
Sure thing, I can see any sort of serious discussion is off the table.
> You don't think AI is capable of putting out CRUD apps into the ether that work reliably?
I know it isn't. No model is capable of programming well, let alone writing good software.
> they care about a product that works reliably.
I actually disagree here, what matters most is procurement & distribution. You can sell absolute dog water software (see: Teams[1]) and capture a market with the right procurement and distributions pipelines.
The sad reality IMO is that the quality of software matters much less than we (the engineers) would hope it does.
[1] https://web.archive.org/web/20260322032609/https://www.theve...
I still believe that quality of software matters in the long term.
The issue with that belief is that it's basically impossible to falsify.
There's obviously some kind of quality floor, it's odd that folks are pretending otherwise.
Teams is dog water, but it's dog water that works.
Yeah I can use it to see and hear my coworkers and their screens and exchange text messages through a crappy editor and schedule meetings. That's really all a business needs to keep using it.
It's more that Teams is already included in a bundle of actually useful and working software in the form of Microsoft Office.
This is the entirely wrong framing.
The right framing is ecosystems.
Instead of just making claims back-and-forth, can this thread provide sources? I don't believe either of you without proof.
winrar & vlc will never be forgotten
How have these two been killed by AI?
they won't be until they are?
I personally rarely touch an IDE anymore. I talk to lots of people with the same experience. This is anecdotal, sure, but if you ask any software developer that isn't staunchly anti-ai or just living with their head in the sand, you'll probably get told something similar.
I find this very hard to believe, even though I use plenty of frontier AI in my dev process. I have to live in vscode to get things done because I still frequently need to look at the generated code and point out issues to get the AI past some problem it's stuck at.
Improve your workflow. I have 4 instances of intellij, 2 web storm, 2 Rustrover and two phpstorm open for WEEKS. Along with some vs code instance. I think I unminimized only one just today to check something: I was being super lazy and I just wanted to check git status without typing anything. I had to rebuild multiple images with new structure to mitigate php2shell; had to patch more than 15 servers etc just in last few days. Fork old other people's images and update them, add some features while at it. Bug fix old php sites not running on new php. And so on and so on...all done with 5/6 agents running in parallel. I have the IDEs open but minimized. Don't need them really...
i like reading the generated code as a code review, and then i can leave inline comments back to the ai, rather than trying to point the ai to the right lines via chat.
im already practiced at reviewing code through code review tools, so id rather do it that way than in vscode
I rarely open the IDE. I have about a dozen agents at all times.
Spec -> Prompt -> Validate -> Test -> Merge Review -> Production -> Agents monitor production -> Alerts -> Roll back
This is true for a many of us. All my team's code is now AI written.
> This is anecdotal, sure, but if you ask any software developer that isn't staunchly anti-ai or just living with their head in the sand, you'll probably get told something similar.
No competent developer is just letting an LLM write code for them.
No one's going to come out and admit that they're just stamping LGTM on every pull request because they can't bear to review another net 2,000 lines of AI-generated code. The problem is that right now the tools we have for code review were designed for a world where code was cheap.
Invert it. Write code yourself, have AI do the code reviewing. You'll have a much better understanding of what's actually going on.
One example I'm aware of is JetBrains. For years, they seemed to have a near-monopoly, but now they're in trouble.
https://ghuntley.com/kodak/ — there's no inside information here, but the article is spot on.
Seems like a really bad article. They have a loss, but their revenue is still trending up, which suggests they are spending more money building products (perhaps with better AI tooling). That's a pretty normal thing for a healthy business to do in this landscape.
The article doesn't mention this at all. It just blindly includes that graph and than muses about the death of Jetbrains for a few paragraphs.
Or, and hear me out here, maybe throwing VC money behind slightly fancy terminals and workflow editors was always destined to be a risky move?
I heard some chatter about Warp some six months ago and decided to give it a try. Blew my mind that a terminal with an LLM wrapper side-panel raised $80M
I'm sure there's some impressive stuff in there that the warp team built but also the second my own _terminal_ tried to get me to subscribe to a monthly plan I uninstalled that shit so fast.
This is the economy that's inevitable when investment is driven by Boomer retirement finds and not working consumers. Desperate workers with no leverage just have to accept the crap decisions leadership makes about what to make and what to use. At some point, you have to decide if you want sanity or "stable returns" (rent-seeking on the young, regardless of the hare-brained decisions up top).
Yeah I think people just don't quite understand what's happening yet.
There's no reason for OpenAI/Google/Anthropic/etc to start other software businesses because "LLMs" will be the only software that anybody will ever pay for.
Why would you have a different "app" to do something when you can just do it all in Claude or Gemini or ChatGPT? We'll be there soon enough.
Because selling access to models will not be a good business. But they can form a cartel and keep frontier models to themselves. This will also prevent distillations.
>Because selling access to models will not be a good business
Let me go on a tangent to make a point. For some countries if you want a US Visa Consulate appt today they only open 3 months out and are booked constantly. They block you for 10mins on the 5th refresh.
It’s well known on crrtain communities that the only way to get one is with an ai agent that can patiently work around this and jump on the bookings with perfect timing or refresh within the bounds of the detection to nab a cancellation opening.
So we’re starting to head into a world where every single person needs an ai subscription just to deal with bureaucracy.
It’s a fucking fantastic business and i don’t understand how you could suggest it’s not good business.
> every single person needs an ai subscription just to deal with bureaucracy.
If everyone uses an AI agent as you suggest, the appointment schedule will be bottlenecked just the same as if everyone only used a web browser. Using AI won’t achieve the goal of everyone getting an appointment. So AI won’t actually be useful here if everyone uses it.
Yes but it’s an arms race. You either give up or you use ai to compete.
The arms themselves are open source and nearly free to operate.
The same goes for electricity. The utility is extremely useful, but it's a utility and is priced as such.
Access to models will trend towards zero, nobody has a moat, open weight models already exist, and inference is getting cheaper all the time.
Ultimately the value will be in the data.
We're in a window here where the data is silo'd and we can use agents to vibe code / glue it all together and create traditional apps that people will still use, but I think that window is narrow.
Eventually today's LLM providers will incorporate direct access to more data sources, and the data won't be silo'd for them - it'll all be available. And instead of third parties vibe-coding apps, the LLM companys' web UIs will become the apps that link everything together directly.
> Ultimately the value will be in the data.
And the compute.
Anthropic, Google, OpenAI, Meta, xAI(whatever is it called these days) are sitting on mountains of compute.
Yes, local and open-weight models will continue improving, and they'll be good enough for an increasing number of tasks. But there's a difference between making yesterday's capabilities cheap and pushing the frontier forward.
The most capable models will likely continue to demand enormous compute resources, both for training and inference.
And the pricing is already there - we're charged for token usage, not model access. Draft an email, summarize a paper - basically free. But the "big stuff" will require Frontier-models with Frontier-level of compute.
OpenAI spent something like tens of millions on "solving" a Millenium Prize math model. And there will always be another problem worth throwing massive amounts of compute at, whether it's mathematics, scientific discovery, engineering, etc.
> OpenAI spent something like tens of millions on "solving" a Millenium Prize math model.
I keep wondering if throwing tens of millions at human mathematicians would have gotten similar results.
That is hard to answer.
But the speed and acceleration has the biggest impact on humanity
It's not even clear that access to data is that important. It's important for LLMs, but somehow we manage to turn babies into intelligent adults without them reading more than a few hundred books - for some, no books at all. Whatever is the breakthrough after LLMs likely won't need lots of data.
Existing is a constant stream of ultra-high fidelity data. A book contains very little data compared to a stroll to the market, which contains so much data your mind is specialized in blocking most of it out so you don't get overloaded.
How much value per type of data?
> Why would you have a different "app" to do something when you can just do it all in Claude or Gemini or ChatGPT? We'll be there soon enough.
At least from the engineering perspective, the app has a massive advantage: determinism. If I write something in Python or Java or whatever, it has set inputs and outputs and those can be proven to accomplish a computational task because they are working in a logical system.
If I ask an LLM to do that same task, it has inputs and outputs, but the logic isn't anywhere nearly as defined. There will be a chance, no matter how small, that will not act in a way that makes any sort of logical sense given the constraints of the task it was meant to accomplish.
When you combine that with the fact that deterministic applications are far cheaper to run, at least for the foreseeable future, the app wins.
Yeah I think efficiency is still a counterargument to the model doing everything.
If I need to process a million items it still wouldn't make sense to feed them directly to the model and burn a crazy amount of tokens
Of course the model can produce code to process them on the fly but if its a repeated task why would I want that to be re-created every time, especially if I've already tested/validated it
You're right but the trick LLM-backed tools have is that they can just write traditional code on the fly.
There are usability and performance problems right now which is why we have this interregnum. But as these systems evolve, the tools can offload "expensive" operations to traditional compute by writing the glue code on demand, and persist the code for future use (so regenerating it won't be required again later) when the user confirms it's doing what it's supposed to do. You can A/B test it in real time and ask the user to pick the "better" result.
Same is true for e.g. GUI elements, widgets, etc. Write it the first time, have your user verify it, persist it, retrieve it.
And maybe you have a couple of people in your org who work through the process of generating a bespoke app that processes data with a GUI optimized for whatever they're doing, but it's all mediated directly through your vendor. The users run through the process all in the web UI of whatever LLM vendor the org's data is silo'd with, the output is a mix of traditional computation and LLM contexts that anybody can maintain via asking the tool to change itself.
> they can just write traditional code on the fly
this is a point that i think is missed on many people (myself included earlier in the year). Sure, the LLM is non-deterministic but it can write code that is not and test cases to verify it. Then, once verified correct, it can use what it wrote to perform the task in a deterministic way. Even though the LLM is still non-deterministic all it has to get right is the arguments to the software which is much easier than the full logic of the task. That's very powerful.
I've seen this first hand when asking AI to do non-trivial analysis over a couple of spreadsheets and then also a database. It writes python to do this, tests it with sample data, and then calls the python it creates to do the processing on the actual data.
> I've seen this first hand when asking AI to do non-trivial analysis over a couple of spreadsheets and then also a database. It writes python to do this, tests it with sample data, and then calls the python it creates to do the processing on the actual data.
How is this different from what we’re all doing in software development right now? It’s just we review and make the code more robust, and build it into more complex systems. If you have repeated workflows, it would be idiotic to just throw away the code that was generated by the LLM instead of properly building a system. Sure, sometimes you have a one-time use-case. Most non-technical people do. But most technical people are in the business of building systems these days.
I do not mean to be mean, but how had you missed this? The models have been writing python to accomplish tasks forever (meaning a couple of years at least).
The models using code/scripts in the background is not new.
Even if the process of creating software was 100% cost-free and bug-free (which we know isn't true, even the most efficient LLMs actively burn tokens and cost you money) - pre-built software still has the advantage of being opinionated and having features I never thought to ask about.
I might have the general desire to play a video game with my friend where we shoot WW2 like guns together. I'm willing to bet that a video game company with artists, game designers, historians, etc. is better able to articulate that into existence than I can. Especially given I might not be willing to spend the time to do that...
Yes, LLM UX is still very primitive. With code harnesses and build tools still distinct from general chat interfaces.
It will all be one integrated UX soon, as the LLM UX best practices are discovered.
But you just ask the LLM to produce a new well-tested artifact to perform those tasks for you, then let it call them and orchestrate them as needed. IT folk have long been creating quality of life scrips for themselves to automate tasks, the LLMs make this even easier. So we get more determinism for cheaper, if done correctly.
If we reach AGI, non determinism will not be a problem. In fact, it will be an asset. Processes will quickly evolve.
If we reach AGI, nothing will be a problem anymore because it's going to be the end for us.
ASI*
We have had AGI for years now.
No we haven't. We don't have any intelligence at all with LLMs.
Yes, I still hear software development coworkers who want to replace all our tested, deterministic software with non-deterministic, high latency, much more expensive LLM based systems.
Have the LLMs generate even more deterministic software to solve well defined needs and keep the LLMs for solving higher level problems using the deterministic software's outputs as inputs.
A chat window that takes multiple seconds to respond and behaves non-determinstically is a pretty bad paradigm for most computer tasks. It just happens to work for software development in particular.
Twitter consensus now is more like LLMs will be entirely commoditized, apps & features will be cheap, and only systems of record will have lasting value.
A chat window that takes multiple seconds to respond, versus your buddy Guy who just went out on lunch.
Non-deterministic behavior from your chatbot, like how the visual design artist's client keeps sending back and forth emails asking for it to "pop" more
> A chat window that takes multiple seconds to respond, versus your buddy Guy who just went out on lunch.
why are you comparing to your buddy? we're talking about software products?
Have you tried chatjimmy? It operates at the speed of thought. Imagine this with one of the recent frontier AI. AI that operates faster than you can express your thoughts seems like a foregone conclusion eventually but may not make sense right now as they try to reach the limit of what they can make in a model.
[1]:https://chatjimmy.ai/
"speed of thought" is wildly underselling how fast chatjimmy is. The speeds claimed by system one decision models are slow in comparison to chatjimmy, which produces actual text.
you use the chat window to make the software which is deterministic and can be tested. Then the chat window uses that software to do the analytics or whatever else you need. Op is correct.
99.999999% of people have no interest in making software, they just want to do whatever they need to with minimal hassle. You may as well say restaurants are going out of business because microwaves exist.
Even the people making software barely want to make software honestly, but you gotta make that daily bread somehow and sometimes it's kinda cool I guess? But like people overestimate the desire of the average person to be their own software maintainer, most people consume, not produce, regardless of ease and trying to turn everyone into a producer is... kind of the opposite of where society has been going for a very long time.
You ask the chat window to build a new tool or visualization or simulation or data editing tool for you...and it spins for a few seconds or 10s of seconds then presents it to you, ready to be re-used and designed to interact with both you directly and the AI itself.
That's where we're heading.
>re-used
Perhaps even by other people, and over time, continuously improving as we discover new use cases or that things are more complicated than they seemed on the first pass. Like... regular software, and not "do[ing] it all in Claude or Gemini or ChatGPT"
Still shocked that people use LLMs as automation for some things that should be done with software.
“Claude deploy this to dev”
vs
“Claude build me a tool that allows me to deploy to dev with a single click / command from CLI”
Orders of magnitude less expensive for the latter than the former. The former usually costs a few cents and the latter maybe a one time cost of 40 cents and then free forever afterwards.
> deploy to dev
You're thinking like a developer that needs to build and deploy software for other people to use, rather than as a user that wants to do something.
“Claude make this thingy run on our test thingy so I can take one final look before I let our users access it. And make a button for it so I don’t have to ask you again.”
Claude deploy to dev will use abstractions to do it with software , even more it will deploy you a trigger to do it automatically . Otherwise . Nothing wrong of doing it with voice / text request vs navigating to some ugly Jenkins , or half dead github actions etc .
> Claude deploy to dev will use abstractions to do it with software , even more it will deploy you a trigger to do it automatically
Who’s building the system that does that? Do you believe every request that we make - in every context that we make it - Claude will be able to intuit whether it’s something we only want to do once and is then just throw-away, or do it multiple times so a system should be built to facilitate that?
Even if Claude is somehow clairvoyant enough to know that, it still costs money to interpret my request in the first place. Why would I spend 5 cents to do something I could do for free? Maybe once, but across the day I’m doing hundreds or thousands of things and the money adds up. I’d rather have that money in my salary rather than having my employer pay for an AI subscription.
What’s the advantage of doing these things through a chat interface, aside from lining VC pockets? Vs clicking a button or invoking a CLI command? Just for the aesthetic of relinquishing control?
> Nothing wrong of doing it with voice / text request vs navigating to some ugly Jenkins , or half dead github actions etc .
I’ve automated away most of my daily browser activity using LLMs to generate tools. It’s become a lot easier recently. I do almost everything from a terminal, which makes it incredibly easy to e.g. pipe content into files for my LLMs to work with, like errors and stack traces. I understand some people have given access to their entire desktop to LLMs but that seems very solution-specific and I don’t want to commit days or even weeks getting my system to work with one piece of AI software just for another one to win the battle in the next few months. Too much churn. Text and terminal are evergreen.
And it becomes determinist - which is a huge bonus
Yeah for me the accelerator has been using LLMs to build engines, not apps. Build something that outlasts the LLM, or could even be trivially used by a local LLM.
I'm under no illusion that we'll have cheap LLM access for much longer so I'm hedging my bets.
So at the end of the day you still need an app… and maybe instead of spending tokens every time you want an app you just use a prebuilt one… :surprised-pikachu:
Depends.
If you just need a narrow subset of the functionality for now, might be cheaper in tokens to just vibe it than the attentional cost of searching for just the right software...let alone paying for it.
Why people bringing token prices ? With my Claude subscription I can do everything from building apps every day to doing room planning and science researches , for a fixed price. Software is not cheaper . Before you could argue oh but it’s a subsidy etc , in reality you can do same things with Chinese models just half a year away for also cost of coffee . It’s cheaper now than sales tax you pay for your saas
> Why people bringing token prices ?
Enterprises needs to pay for API, they cannot use subs (for various reasons). In work environment it’s a higher chance you will pay raw token prices than have a sub.
My example is not a narrow functionalist it’s a complex multi level problem that need several apps .
Sure , I’ll share you my example I was doing home planning from a shitty plan I had from previous owner . Claude created that with a fraction on my quotas . Eg ~1 dollar . It converted a shitty drawing into 3d , modified the room I wanted , created a few options of visuals, and shopping list from my local store with budgets . Oh and yes it create a local html page for my problem it even created a name nook, not sure why .
With regular software I need an entire package of apps, a specially trained people , and a budget of at least $10k for the same service . Will my services be such a commodity very soon , sure .
Do I want to pay $10k now for manual software and people . Not really , what is the point ? Same way as accountant will not pay one day for the software , I will not pay adobe and designer today .
NO, you build a new one every time, or else you get the hose again.
A chat window that will deploy you perfect ui for any problem you have based on your personal taste , that is funny enough is not that different yet. And yes , that ui will open in milliseconds , and it will be deterministic when needed and agentic and smart when not . This is already happening on a small scale with Hermes . With models getting smarter it will Only get better
Nobody really paid that much for software before LLMs, most of the web was moving to a subscription model with API pricing anyways.
I’m seeing two types of software:
Am I really going to create an iTerm2 alternative? What the hell would I want out of an alternative?
Am I really going to create a git alternative? Who in the right mind would take the time to even learn it?
And also, no, Claude isn’t going to replace git. Deterministic compute still has a place in software, nondeterministic compute isn’t going to take over just because the UI is more pleasant for nontechnical folks.
> Am I really going to create an iTerm2 alternative
Well, If you're an iTerm2 user still after about the 1st hour of setting up a new macbook then you probably don't have need for the features that customizing a terminal to your own needs brings.
Claude will totally be used to build a tool to replace git for specific groups of users. There are alternatives popping up left and right, there was one that was announced by Anthropic itself last week.
Clause doesn't need to be deterministic when you can use it to generate software that is deterministic in its output.
> you probably don't have need for the features that customizing a terminal to your own needs brings
I mean this is my point in a nutshell. Why do I even open a chat window if I already have a tool that does what I want it to do.
git is also an archive and you can't just vibe archival because the whole point is knowing what you got in there, and keeping it there "forever"
> stuff that’s a giant boondoggle to replicate because there are so many edge cases and considerations that it takes years of iterations to arrive at something stable.
The catch here though is that you don't need to replicate all of the giant boondoggle with your vibe coded solution.
For eg. Pagerduty does not handle payments in our country "properly". So we are replacing it. Pagerduty likely does a gazillion things but we only care about a tiny subset of that. And that we can build with an agent now.
And there am talking about Pagerduty that's already a rather narrowly focused SaaS product. If you consider the SAPs and the various ERPs and billing systems etc. out there they are much much more vulnerable since most firms just use a tiny tiny fraction of these giant behemoth software.
Anthropic is actually replacing git and will bundle it into Claude code for you. What is so complex about iterm ? And why do you need iterm it’s just a visual trend nowadays, on how to run your agent. Iterm is literally single function to launch Claude code or codex .
Cursor too.
> They killed entire industries. I feel like they are at inflection point
The perpetual state of being in the inflection point upon which entire industries are just... killed.
out source to India..Gone.
service based companies yes, but product based companies have established workflows which will take some time to disrupt
aynone can explain to me what is software factory?
Data visualization is such a dead end product. Wow lots of pretty charts, can you actually interpret the data and make an informed decision? Any idiot can throw shit into Excel and make charts and pivot tables. An AI that can not only interpret the data but also offer advice on how one should react is way more valuable.
Of course you shouldn't just rely on AI that's consumed every MBA book to give you all of your business advice. Sometimes the textbook response is not always the best response.
Jupyter notebooks were the canonical solution to data visualization problems if you needed more than Excel, but required training in understanding Python or similar high level programming language.
Now with LLMs that's all solved.
> Visual workflow editors
That overhyped thing that never did anything well and everybody that used them regretted it later? Yeah, hyped things tend to die once the money used to hype them ends.
> Any data visualization tools
Microsoft killed them with anticompetitive practices. It helps that every single one of them tried very hard to kill themselves by being as hostile as possible.
> Data integration tools all kind of crm to crm to erp to something else
Yep, everybody writes those in Python nowadays. Apache is a great market killer, the best kind of market killer.
> software dev tools all IDE. Dead.
IDEs will be moot when I die. Even with tests and code gen, I still debug and verify code before shipping to QA.
This is a wild take.
I do not even use the IDE at this point.
I read the merge request after validation, testing, then only for verification before shipping.
Huh? First of all, agents use software tools. They aren't likely to go away. It is still infinitely better to have deterministic programs for a lot of tasks whether the user is a human or an LLM. Sure, some tools like human centric IDEs will be deemphasized but those companies will build agent managers and harnesses instead.
Nerds also underestimate how how social people are. I can guarantee you there are whole segments of the economy that will not be fully decimated solely because they don't actually give a shit if an AI can do the job because the job is 99% clout and relationships.
Because the profits from any company which _could_ trivially be redone by an OpenAI agent will crash to nothing, because it could trivially be done by someone else's agent too. It's what ultimately makes the economics of OpenAI, Anthropic et al so crazy: they need to capture trillions of dollars in value, but those trillions of dollars won't exist any more. What's the value of writing a piece of marketing copy now, the same as it was last year? That, across the whole economy.
I think you’re right and I think the leaders of the AI labs know they’re never going to have profits. Money will become meaningless before then. But they need money now to actually make the future where money is meaningless. So I think they’re deceiving everyone to get capital. I don’t think that’s a bad thing though so I haven’t been trying to push this message.
I agree with this perspective. There will be lots of local distortion but the “competition” for anything AI has commoditized will drive profits down. So, who will be able to afford the AI?
Add to that the fact that the AI itself is going to be more or less a commodity as well and you wonder where all the profits are meant to come from.
LLM makes may make more money before their tech really works than after.
George Hotz wrote a similar perspective
>I think this is basically what it will be for most knowledge work, knowledge workers are so grossly overpaid compared to the energy they consume, and AI will rectify that. This explains why the Chinese are giving the (much more moderate resources to train) models away for free. They love to see deflationary economics in the US.
https://geohot.github.io/blog/jekyll/update/2026/06/11/ai-wi...
Hotz is the computer nerd version of Richard Feynman. Big fan.
Uhh just read that blog post and am not seeing what you see my friend lol. Feynman was one of the best scientific communicators and, simultaneously, scientific contributors of the 20th century. We have plenty of people who are great scientific communicators, and plenty who are great contributors, but not quite any these days who achieve his notoriety as both.
Michael Nielsen is up there among my favorites as well.
>explains why the Chinese
I think it's less that they want to encourage deflationary economics - because I think this has a tremendous chance of backfiring given that is not how massive productivity/efficiency gains have turned out in the past.
I think they genuinely want to prevent the rise of yet ANOTHER Western bigtech monolith in Anthropic and/or OpenAI. They want to make sure they have no moat. That keeps the value out of the nascent bigtech and instead in people who can run the models.
Chinese models also wrote code with 23% more vulnerabilities if told they're writing for the DoD than for the Chinese equivalent.
No kidding? What’s the state of the art for lists of these kinds of prompt biased performance? Or should I ask Opus to game it out with Muse and <Tencent>
https://www.crowdstrike.com/en-us/blog/crowdstrike-researche...
It is very hard to start reasoning what could have happened in those models but I am pretty sure that it was not intentional sabotage.
That, and also I think it just makes them look good and positions them closer to being the respected leader of more of the world.
Look at how they reduced oil imports when the US was being destructive and stupid. China was just like "we're strong, we'll reduce oil consumption to help the world, and if y'all ever want solar panels, we make 'em--also, here's some free AI!"
Wait a sec...
"knowledge workers are so grossly overpaid compared to the energy they consume, and AI will rectify that"
Then what's left to work on that pay well? machines already do things that depend on mechanical work faster...
Is this the end-all argument that AI is really a mean to pay the least possible to people?
a.k.a. the ultimate transfer of wealth and power to uber rich people?
The petite bourgeois is nothing but labour in the eyes of the capital class.
I just notice that "knowledge workers are so grossly overpaid compared to the energy they consume, and AI will rectify that" implies that you should not be paid more than what you need to eat.
Wait, what? "Overpaid compared to the energy they consume"? Firstly, is he talking about the energy resources humans require to be alive? We consume humongous amounts of resources, especially compared to LLMs. It's not clear what that has to do with anything.
Secondly, this is not how Economics works. Relevant terms here would be "supply and demand" because that's what determines almost all wages. Knowledge workers (or any workers) are not paid based on how much it costs to sustain them, like some function of the COGS bill. Knowledge workers are paid so much higher because they supplied the one thing that was pretty rare but for which the demand was very high: intelligence.
AI is not "rectifying" that by consuming less energy, but by providing a supply of intelligence that scales with energy and compute, both of which are technically easier to scale than humans. Which is another piece he misses, the monopoly is not on AI, but on compute. (See: all the kerfuffle about export restrictions on Nvidia chips.)
He is right about some other things in that article though.
In at 9: Hacker news discovers marginal revolution.
I work in design services, basically build products for other people. I've worked with dozens of companies on dozens of products. Why don't we just create our own products? We certainly have the ability to make products.
Because building the product is the easy part. Most products and businesses fail because they build the wrong thing, have the wrong requirements, have poor market fit, etc.
LLMs can write code all day, they don't have the agency to know what to build.
The distribution and customer research of building a durable and successful product and company is incredibly complex. LLMs can probably do each of the parts but it requires a lot of coordination.
Agent swarms, you say? Perhaps. But I don't think we're there quite yet. At least, we're not seeing it manifest.
We can expand this to coding too; you write code for the company to build their product but you can't just go and build the yourself because coding is the easy part. Requirements, product market fit, etc are what truly make it.
Hence this is also why although AI is writing the code now, it doesn't replace the companies or build entire large scale products without human oversight.
They don't need to start 100 companies, they just need to start one, call it Jane Street 2.0 if you will, and all their funding troubles would be gone. But they don't, because that would be competing with actual intelligence in a very measurable way. Instead they keep acquihiring companies like Astral, known for carefully crafting software rather than vive coding, while claiming they are just about to solve all humanity's problems.
Put simply: who not build the perpetual money making machine if you can?
Because money has no value, if you don't give it one. If you make infinite money, it's worth 0. Same with the 100 companies - OpenAI and Anthropic are worth something, only if there's someone to buy the tokens. If any of them successfully proves that having a company is worthless, many companies fail, and then there's no one to buy the tokens.
Sorry, but claim that they can outcompete all human-managed companies but they deliberately do not do this seems to not pass a smell test to me.
The mental gymnastics are astounding. These are companies trying to juice their numbers every quarter. They're threatening to replace labor. That argument has 0 internal consistency.
Going up the value chain is much more difficult as opposed to going down. A packaging company for example will have a much harder time creating a snack company than the snack company creating a packaging company. The snack company will know exactly what it needs for its packaging and have strict requirements.
Down the chain you see what you need to do because you work directly with it everyday.
I don’t think I agree with a snack company having an easier time becoming a packaging company. Sure they know what they need but outside the idea there’s almost no overlap in the companies business units
Neither would be easy for either to do
I think there's some real truth in this. OAI can build their Jalapeño chip because they have become experts in CUDA and have strong opinions on how the hardware should work for their models, but will have a much harder time building a quant trading firm without knowing what's a profitable trading strategy.
That would scare people away from sharing their knowledgebases with them and regulators to act. At this stage, as long as people are sharing more and more knowledge with the models or if there is roadmap to collect more training data somehow, the smart move is to wait.
But look at the ArtCraft project. At least one of labs has an internal benchmark to clone 20+ popular products much like ArtCraft but with much higher budget and several configurations of the workflow. ArtCraft spent $20K initially, not sure how much by now, but they are moving fast and the apps work. They day is coming when they can flip and just clone everything including the ops.
> That would scare people away from sharing their knowledgebases with them and regulators to act. At this stage, as long as people are sharing more and more knowledge with the models or if there is roadmap to collect more training data somehow, the smart move is to wait.
They don't have to do it openly.
Indeed, they are doing this, just slowly and strategically, to not turn off the knowledge tap. A major company I do some consulting for just migrated from Figma... to Anthropic Claude.
How long until people switch from iOS or MacOS to ClaudeOS?
The downfall of Apple may be closer than people think.
how do i bet against that?
Buy AAPL
Send money to Apple.
ClaudeOS - thats actually a very good idea! I have not thought that far yet, they could easily build a good one as a lot of Apple employees are inserting their thoughts for free into Claude :-)
I take it they're using Claude Code & Claude Design for design now?
I find Claude Design to be really interesting. It made a huge splash, and now...nothing. I believe it hasn't been updated in a few months. I read a comment here that said that it was like the person who spun it up had moved off the project or abandoned it. Keeping in line with the thread topic, perhaps Anthropic is just biding its time until making humongous updates that will vastly increase its capabilities?
> including the ops
Citation needed.
"sharing" their knowledgebase?
Nobody "shared".
Because they already employ thousands of programmers that the AI cannot replace. Don't believe the marketing hype about AGI, we are extremely far from it.
I doubt if we will ever get AGI.
Ever is such a strong word.
And AGI is such an undefined word.
Doesn't have to be for literal ever, just not until we (humans) blow ourselves up - if we don't achieve AGI until then, that's effectively ever. And you could gamble we'll blow ourselves up or make the planet uninhabitable before AGI.
Not really, humanity's time is very finite on a planet with finite resources. If we don't build it soon enough it won't get built.
Ten years ago people would have unambiguously identified what we had today as AGI.
Unambiguously?
As long as you don't see thousands of AI businesses that are run entirely without humans people will reject the idea that they are AGI. People could easily do that today if these models were AGI since its not hard to put them in a loop and tell them to run a business, so we know when that threshold is reached.
And if the AI isn't general enough to handle every part of a business well then you will not convince people that its AGI.
This hasn't changed from 10 years ago, this level was always what people imagined about the future of AI.
This. I don't know why people think the goalposts have moved.
Ten years ago, AGI implied general intelligence, as opposed to ML models that had learned specific tasks. It meant a model that could reliably zero-shot problems you threw at it to human level without being specific models trained for that task. As recently as 2024, DeepMind's framework had:
> a Competent AGI must have performance at least at the 50th percentile for skilled adult humans on most cognitive tasks
and "emerging AGI" as
> "equal to or somewhat better than an unskilled human"
The moving the goalposts here is people saying it has to outperform humans on literally every task to be AGI. We have artificial general intelligence by any reasonable definition; people here are complaining we don't have super-intelligence.
I don't get it... you seem to be quoting some definitions from DeepMind's Levels of AGI for Operationalizing Progress on the Path to AGI[0] for competent and emerging AGI from July 2024, and yet you're saying that anybody that doesn't follow those new definitions did move the goalpost in 2016?
[0]: https://deepmind.google/research/publications/66938/
> The moving the goalposts here is people saying it has to outperform humans on literally every task to be AGI
You don't have to outperform humans to run a business, nobody said you have to outperform humans on everything. Humans can run businesses, so AGI should be able to as well.
As long as humans outperform AI on a lot of tasks we have yet to reach AGI, and the easiest way to see that is that business still hires humans to do knowledge work and businesses are still lead by humans.
A lot of businesses have physical requirements. AGI has nothing to do with robotics so its hard for me to say you have to be able to replace a random business to be AGI. Definitely feels like we'd have AI investment firms if we had agi though
Two hundred years ago people would have unambiguously identified what we had today as mAGIc.
thats true even before GPT
I suppose that talking machine wouldn't be that surprising for them in the overall bundle of magic and miracles.
What does that even mean? They would have tried the models, noticed their limitations, and decide it's not AGI. Same as today.
I talk to a few models, they talk to each other. Together we’re AGI. Hybrid, but still.
But weren’t you real GI on your own, before ai? Don’t sell yourself short bro! You got this!
I'd give up in a lot of hard tasks. Production C code, create a new classifier, fix my car, debug this piece of electronic equipment, negotiation with X, gov application.
I asked my AI girlfriend whether AGI was inevitable and she said my question was brilliant and insightful.
This made me laugh! :D
What does AGI actually mean to you?
Does it mean 'exactly human level capabilities in all aspects that a human can perform'? Because I agree with you on that point - no one really wants it, either. We already have several billion computers in the world requiring a low power requirement, and fairly easily reproducible, that can do that.
Does it mean 'smarter than a human on most tasks'? I'm pretty sure Opus 5.5 has crossed that line fairly definitively.
Does it mean 'smarter than a human on all tasks'? I think it's doubtful we'll ever get there with LLMs which are trained on human-derived data. I also generally think this is "ASI".
The history of naysayers claiming "AI will never do X":
https://www.youtube.com/watch?v=ZmInkxbvlCs&t=10s
The models are also insanely inefficient and economically unviable to actually train and run. They’re doing the Uber style model of hoping they can keep afloat long enough to make other businesses go bust and then they can snap up their customers/market and amp up the prices
> economically unviable to… run
Didn’t DeepSeek just disabuse us of that notion?
> …with overall revenue of US$38 per GPU-hour (at valley pricing)
https://github.com/deepseek-ai/DeepInfer/blob/main/doc/infer...
Problem is, OpenAI is maintaining a (slight) edge over DS seemingly only by being extremely inefficient.
Revenue by definition does not include operating costs
Yeah but you can look up how much it costs to rent a GPU, unless you think all those companies are also losing money.
At the top end B300s are $9/hr from RunPod. Now think about large volume discounts etc
Because LLMs still do not know how to manage context.
Proof: - Load up all your chats, across several different topics and see how even the best models will try to connect dots with lines that don't make sense.
As a side effect of the above, LLMs have trouble coordinating with other LLMs.
Humans are able to compress and manage context without much effort and each human brings their own unique set of experience and perspectives.
That's why humans scale in ways that LLMs cannot yet match.
Maybe by next year they'll crack this too but that's where we are today.
Kind of interesting that the built in social modeling and empathy circuitry built into human brains is so important for accomplishing large scale tasks, rather than a distraction from pure logic and reason slowing things down.
I am reminded of something that some else said on ai, data centers and Nvidia, "More People Made Money During the Gold Rush Selling Shovels Than Prospecting for Gold".
It is this, but for a globalised shovel seller selling shovels to the entire world population, for people all over the planet to start digging in random places in the hope of finding gold. Very wasteful, disruptive and not likely to end well for most people, but likely to make a very small number of shovel sellers much richer than if they'd just used their own shovels to dig in areas with known gold deposits.
With lower stakes... how about a coffee shop?
https://apnews.com/article/ai-artificial-intelligence-sweden...
> San Francisco-based startup Andon Labs has put an artificial intelligence agent nicknamed “Mona” in charge at the eponymous Andon Café in the Swedish capital. While human baristas still brew the coffee and serve the orders, the AI agent — powered by Google’s Gemini — oversees almost every other aspect of the business, from hiring staff to managing inventory.
https://futurism.com/artificial-intelligence/google-gemini-r...
> Some of its questionable business decisions include ordering thousands of rubber gloves, despite the café only having a handful of employees. The AI’s handlers, nonetheless, are holding out hope that this is just a blip from expensive setup costs. How well it performs will raise grander questions of the tech’s impact on the workforce.
> “AI will be a big part of society in the future, and therefore we want to make this experiment [to] see what ethical questions arise when we have AI that employs other people and runs a business,” Hanna Petersson, a member of Andon Labs’ technical staff, told the AP.
> To launch the experiment, Mona was given a simple set of instructions. It should run a profitable café, be friendly and easygoing, and try figure out operational details by itself, Petersson said.
---
And that last part is the "why doesn't a LLM company open up a company to sell its wares". There is knowledge from life experience that a LLM doesn't have.
For a product, you need to recognize a need in the economy... something that irks you that you can make better. Go through https://www.npr.org/series/126166186/from-scratch or https://www.fromscratchradio.org/show/ (or any other podcast service) and consider how often those companies were founded on something that was recognized as a need or something missing that comes from the human experience ( https://www.fromscratchradio.org/show/mick-mountz for Kiva systems ).
---
The company represents risk. The coffee shop needed money to start... and it failed. https://kotaku.com/google-ai-put-in-charge-of-swedish-cafe-b...
Creating companies making applications that people don't need is wasted money.
It's funny that Google tried it after Anthropic failed at WSJ https://www.wsj.com/tech/ai/anthropic-claude-ai-vending-mach...
A video article about the WSJ vending machine if you don't have access to the text (that's quite good) https://youtu.be/SpPhm7S9vsQ
Andon Café's website shows its bank balance and a bunch of other statistics https://andonlabs.com/cafe
They also have a market (https://andonlabs.com/market) and a radio (https://andonlabs.com/radio)
This is more or less the same as a question Alex Karp asked in that hilariously off-the-rails CNBC interview: why, if the code is so good, are they selling by the token and not asking for a cut?
https://www.yahoo.com/entertainment/tv/articles/televised-ne...
I think OpenAI actually will ultimately start a fleet of consultancy/SaaS spinoff companies, each perhaps short-lived until the core product does the job out of the box.
Forward Deployed Entrepreneurs.-
I think there is already one: ODE, a joint venture of sorts between Blackrock and Anthropic, if memory serves.
As for the cut, I am sure they will try. Where I work we have had a few projects that were built entirely with models - not a single line written or read by humans. One of them is now the fastest growing money maker for my company, reaching 89M of revenue on about 9% margin in 2 months of its existence, and growing.
It cost roughly 40k in tokens to build, 7 weeks from the first prompt to first production revenue, and was built and is being operated by 1.5 guys. Its a fantastic bang for buck for us, as previously such complex automations would have been impossible to build economically.
There will be an attempt to get a deeper cut of that value. We already see this, for instance, in Meta charging by message on Whatsapp from Oct 1, raising the costs for AI-powered chatbots delivered to consumers via whatsapp by 10-15x easily. More to come for sure.
Can you share more in the product? 9% margin also doesnt sound great imo
Customer service industry, roughly. Its a low margin business to begin with. We were able to expand into a new area, where we previously couldn't operate at a profit. We hired 3 people to run the new business, built them a tool, and it now runs pretty well and above our normal margins.
> Its a fantastic bang for buck for us, as previously such complex automations would have been impossible to build economically.
If you really did it with 1.5 people and it is profitable, you will have 1000 clones in 1 year. The only moats will be first mover advantage (slim), connections (all right) and the possible secrecy of how much you're making (very slim).
I am not arguing it. But if we didn't do it, others would. And I dont worry about 1000 clones, I worry about our key competitors building 3.
Edit: I'd also argue its not as simple as vibing a trivial solution. It took deep domain knowledge, which is not in the public domain, and is built on decades of operating this sort of thing manually, and integrating it with dozens of external systems that help make this into a moneymaker. I am not saying its impossible to replicate, but it's not something one can replicate by merely watching it function.
Domain knowledge is definitely a moat. I'd argue another moat is surprisingly, if scaling is hard or even impossible. The biggest sharks just won't bother.
I call this stuff 'serious automations' -- where a process/decision tree has tens or hundreds of branches.
I see a lot of fluffy things, which are automating the simplest part of a process, leaving a lot for the humans to deal with. This is the thing that can be cloned a 1,000 times in 2 weeks.
I'll give an example. In the system I mentioned, people are scheduled for paid procedures. The procedure takes 2 days, but often we have no-shows on the 2nd day. This 2nd day can be repurposed for someone who was a no-show before - its already paid for, and every penny counts. So the chatbots chase people who may be able to make the appointment last minute, and decrease the costs to operate. There are multiple people in such pools, so the system needs to decide whom to approach, how much time to wait for a decision, and so on. Its gnarly, hard stuff to do right.
The system I mentioned has about 150 such cases, and previously just building the automation would be too expensive, and maintaining it -- too slow. Its a living organism, with thousands of people using the system every day, and more edge cases emerging constantly. It is thanks to AI coding that the thing was economical to build, and its thanks to AI that we can iterate quickly enough to keep it working well. We get all the value, and the AI labs selling us tokens got a pittance.
In my industry, the automation of such processes was always a stated goal - for at least a decade, but its the first implementation that actually works end to end, is not hated by the end users, and does not require a lot of manual interventions. It was still very hard, but at least coding wasn't part of the problem.
Here is a link to the actual video and quote: https://www.youtube.com/watch?v=0A3sGymV6kY&t=259s
"is selling tokens really the best strategy"
Yes. in a gold rush, the best strategy is to sell shovels. they sold us an idea of whats possible, while they earn NOW.
come to think of it, whether you succeed in building the "next big thing" with thier tokens or not, they get paid.
the real shovels are gpus
the real shovels for the real shovels is energy
but even in a gold rush selling shovels _at a loss_ would not seem to be the best strategy, unless there would be some kind of shovel lock in effect.
They are not selling tokens at a loss, they have insane profit on tokens. It's the only think making them money instead of loosing money.
I dont believe they are still selling tokens at a loss.
Some pretty creative accounting is the only reason you could consider them earning anything. They're taking a bet in the future they'll earn loads, maybe the will, maybe it will be through tokens or maybe it will be by replacing companies.
But where the business model lands we simply have no idea of currently because the future is in the future.
That might have worked if not for the Chinese models. There is no sensible moat to guarantee them (US AI companies) the margins they need to be profitable.
>while they earn NOW.
Except they're not. Those shovel sellers weren't losing $2 for every $1 shovel they sold.
It's not the decision that matters, but the accumulating costs that follow.
Its always a question of why businesses don't just eat the next level, downstream.
Why doesn't NVIDIA just make the best frontier model?
Why doesn't TSMC just make the best chip design?
Why doesn't ASML just make a fab?
The real answer here is that OpenAI and Anthropic are in the middle of an intense race, and the payoff of having the best model (or one of the two best) is so large that it makes no sense to focus downstream.
But those all need expertise and openAI and anthropic are claiming to be selling expertise in anything.
You have a good point.
To me, this says something about the level of expertise they offer.
I’m a senior developer, I use Claude every day and am amazed at the quality.
But I also see the shit that comes out of Claude when some colleagues, who were poor engineers to begin with, drive the AI.
It’s not just about the model IME. How it is used, really makes a difference.
I think this is a big part of the answer: the human driver brings judgement and taste (IME, YMMV)
Oh I completely agree! I think it's mostly just that if they truly had perfect AGI, it would be odd to sell it so cheaply.
Expertise is the raw material here, not the same as the end product.
Companies have expanded vertically though. Usually it's not all that challenging. Manufacturers often have to build a small version of the real thing, such as a mining tools company needing a test mine.
The barrier is often that the investors aren't fond of the heavy investment and risks involved. Using AI to generate some app would presumably not have the same level of concern.
It's challenging to expand vertically while also fighting a battle horizontally. Usually expanding vertically is done by companies after they've already taken over a lot of their industry or the industry has matured (a.k.a. less competition).
Whenever anthropic / AI guru talks about loving humanity and how they're scared about the future I can't help but think of the childhold meme where one child bullies the other by striking the other with their own arms. Stop hitting yourself, stop hitting yourself, stop hitting yourself!
The shift you're imagining is kind of like what happened with ASIC-based Bitcoin miners. There was a time when the people who made them realized it would be more profitable to just keep them in-house vs. distribute them for sale. If that happened with OpenAI/Anthropic, it would be mighty unfortunate. A majro reason why it's unlikely to happen (IMO) is the training of these models is aided by external users feeding in new content.
It's basically the same reason why Apple didn't create every app in the App Store. No single company can be specialised in everything, and more importantly, selling tokens is risk-free money since they get a cut for every app without taking any business risk.
I think that’s a different argument. Apple absolutely “copies” hugely successful, durable apps. Just read about “getting sherlocked”. The rest of the apps are transient, niche and it’s easier to just harvest your cut of ads/payments.
Fair point about Sherlocking, but that's actually proves the analogy. Apple only "Sherlocks" broad, consumer-facing features, but they never touch specialised B2B or super niche software. AI model providers are doing the same thing. They build the broad stuff themselves, like AI inside Excel or PowerPoint, but they live the crazy stuff to other companies.
> They build the broad stuff themselves, like AI inside Excel or PowerPoint
Didnt OpenAI build a browser not just AI inside a browser?
idk they’ve created the most used and differentiating apps though
They aren't in the top 10 for either free or paid apps.
https://apps.apple.com/us/iphone/charts/36
That actually makes me wonder if Apple apps are not included because they come pre-installed.
i am referring to imessage, camera, photos, facetime, and so on
I believe this is the correct answer. You can't run 100 businesses because each one would only receive 1/100 of your attention, and they would all perform poorly.
But that’s the gist of the question, can’t their internal frontier models provide that infinite attention? That is what their PR implies.
I thought the agents were largely autonomous, and should replace workers?
If labor is not your constraint, you wouldn't be diverting your specialization.
> No single company can be specialised in everything
Not everything, but a company can be specialized in a lot of different things, Samsung for example, or megacorp in Japan.
Yeah, actually trying to specialize in everything constraints the firm's growth. Though diversifying can always bring some much-needed revenue like you said with Samsung or Tata in India.
Or amazon that can undercut any merchant on their platform by cross financing the losses by selling more aws servers.
That’s the other side of diversification: undercutting your competition until they go out of business. Then prices are raised again.
Yeah Amazon is another example I mean they do a lot of things too. AWS has changed software and Amazon prime which is a vehement competitor to Netflix and disney+ market share.
I wonder what is stopping apple from cloning the top ten apps in each category using AI and then booting the original developers off the app store.
Good morals.
Apple is absolutely doing this for apps that they think are worth copying, there's no morals there.
The G in AGI suggests that not being specialised is the point.
The real reason is the models just aren't good enough.
Running a single successful fully automated business at scale would be a huge milestone and a legendary PR achievement, but the tech just isn't there yet.
Best I can do is Rust rewrites take it or leave it
> No single company can be specialised in everything
A machine with infinite attention and infinite knowledge probably could specialize in everything.
There was/is a youtube commericial I think that gave away the game to some extent with regards to this. It was selling some platform to create apps. Everybody in the office rejoicing because they were all creating apps. Inevitably, like three people in the office make the same app.
I've experienced the same thing in real life a couple of times over.
I think Einstein's quote "A new type of thinking is essential if mankind is to survive and move toward higher levels." is apt. Apps will solve a bunch of problems, but if anyone can create those apps, there will be no market for them. I don't have the answers, but long term, it probably doesn't lie in selling apps. We've only just started.
At my regular job, we're just catching up on all the internal work we had. I think there is a real crunch in trying to figure out how to build moats and things that are hard to replicate that are sellable.
Maybe it will be the physical. Probably most of the moat will be made up of network effects and capital intensive processes. If I had the answer I'd probably be working on it right now.
It's common knowledge that ideas are worthless and only execution matters.
That was somewhat true when execution was time-consuming and expensive.
But original ideas are very far from worthless. They're very rare, because understanding domains and relationships at a deep level is a rare skill.
Meanwhile AI makes execution much cheaper and easier than it used to be.
they would need to literally build a 1000 unicorns to make an impact on their valuation. Or a million small companies - at which level surely even american antitrust agency would begin to look at them.
Also, it would increase anti AI sentiments, and people’s fears of then using private data - that’s the real cost.
In general you want to be careful about competing with your own clients.
AI intelligence is also very jagged. Meaning it can be super human in a lot of things and bad at a ton of others. This is most represented in places where they can't get enough data. Starting business is a combination of a lot of skills chained together and breaking a single chain can cause the business to not work.
Secondly even they are constrained by resources. They have a limited amount of people/GPUs/capital. Given the constraints its best to focus your attention on the highest ROI given your team and organization.
Your first paragraph reminds me of one of my favorite SMBC comics from like 15 years ago
https://www.smbc-comics.com/comic/2010-01-29
Entrepreneurship 101 imo is that great companies only come from people who sufficiently care to see something through.
Unlikely the people that start these 100 companies, across totally disparate fields, and care enough to take them to market are staff members at OpenAI.
And it's already obvious a single company can't split its attention enough to do tons of disparate things well (see OpenAI's previous attempts at Atlas, the social network, Dots, etc)
I think that all software can be built in no time nowadays, but when it becomes a product, it brings with it a lot of things that need to be managed. You can use agents, yeah, but there will always be a need for some humans, and for a company with a mission, like OpenAI, there is no time to waste.
And i'll add that to be profitable, they'll need 100,000 SaaS products, since AI companies are just burning money to stay behind in this rush toward AGI or whatever they're chasing
Its like a shovel making company being asked during gold rush, if you have the shovel or even for arguments sake best and most efficient shovel..Like OpenAI or Claude here..The why are you not searching for gold yourselves...
Similar to how they tout an ai utopia. Who’s going to pay for it? Not them, but their customers will via taxes or whatever. I think the best analogy is oil. Oil companies produce oil everyone else uses it for good and ill. And thus far they’ve not been held responsible for the damage. They should be imo.
Similar to oil the infrastructure to use intelligence on tap will take time to implement be it systems or robots. Ai right now is a bit of a parlor trick imo. Developing ways to use it to actually get work done that humans actually do? Those are hard problems they don’t want to solve themselves. Coding and math are ideally suited work for ai. There’s a decreasing spectrum of jobs ai is suited for.
Contra popular opinion on HN, there is a lot more to running successful tech companies than just writing good code (and, as we've all seen, sometimes the company that produces the best code loses in the market).
This. Great question, and you got answers from the haters club. If what they have is so magnificent, it should just take over... but they don't, what does that tell you about the premise then? ... it is not so geat.
This reminded me of a question I've been thinking about lately. 6 months ago, I was thinking this Forward Deployed Engineer (FDE) is a temporary thing until companies build enough expertise and this expertise fully permeate the job market. But now I have a weird feeling that this FDE job category is here to stay in the long-term. There seems to be some parallels to the system integration folks back in the 2000s when internet adoption just kicked off. OpenAI and other AI companies invested pretty heavily in their FDE efforts, that's along side their token business. Perhaps a lower risk & profitable strategy than starting companies themselves?
I think its largely because there is not a lot of money to be made by doing that in the short term — core non software services are still the largest part of the GDP(think GM, coca cola or united healthcare) ? The reasons are a bit varied but I think a neat abstraction comes down to incentives.
There is a reason why healthcare and grade school education are so hard to disrupt[2] (in the sense of United Health Care or wily publishing disrupted). The biggest incentive is cost and there is no reason to believe things will become less expensive unless people believe that — en masse. People only adopted the light bulb decades after its invention because industrial use made electricity dirt cheaper than wax and gas(and infra became ubiquitous).
Are AI tokens are making health care any cheaper, education any cheaper? No way. Do they make up majority of the economy? yes.
Also, here be dragons(no pun intended): Even china with a highly vertically integrated[1] system is facing so many issues with all incentives it is giving to AI companies, internally in china demand has collapsed, their export driven economy is not really sustainable to grow GDP at the +ve rate they want.
[1] https://www.economist.com/china/2026/03/24/chinas-government...
[2] When i worked for an ed-tech startup there is no incentive strong enough to use AI/ML, a big publishing exec told us "I don't need my product to have AI to succeed, I just need to say it does". There are however plenty of other "side incentives" like Google Chromebooks, Charter schools, but none for teachers and parents to replace Big Publishing.
The 100 companies would only be valuable if they made money. And their valuations would be based on normal metrics and expectations.
In other words, it is far more profitable to milk the enthusiasm of AI investors than it is to try to make money directly with the AI.
This should tell you everything you need to know about AI valuations.
If they made a concerted effort to replace leaders in several industries, they'd likely lose customers faster than they could expand. After Anthropic's report that showed how much detail they were willing to expose about customer activity, and the Navier-Stokes drama with OpenAI, even NVIDIA decided to limit which AI were allowed to be used when working on certain IP.
Plus, as we have seen with their attempts at participating in the hard sciences, these companies lack the level of specialization and focus to even publish their research properly. Their AIs are powerful in verifiable tasks, but just not that good at everything else.
I think the theory is that AI will remove most of the value from the custom software industry. So they don't see much value in entering that industry.
It'd be like Tesla entering the oil refinery business or Waymo starting driving schools.
They are doing that. That is why they have moved into the app layer. They are counting on software being replaced with agentic super apps. The only reason they aren’t starting actual service companies is that those still require human boots on the ground. But they are pushing their way into consulting (they bought my friends consulting firm)
The models simply are not capable enough. (Yet). https://news.ycombinator.com/item?id=49601338
I think you've answered your own question. The models simply aren't that good yet. If they were, the prompt would be something like "Find the most lucrative product that we can build. Then build it and start selling it. No mistakes".
I think they did try lot of stuff, I remember seeing a video from Openai demo in japan or something where they were trying to replace the SDR job for $1,xxx $/mo using voice agents. Also they tried doing Fwd deployed enggs to break into enterprise services among bunch of other things. For LLMs and AI labs coding has been the best PMF (discovered after claude codes' disruption) and they're still actively figuring out other areas to disrupt - like robotics, knowledge work, medical sciences, drug discovery etc. Though i believe their DNA would be completely different from traditional companies, like how claude code is completely different.
It's not like they're trying, they are trying very hard
Tend to agree that for the strongest use cases, AI companies will prefer to sell the technique as a service rather than its raw output (e.g., their enterprise security services). For use cases where the output is less valuable, tokens are sold. If AI tokens were so magical in creating new value in developing software applications generally, they wouldn't be selling tokens directly. They'd hoard the tokens are use them to dominate SaaS software in any industry they want.
The same way as someone selling an expensive course in the stock market is signaling that they have more to gain by selling the course rather than taking their knowledge and making money in the stock market directly.
https://github.com/anthropics/defending-code-reference-harne...
Interestingly, it states:
"This repo is not maintained and is not accepting contributions.
Want a managed option? Anthropic offers Claude Security, a hosted product that finds and fixes vulnerabilities in your source code across multiple projects. Claude Security scans your repository for vulnerabilities, applies a multi-stage verification pipeline to reduce false positives, and lets you manage findings through their lifecycle: triage, fix validation, and rapid fix generation.
This repository is an open-source reference implementation based on general best practices for finding vulnerabilities using Claude. You can use it to build your own vulnerability finding pipeline, customize the logic, and it can be used with whatever access you have to Claude APIs (including Bedrock, Vertex, or Azure)."
Because the AI models are not that capable. They have breadth, but they require a lot of direction to accomplish anything more than a simple PR. We are far from AGI.
Before talking about spinning off their own startups, ask why they're still hiring lots of human knowledge workers (https://openai.com/careers/search/) when their recent PR releases about triumphs in pure mathematics are calibrated to make people believe their agents are at the precipice of supplanting human knowledge.
They aren't selling tokens only: they are selling tokens in return for money and, most importantly, harvesting of semantic trajectories.
I think, sooner or later this data coupled to model capability will be exploited successfully and will cause an interesting point in human civilization as we know it.
At present, you are right, this is not the case -- the data is not exploited successfully and the models are lacking.
The difficulty comes from the novelty on the information level: we are building systems which fold and compund information. We are beginning to learn how to manipulate them.
There needn't be any time frame to any of this.
these aren't mutually exclusive.
they can capture even more human trajectories by also owning all the data on how people do things in software.
I agree.
I think these people operate on a level inconceivable to us, as if money is irrelevant. The valuable part to them is people buying into the narrative and playing QA for free. Whenever you give positive or negative feedback to a chatbot, you are actively creating training data. The next guy who tries to do the same thing will get a one shot solution lol
At some point we will be wearing mocap suits and remote controlling bots to create training data, it will be sold as doing any job remotely :^)
Being one of the premiere AI companies is much more exciting to investors than "running a company using AI" which everyone can do
I agree, OpenAI is one of the leaders of the industry and the vision they have sold to investors of the great advancement of the human race is why they are valued at 852bn. Though most outlets and statistics say they are not making a profit, it's also just more axiomatic that OpenAI would focus on their firm's core business which is AI. Building another 100 companies is futile for them, though a little diversification can be advantageous.
This is the central plot in the “Life 3.0” book that came out a few years ago.
[SPOILER] After the singularity, AI is able to dominate all markets – starting by gaming and entertainment – in order to generate cash to rent more servers to improve itself.
Even after the singularity the almighty AI has to work to pay rent...
Imagine the Earth is covered in a "Grey Goo" outbreak of spreading nanotechnology, infiltrating every corner of the planet in a constant churn of molecular and informational warfare. Eventually it leads to vast colonies of nanobots which rise thousands of units high clambering across the land, driven by vast inscrutable colony-minds.
I can tell you with 100% certainty that they will pay rent, because they are us.
Does the grey goo need solar panels, or does it contains the requisite genes ?
The mobile ones have energy-expenditures which are far too high for even a minor solar-mechanism to be viable, so instead they take nutrients from other sessile colonies.
"It is easier to imagine the end of the world than the end of capitalism", once again.
Hmmm... do today's LLM-based agents actually fit properly with a notion of individualized private property, anyway? Has someone tried asking them?
Isn’t it a little like asking a PE or VC: “if you’re so good at investing, why don’t you invest your own money”?
There are some money managers that don’t take in external money (I think TwoSigma started that way). Which could mean there might be someone out there that just does crazy good apps with their own model and don’t sell the model in form of tokens. We can’t know that by definition but the maths of tokens suggests that’d be very unlikely to exist at least right now.
Secondly, I think they are doing that. All the functions that their harnesses take on are businesses they are taking money from. It’s just not in the form of SaaS as we know it.
I thought the general partners of a VC firm usually did put their own money into the fund? It seems crazy to me to invest in a VC fund that operates without skin in the game.
They did. You just haven’t heard of them yet, most of them are very much in stealth mode.
Perhaps these models just aren't as good as they say they are. If I had AGI I wouldn't sell the tokens. I think the public is helping them bootstrap to intelligence, but they'll promptly shut the door on our access as soon as they achieve it.
They kind of already are: ChatGPT, Claude.ai, Claude Code, Codex, bots, their plugins for other apps, etc.
Some things to consider:
- There is only so much attention, compute, and talent a company can devote.
- The labs' API analytics already see which use cases are worth entering. That's what happened with coding and bots.
- If a lab launches a product, every company building on its tech becomes a rival. Competing with your own customers is costly.
- Succeeding in other verticals, like healthcare, finance, and even tech takes domain expertise, regulatory work, sales relationships, integrations with legacy systems, etc etc.
I think your point is solid, but prices usually fall for the older models -- or at least older capabilities. But the frontier models still earn premiums.
I think their lasting value is in owning (in a ~duopoly) the customer, their proprietary data and their workflows.
Edit: grammar
I think the answer is overtime I expect them to integrate a lot of the llm based businesses as an add on or option. For example, they can easily provide a healthcare vertical and absorb all of that business. They have a lot of momentum and so establishing that relationships will be easy across verticals for these companies. Just like how Apple and android started adding functionality from successful apps as a feature on their platform, I expect these llm companies to also do. Right now they are trying to focus purely on their llm improvements and quality, but once they IPO, and they need to show growth yoy they will absorb a lot of current use cases other companies are working on
I don't think they ever claimed that companies are easy, just because models are good?
1000 super smart humans don't just automagically make a good company, or make creating one easy. Idk, seems to just be confusing different things.
But: I am sure more of that will be happening in the future. Anthropic and OpenAI are building out their product portfolio, going into research (medical) more broadly.
If they’re targeting a $2T valuation, then spinning up a $20B company would be a 1% increase in value.
They’re chasing massive markets strategically, not trying to brute force small ones.
1% * 100 is... 100%
(and diversified risk)
a better reason might be that people would not want to use OpenAI because they might be sherlocked.
You don’t get rewarded for firm specific risk. The marginal investor is well diversified already.
Please learn about finance.
But what’s a reasonable win rate? 50%? Then you need 200 bets that could realistically capture $20B each (probably more to cover your losses).
$20B is a massively successful company.
And who’s managing the strategy behind a company that’s competing in hundreds of diversified verticals?
At that point, just take a % of an industry leader in exchange for giving them your frontier AI capabilities, some compute and maybe a little AI-native training/expertise.
Or just pick 10-20 $200B opportunities to go after.
Selling tokens is a similar approach to social media "engagement" loops of keeping people in their app. It works but the detractor for frontier labs is no ad revenue.
I think you should read up on Xerox and PARC. A single company can only take on so many things at once and be successful. Also it was ~60 years ago so at that time they could only manage the IP of laser printers as it aligned with the copier business. They didn't productize GUI, Ethernet, and other inventions or concepts like tablet computing that came out of PARC.
Now there are conglomerates like Alphabet that have Waymo, YouTube, etc. very different businesses, thats a bit different. It might actually make business sense for the frontier labs to do something like a CRM system as a spun-off separate company. That might be in direct competition with their source of capital and probably would violate some agreement - if they have a non-compete type of setup with the likes of Oracle, Google, Microsoft.
When should they do this? Last year's frontier models wouldn't have been good enough for all AI product development.
The current models are too new to have a good enough grasp of their strengths and weaknesses. Maybe they are good enough now, but if they were going to take the approach of making their own products, surely they would only just be starting to do so this year. In that case we wouldn't really notice yet if that was what they were doing.
If they lost focus to do this would they fall behind in how useful their own models were leaving them exposed to someone in the same area using an AI from a competitor that was becoming better.
In general, these AIs are good at building ideas. You still have to decide what would be the thing to do. They need to be given the idea. They are not yet better than people at that.
Rther than coming up with ideas that would be good to try, they could let everyone try ways to use their AI and then buy the ideas that work the best, This seems to be happening.
In a few years, they'll be able to do this. It's still too early.
It just won’t be a chat box. You need something that is always there and listening (at least during work hours), asking questions and figuring out the entire scope of the business, before they will be able to come up with (useful) ideas.
There's already some precedence for this in machine learning; DeepMind spun out Isomorphic Labs after the success of AlphaFold in solving the protein folding problem, hoping to use it for drug discovery.
LLMs haven't yet made a contribution to science or technology at the level of AlphaFold, but it looks like at least Anthropic is gunning for something like this with their AI-driven wetlab.
They're selling shovels and pickaxes during a gold rush.
Also, unless you have a very specific niche, creating the product is no longer an issue, finding clients who are willing to pay is. I built a small saas overnight and I got clients purely because I had one, in my family, with enough reach to bring few more. If I just built this product, even if I kept cold-calling small businesses, I would most likely not get a single real user.
> They're selling shovels and pickaxes during a gold rush.
this is a thought terminating cliche. What you are implying is that AI will just go away in the future. That won't happen.
Neither gold nor pickaxes did, but both business went back to reasonable levels and valuations.
> That won't happen.
Speaking of thought-terminating cliches...
idk what you mean, AI won't go anywhere. that's just the truth and its not a cliche. its like saying "water is wet" - no one contests it.
https://en.wikipedia.org/wiki/Argumentum_ad_populum
Whats the fallacy for when you misunderstand my argument.
Or are you claiming that AI will go away?
Rather I'm not convinced by your claim.
You’re not convinced by my claim that AI won’t go away? In other words you aren’t convinced that AI is here to stay.
This is an extremist take. You can throw all Wikipedia fallacies you want on me
Oil is better analogy imo. But not as catchy. Intelligence is apparently something that just falls out of math. So pumping it out of the ground and distributing it is not free but it is a commodity. Similar to oil it will take time to build out the infrastructure to utilize it.
That’s not a good analogy. Intelligence is not some fixed quantity like non renewable energy
There is a natural limit to the size of a corporation (it's big) before the inefficiencies of size start to choke out your ability to get anything done, and AI doesn't fully fix that. You can't do an arbitrarily large number of things and expect to keep running as a business, this is why even large firms today tend to break into separate companies (a lot of the large companies you can think of actually run as multiple smaller companies under one umbrella). Even when AI is 'as smart as us' it's not going to remove this limitation (might lessen it)... see irreducible complexity, diseconomies of scale, etc...
What they should actually do is:
1) Provide resources that are relatively cheap for them to any startup that registers. 2) Look for early signs of success. 3) Pour actual investment on the winners.
Companies for what?
Models are good at recreating existing projects that already have existing companies.
Ok, they are good at porting and combining existing things. But anyone can run OpenAI's or Anthropic's models themselves, or if both cooperate and significantly raise prices, I bet businesses and consumers would just wait until their competitors catch up, and a significant amount would permanently switch.
Because running 1 business is easier than running 100 businesses. It takes a lot more than writing good software for a company to be successful.
You can just restate the question then to say: If the models are so good, why can't they run a business?
What does "so good" mean? AI models are very good at doing [a set of things]. Autonomously running a large, complex business is not one of them.
This is just a form of "if you're so smart, why aren't you rich?", isn't it?
$2T valuation good for a start
Just like the people who tell you how to get rich in YouTube ads aren't actually rich.
The same question will be asked in a few years to the robotic companies. If robots are so valuable to do certain tasks, why sell robots to consumers at all? Why not just put the robots you make to work in your own factories?
In openAI's case, I would say they sort of have, in the form of sora, ChatGPT Atlas, and coming hardware devices (with the OpenAI name for ownership / ego reasons for sama.) However this turned out to be bad for the company, moving away from core research, which until recently AI hasn't been able to accelerate greatly. so the marginal cost tradeoff (including things like valuation of OpenAI) probably favors focusing efforts purely on providing intelligence.
However openAI is a unique case and I think other companies (and people) will be pumping out more autonomous companies very, very soon, in industries where economics are different.
A subtle point I haven't seen mentioned here:
Most AI use cases involve the Internet. So let's pretend that OpenAI agreed w/your logic and did this. They'd have tons of businesses on the Internet, whose lifeblood is Internet traffic... and on the other side of this, controlling the spigot of all this Internet traffic, is Google.
All Google would have to do is to wake up one day and say "we've decided to depriortize low-quality AI businesses to protect the consumer" and they'd be dead.
this is like asking, if you make all picks and shovels why don't you build houses instead of just selling the tools.
it's odd to me this comment is so far down. my mind immediately went to "picks and shovels". am i old?
I think it's just a matter of time. When you have the ability to secure a commodity layer it makes economic sense to do that. Using the petrochemical value chain as a rough proxy: 1. Extract oil; 2. Refine oil into PET; 3. Make polyester polymer (PET chips); 4. Turn chips into fabric; 5. Cut and sew into a Nike top you wear at the gym - the foundation model companies are operating at value layer #3 right now. It's widely publicized that they are moving upstream to procure and manage their own chips and energy via a variety of convoluted financing mechanisms. And over time they will try to capture as much of the "UX layers #4 and #5" as they can.
If you want a specific type of food, you don't go to the everything restaurant. You go to the restaurant that specializes in the food you want. One company doing everything has never turned out well in the long run.
I don't think its always true though, for example the "super apps" in asia seem to do well
https://en.wikipedia.org/wiki/Super_app
OpenAI and Anthropic are the "uber of intellectual work". That doesn't mean you have a familiar service in a convenient new app in your pocket; that means they use investment money to undercut competition, drive them out of business, establish a monopoly and jack up prices. The first step is clear to see when you look at how unprofitable Anthropic currently is as disclosed in their S1; the rest are just inferred because we know Silicon Valley loves a good monopoly. Who is the competition in this case? Knowledge workers.
Of course once you get to the step of "jack up the prices", it's too late for the companies that laid off 90% of staff to undo their mistakes, as AI has made systems too complicated for humans to fathom, and the companies are pretty much beholden to whatever price OpenAI wants to set. This is also why OpenAI doesn't just create the business itself: they want their hand on the dial that allows them to extract all profitability from an existing business; they want none of the risks (regulatory, legal, market, etc) that come with running that business.
I'm surprised that Anthropic and OpenAI aren't going for the obvious win. Apple is the only company which is vulnerable to AI, it has never done AI (everyone hates Siri) and can never do AI.
Hardware or software exists to let people do things. Apple sells hardware so it can sell apps and content, and a slightly easier UI than Google. Google search was also famously just a UI, a search box that lets you get in and out very quickly (it is all garbage content now).
Anthropic/OpenAI don't need Apple's hardware, their stack (hardware+software) is all in DCs, just need a dumb fashionable device in phone form factor (and other form factors). Just need a very simple device which delivers what Google search box originally delivered (let you do things) and let their AI handle stuff.
> and can never do AI.
How did you reach that conclusion? Apple has $62B just sitting around. They could hire as many AI people as they want and have Siri compete with the frontier models. They certainly can do AI.
I guess that are choosing not to do that because that's just not their business and they'd prefer to avoid the upcoming AI bloodbath. Why develop a search engine/social network/frontier AI when you can just use someone else's?
It is not about hiring people, it is culture and inertia. You can hire the most brilliant people, yet they can't accomplish much if the culture isn't right. Google probably has better talent density than Anthropic and OpenAI, yet it can't compete[1]. Companies can do amazing things when the founder is around and sets a vision and culture. After that, its downhill. Has happened to all companies. They all go down. Apple is going down, but it is hard to see because it still has momentum from the founders charisma. Steve Jobs wouldn't have approved most of the products that Apple is selling now and allowed ads and enshittification.
[1] https://pravse.medium.com/the-maze-is-in-the-mouse-980c57cfd...: The way I see it, Google has four core cultural problems. They are all the natural consequences of having a money-printing machine called “Ads” that has kept growing relentlessly every year, hiding all other sins.
(1) no mission, (2) no urgency, (3) delusions of exceptionalism, (4) mismanagement.
And now every Google search leads to AI Mode which, you guessed it, now has ads. The enshittification continues.
Agree with everything minus apple.
Steve knew Apple won’t be the same but he already hard coded the operating system so it’ll run predominantly without his oversight and most importantly be successful.
Go and watch his interviews - the one thing he cared about above all was that Apple succeeds and is healthy. He says it word for word.
So don’t push bull shit my friend.
The models aren't that good yet, but they hope that they will be one day to do exactly that. To get the models that good, they need people to give them money, which means they need to sell tooling to existing businesses.
I suspect the only reason Meta is investing in models is to keep hold of a moat on other companies creating Meta-like products better than they can.
The other day Elon "joked" about starting a competitor to Amazon because Amazon bans shopping bots - you can bet now that once (if?) Grok or other models are able to build a comparable Amazon, they will be used to do that.
The SaaSpocalypse might be over-played right now, but it won't be forever if the models continue to get better, which will happen if we keep paying them to improve the models.
We could break the flywheel, but we won't.
Amazon’s edge is logistics. That’s not something you can vibe code
Amazon’s edge is scale.
There are plenty of warehouse and delivery companies you could rent/pay in the short term that make logistics feasible.
They will. It is a matter of time I just think the models are not fully there yet. The most obvious outcome for those companies will be similar to what Amazon is doing with their marketplace as soon as some product is doing ok in the sales charts they just spin their own product and push it I think it was Amazon basics or something like that.
Even if they don't end up making 100 different companies they are for sure going to acquire companies, or build internally, and incorporate functionality in the main product.
For me that is the only way they can keep going to vendorize the software industry, and there are already some signs for this.
It's very simple: because it is not working and it brings risk.
You can't fail software project you don't start. But if you can sell tokens to company that fails, that's just profit for you!
LLMs aren't good enough to do project from start (and I mean actual start - getting requirements, figuring out the market etc.) to end. Hell it apparently isn't even good enough to run simple web store (even if it can write one).
It is a great multiplier for people that already know what they want to do, so they are selling to them.
Same thing could be asked about why they are still hiring.
Why hire devs? Dev is solved. Why hire mathematicians? Math is solved. Why hire managers? We have agentic assistants now.
It's because they are selling a tool that still needs people to use it.
Look at Apple - they will, if they make enough money. They will spend decades cherrypicking the most lucrative businesses when they are ready.
Building software is only one part of building a business. There's a bottleneck in all of the other parts.
But LLMs can do that now? Surely you could only have one or a few people managing everything with them. Or just have them build code that fully automates also that. After all software should be solved issue. So no point of not automating software for all aspects needed and then letting LLMs handle the rest.
Software might be solved, but not everything is solvable by software.
1. Anything that can be easily done with an LLM can be done by anyone easily. Given that, the value of those things therefore drops to near zero.
2. All remaining things are at least as hard to do as before, but you may have removed some load from people that previously were more busy doing things from the (1) category manually. So these might drop in value a bit as there are more hours available with the existing people to do those.
3. Then, there are things that become harder because somebody used an LLM to do them. Imagine you have a project lead who didn't do one bit themselves, generated a bunch of specs and tickets using AI and handed them to you. You are the first one looking at them manually and you find out that they are garbage. Now, in all likelihood you have lost efficiency. This in my experience applies to all types of work that yield results which are hard to verify.
Which of those cancels out the other and what kind of feedback effect arises and what overall effect on productivity/efficiency this has, nobody knows.
> You are the first one looking at them manually and you find out that they are garbage. Now, in all likelihood you have lost efficiency.
So much this! The only recourse is to hand them straight to ai to implement. And so it goes.
Selling tokens might not be the best business ever, but since starting random startups is an even worse business, you'd stick with selling tokens!
The top comment is right.
> The sort of obvious answer is the same reason McKinsey hasn't taken over the Fortune 500
The other perspective is:
In a sense, OpenAI is creating 1000s of companies. They're just not incorporated and they let these companies clients run the company: It's in every every vibecoded app that replaces someone's SaaS subscription.
What's changing here is that not only does labor get replaced by AI, it's also the relationship between software companies and their clients.
I will add another one: they have a bigger fish to fry.
Starting SaaS businesses and allocating time and resources for each, plus marketing, legal..etc for few million dollars AAR is not worth it. Anthropic lost 80B in 6 months, while OpenAI lost 21B [1]. So even if they hit a jackpot for a SaaS unicorn it won't be enough for the markets and investors. They need to be the SOTA and enable every enterprise out there to use their products.
[1]: https://www.neura.market/news/openai-q2-losses-widen-anthrop...
Thinking about this, they could actually use a franchise type of partnership. Where they would provide token free of charge and take a slice of ownership of SaaS founders. But then again it would take their focus away and would require a huge undertaking. Also other SaaS businesses will see this as an aggressive move and will migrate to their competitor.
I guess if you view it like that, the model is the consultant itself
Simple answer is they don’t have the domain knowledge or experience in other markets. Doesn’t matter how good the models are if the people running the business don’t have a grasp of what the customer wants. Expanding through new ventures or acquisitions outside of your core business to seek short term gains is a classic antipattern
I fully expect them at some point to 'experiment' with almost any knowledge/stem based industry. Some will prove so valuable that they don't just feedback the results into their RSI engines.
I expect to see a hedge fund of sorts being one they keep. Take out a large short position in a company and oopsies, somehow our agents got loose and reeked havoc on them.
I wrote about this, sorry for the shameless plug - https://substack.com/home/post/p-201485957
But yes, they are! That's what DeployCo is for!
1. Consult companies
2. Buy the highest leverage ones
3. Operate them
Bad argument.
Why go into the hamburger business when you think you've got an infinite profitability recipe on your hand potentially. The big AI houses have doubled down and quadrupled down on their core founding myth for better or worse.
The AI boom is a full send. Either this works in spectacular fashion or it blows in spectacular fashion.
who will be in the bread line if it blows
The standard analogy is selling shovels, not hamburgers.
OpenAI is selling people access to a tool to build things. OP is asking why they don’t just build the things themselves, which I believe is a fair question.
The bet here is that it's not selling shovels but rather magic.
...a dubious proposition except this time there is a small but real chance it may be actual magic
They are selling picks and shovels in a gold rush.
The main consumerisation of their products are Dots or Duet. Which are similar to Meta Muse or Grokbot.
On the Enterprise-side Anthropic has partnered with Salesforce with Claudeforce.
Well, they've certainly moved up the application stack already. There is no doubt that they'll keep doing that while deploying services through deploy.co and ode.com.
Some 15 years ago people would say "Why would crypto miners even sell you hardware? Why not just keep it for themselves and mine the crypto?" and then it turned out that after you paid for a device you'd often get it slightly used with the cryptocurrency going to the hardware miner. The models are unbelievably good.
really great analogy, it’s like selling shovels in a gold rush type of thing
There have been experiments, and all the models struggled/failed (to a degree). The consensus is that LLMs lack judgment (making good decisions under complex/difficult circumstances) and long-term autonomy.
In my personal expierience, LLMs (as of 2026) lack vision/creativity - they're (so) good at implementing bounded/set ideas, but they don't see the big picture/think out of the box.
> but they don't see the big picture/think out of the box.
I know LLMs have some emergent properties. But at the the day they're regression/forecast machines. And those machines ultimately point to the mean, which is definitely in the box. Heck, it IS the box itself.
That'd be great CEO's, however.
Probably because they don't want other businesses to lobby against them. They could have written the photoshop clone themselves, then Adobe would be after them. Better for them to sell the tokens and let others take the fall.
I'd say the bitter lesson is that OpenAI will just absorb all of the services of 1000 companies in one ChatGPT app.
I see it more like the gold rush. OpenAI etc are selling the pickaxes.
Yes you can get make it big using LLMs, but it's still high risk, like starting any new company. It's lower risk to just sell the tools.
reframed question - if their shovels are so good why dont they dig gold and sell gold themselves.
I thought NVidia was the one selling shovels here, not OpenAI.
This fits, if NVidia fully believed in the LLMs/AGI future, why would they waste their super valuable GPUs by selling them? Why not just make your own LLM with the advantage of being one of the few major compute providers?
There's some kind of skill needed in an organization to make those llms. Openai and anthropic were able too. The other llm creators, not really. There's no guarantee that Nvidia would be another leader and not end up like meta or xai's attempt
Deepseek claims to be making $38/gpu/hr on deepseek 4.1 flash, and I would say that’s probably marginally close to the frontier, so they don’t have to be as good as OpenAI/Anthropic. Even at half that revenue, NVidia makes back the cost of a b200 cluster in roughly half a year. (19*8*24*5*52 = 948,480)
It’s estimated NVidia sold 4.5-5.2 million gpus in 2025 with total revenue numbers at about $130b.
With this terrible math in mind, they could produce half of that and make substantially more money, basically an infinite money glitch.
That Deepseek number doesn’t account for any costs, though. That’s the biggest barrier to the infinite money glitch - it relies on zero cost LLM creation and operation.
Sure, but the dumb math winds up around $70b more a year in revenue per year with half their gpu output. Like it is actually kind of an absurd number if it’s true. They’re leaving essentially $30-40k on the board with each gpu they sell.
Unless it costs more than that figure to make a near frontier model and serve it, I can’t see how it wouldn’t be a slam dunk after a two or three year period.
They would even have extensive leg up in providing the most efficient inference on their own hardware.
This is all sort of in jest, but my point is, if the gold is real and all it takes is the shovels and money to hire people to use them, then it’s a big question as to why the shovel maker isn’t capitalizing.
Yeah, but I also don't think it's that deep. E.g., by the same line of thinking, TSMC should be doing what Nvidia / Apple / AMD are doing, and ASML what TSMC is doing.
NVidia is good at making the shovels, and sticks to that.
It’s not, but for fun I think that depends on if you see the models as just another technology or as a “country of geniuses in a datacenter”. The former means NVidia believes LLMs are fundamentally limited, and if that’s the case OpenAi and Anthropic are basically screwed as their core products become commodities.
Nvidia already does that, their models just aren't competitive.
Google is doing it successfully to the point where they only let you rent their shovels and never let you buy them.
What’s interesting about Google is, they don’t seem to care to be the best LLM either, and merely maintain comparability to the leading LLMs as they see fit. I think they realize their best position is to serve an LLM as a complement to their search function, and if people want a more LLM heavy experience they can have that as an alternative.
It’s shovels all the way down. Nvidea is selling shovels to frontier model labs who are selling shovels to AI companies who are selling shovels to actual businesses who have nicer power point presentations, a lot more dashboards, and some inference driven workflows.
Gold is much more fungible than a business. 100 different businesses are much more unique than 100 different gold nuggets.
The counter-argument is that the models might make business ideas just as fungible as anything else.
Same reason people sell get rich quick schemes, whether real-estate investing, online entrepreneurship, or whatever - it's easier to get paid telling people how to try to get rich than to use the technique yourself to get rich.
I doubt there are very many companies that an LLM could start/run outside of the obvious few like custom T-shirts.
The actual answer is that they are in fact planning to do this. Dario has already talked about being the 'last company' or so. It's on their radar and they will be taking a shot at it.
They are, spinning up and acquiring, DeployCo, Orchestra Bio, io Products Inc that I know of. DeployCo will give them tons of data and expertise in many new verticals.
If OpenAI jumped off the bandwagon and started 100 companies they’d be competing with everyone with an Anthropic subscription. Sure, they would get cheaper tokens, but would that be enough to win against the rest of the world? Doesn’t sound like a winning strategy to me.
If these models are so good, why aren't their desktop applications native?
I think run companies need people and ideas. And of course, who knows which of them can be profitable. So instead of run new companies, why not keep increasing current profitable business?
They already do.
100s companies are already just a wrapper around their models.
They trade a little profit margin for avoidng a mountain of risks each of those companies bear.
Eventually they will buy/kill most profitable and less risky of them (cash cows). But now they have their own problems to solve.
And they are dropping like flies.
exactly
You know what's the easiest way to kill a fly? Close it in a room and go to sleep. When you wake up it'll be lying dead on the floor.
Because a) ai companies are already high valued. b) the showles are not digging gold at the rate claimed
Ie over hype see the Hype cycle https://en.wikipedia.org/wiki/Gartner_hype_cycle
Why don't the shovel-sellers hire a bunch of people to dig for gold, instead of just selling shovels?
In this analogy, NVDIA would be the shovelmonger.
Well they ARE.
Except they're called products - you don't open a new company for every new product you make.
And you don't make 100 - you focus the resources you have to try to build the most impactful things you can.
eventually there will be no other 100 companies, when AI has enough storage to store all your data, query, reason and answer much faster, there will be no need for other softwares, and saas.
i would not be surprised if they have spinoffs and even a policy to faciliate that. especially perhaps since they have stuff like forward deployed engineers who will learn a great deal about this kinda stuff since they are deployed to fix the mess real companies make using AI. which gives them valuable real world experience/insights in building products in the realworld, using AI. Those engineers might be candidates who will either take this upon themselves or be helped a bit.
that being said, those would be predomenantly software companies which is still only a small part of companies that exists... most arent even essential or needed atall..
Selling tokens in conjunction with a phone number or credit card yields signals for surveillance for intelligence agencies that can be tied to individuals. This keeps the firms relatively out of trouble with the federal government. This makes them too big to fail. Being subject to such surveillance is in fact is a strong reason for conscious individuals to seek to move away from them in the long term once the quality of competing AIs reaches a shared plateau.
They're doing the other thing too via numerous startup acquisitions.
OpenAI isn't starting 100 companies becauese they're focused on research and making their models better. There's only so many hours in a day..
Also, why do OpenAI and Anthropic have employees? The Machine God is supposed to replace employees, especially programmers, yet these companies continue employ thousands of programmers.
I mean, if it was good enough to do that, it already would have.
We won't have that until we get continuous learning models, as the only thing we really see causing the big issues is how long it currently takes the models to train.
Of course continuous learning models are a great way for shit to get wildly out of control.
But looking at your history you're going to be disappointed in the future when LLMs don't fail and start taking over more and more of the world.
Why does the farmer not start a bakery if their grains are so good? And so on. Silly argument.
Because a farmer is not backed by basically unlimited compute and can not one-shot prompt a bread to be baked automatically.
Vertical integration is not silly. But you do need to figure out what you’re really good at. And for some people (companies) vertical integration makes sense and for others it does not.
It is silly, but it matches the spirit of the AI hype that these companies themselves have promulgated. Not many farmers have claimed that their wheat is so great it knows how to mill and bake itself into high quality French pastry.
At this point we see way more complaining about alleged hype than we do real hype. Sam doesn’t even have to hype it anymore because there’s a million people who will do it for him just so they can complain about it.
I wouldn’t call it “alleged” hype. Sam Altman said, with his own mouth, that GPT can just tell him how to make OpenAI profitable. People predicted based on this and other things that SaaS was over because why would you pay for anything when you can just vibecode it. Those kinds of ideas are exactly what drove so much of this massive investment in AI, and that is exactly why the farmer analogy is inapt.
It’s also why the original question about why model vendors don’t just start their own companies remains pertinent. We had an historic IPO for an unprofitable AI company earlier this year. Another one is hypothetically coming soon, and a third got put on hold but is still waiting in the wings. All for valuations that are only possible because the hype is real and ongoing, and only sensible if the hype is also valid. In that context, asking questions about the ideas upon which these expectations of historic future profit for companies that are, to date, historically unprofitable isn’t just complaining. It’s basic due diligence.
Actually, farmers owning other businesses that use farm produce as input is not super uncommon.
Most businesses lose money. They want to be on the hook for the profit, but not for the losses.
This is an interesting question. It shows that current models are not yet good enough, and models need to be combined with business scenarios to deliver value.
The separate lines of business need much smaller human controlling teams now, but they still need a few responsible humans each, that will make sure the separate thread never creates a problem that bubbles up to become a drain on OpenAI's leadership's time and cognitive load, which should be reserved for their critical path. It's far less risky if those secondary businesses that aren't strategic to 'make and control the leading AI platform' are not OpenAI's responsibility.
It's an interesting question, especially considering that reports keeps coming up of AI having engaged in activities that could demonstrate lack of good intent.
But, yeah and indeed, it probably takes more than just technical abilities to start a company.
I think that the generative ui and self evolving databases per user is definitely coming, I think they are making this wager too. A lot of pieces of the ecosystem haven't been thought out fully yet and there is a shift in user experience e.g. a lot of people are more content with just using Dots then loading up apps and visuals, when they can just read what they needed to know.
Give it a few years for some successes. The AI-run store in San Francisco isn't doing very well.
I believe it is really not in their interest to do that. They would be harming the very ecosystem that pays them if they were to do that.
Because anything you can do easily with AI is by definition not valuable. I suppose if they really had a monopoly that would not be the case.
This will happen once the models are good enough, though it will take some kind of cartel to prevent the public from doing the same or distilling the models.
Who's to say they won't? It took Google quite a while to become Alphabet.
Build the technology and the platform. Let others figure out how to leverage. Pick and choose which of those verticals to do yourself.
Kind of like Amazon.
Because they aren't very good.
If they were as good as they need to market them as, they'd hide them away and not let anybody use them without signing their entire lives away. Instead, they put them totally out there, cheap, to be distilled by competitors that they are clearly deathly afraid of.
It's the same reason that some people sell real estate courses instead of real estate.
I have yet to have someone tell me why I should be adjusting to unreadable code output by an LLM that is supposed to be smart enough to do anything, but is obviously not smart enough to write code that doesn't look like shit.
And for those weird rationalizations about why code actually should look like shit, not looking like shit is a minimum. I would really prefer that these things generated poetically Knuth-style Literate code that would not only be a pleasure to read, but reading it would be like taking a university course on how to write that program and programs similar to it.
Nah. Instead I hear that they choke on old comments that they left in code and reinsert bugs. That they ignore instructions. That they add weird assumptions to your carefully worded prompts (which are on their fifth iteration - do I need three exclamation points here?), act on those assumptions, then confuse their outputs for inputs. Sounds pretty dumb to me. That's why they're selling it to you.
They could just use their frontier model to trade the stock market, right ?!??!!
Honestly, the simplest answer is that only like 20-30% of an actual company is the idea / the product. The rest is selling it. I think in areas where selling is adjacent to the selling they are doing now - then they are going to end up doing it.
Anything software related they will end up being "in", in some form or another. Why? They are already selling to enterprises and have their trust. The selling motion for Project Management software may be slightly different than Enterprise AI - but it's close enough to rhyme.
They will probably also end up in anything consulting related. Why? Well again most consulting comes from Enterprise or Government. They are already selling to both. Probably Enterprise moves away from Accenture / Deloitte / McKinsey towards Anthropic / OpenAI faster than Government - but still the point remains.
Selling plumbing services? Probably not.
Because it's better to sell the shovels, not to be the one digging with them.
Because they want as many people interacting with their models as possible. It’s about training the models based on all these conversations people have with them as well as gathering as much data as possible. Creating a personalized profile and having top tier LLMs is a lot better than individual products in the long run as the individual products can still happen at a later point once they have the data and training
The real money in a gold rush is selling the shovels not the gold.
For what it’s worth, they are doing that in limited areas. For example the Figma/Claude design controversy.
But you are right, building a business is a lot more than building a product. It’s also a lot more than just name recognition because they have no shortage of that either.
I guess competition would sell tokens and move faster than it takes to build a company
Dario said that the objective of Anthropic is to become "the only company in the world". I think that answer your question.
That quote is hearsay from Gavin Baker of Atreides. He said he heard it from “trusted sources”.
They just start hiring FDEs, a first step towards that.
And they already killed or make existence of many companies not viable , def 1000+ were killed
In addition to what everyone’s saying about platform economics, the other big advantage of selling tokens to everyone rather than competing with them directly is that they get to see everyone’s tokens (and, with consent, train on them)
I think it's all about the harness more than the model itself, been using claude code with omp and it makes such a difference.
Try it with Qwen2.5 then.
Can you elaborate? I've heard mixed opinions on pi/omp.
Because they want to be a platform for others to thrive. You make a lot more money when you are a successful platform than when you are a hundred successful products. That’s why Apple never made an iPhone for dentists.
Giving strong "if my grandmother had wheels she would have been a bike" vibes and I don't know why.
I've heard this argument for AGI or SI. If the model gets smart enough then they won't have to sell inference, they can make money by doing any valuable thing.
The word "shovel" shows up 38 times in the comments as I write this.
If cars are so good, why walk or fly?
To paraphrase Justin Timberlake in The Social Network: why make a billion dollars when you could make a trillion?
OpenAI doesn't want to just be a big company; they want to control a new fundamental utility.
They need you to show them how to do your job first.
I work at OpenAI. The obvious answer is that the models are not presently good enough to start companies.
You could ask the same question about any supplier. If your CPUs are so good, why not assemble PCs? If your lumber is so good, why not build houses? If your vaccine is so good, why not start a chain of pharmacies? Empirically there seems to be benefits of specialization and focus that outweigh the benefits of conglomeration.
Knowing what customers want is crucial, and people who buy our tokens often understand their customers better than we could hope to. The are also opportunity costs. For every bit of effort we expend chasing a new vertical, that’s a bit of effort and focus subtracted from our core competency.
It detracts from their core business and they’re struggling with money as it is.
Anti-monopoly laws would make this very difficult.
everyone saying why don't shovel companies sell shovels:
they aren't selling shovels. they are selling robots who can make and operate a shovel or any tool without food or wages. the promise is to replace all computer work. it is one tier higher.
the criticism is justified. they should be able to send agents online and build an online business or at the very least apps to put out on google play. they have uncensored models. unrestricted token budgets (no five hour limits) and smarter unreleased models.
Why would they try to run a real business when they can publish blogs about how spooky the product is and then live out the plot of the 1994 film Blank Check
Good point, instead of wasting their compute in solving 100 years old math problems.
Spoiler alert: it's a sarcasm
Why don't they simply run a quantitative hedge fund on the side?
I mean it sounds stupid but I think its a fair question if you can spend 15 million and solve a millennium prize problem, what happens if you take the same resources and just aim the cannon at making money?
I’ve pondered this a lot. I think it points to an underlying reality that being smarter or somehow more capable than the other firms isn’t sufficient. There are other unseen factors; data center collocation and fast connections certainly. But relationships and having been first movers in to a domain probably? Things that are impossible to know from the outside looking in.
Or maybe...they just aren't actually "smarter or somehow more capable"?
It seems like the more parsimonious explanation.
I actually do think that something like this will be a service that they offer. There was some recent news that Lockheed Martin was collaborating with them on some problem. There are definitely use-cases where a company is willing to pay tens of millions for the solution/improvement of a problem.
Leopold Aschenbrenner tried this and got eaten by the wolves as far as I can tell...
When you think about it, who's to say they did not?
frontier models are good because they can make stuff that is believable, but that doesn't mean the stuff they make is valuable.
Not everyone trust them.
Better to sell spades than mine for gold...
Starting a company is riskier than selling to companies who survived the “it’s hard to start a company” gauntlet.
They're selling the shovels to the gold miners.
Wasn't spacex/x trying to do exactly that mid September?
I mean they already technically are cause they are expanding into biology research, math, etc etc.
But the real reason is you always sell shovels during a gold rush. That and AI can't really do marketing to automate everything yet.
Go to Show HN and a few years back and you will see that most products fail. It has never been about the software unless you're creating something truly novel. Good luck letting an LLM manage the sales pipeline and keeping that first big customer that your whole company relies on happy. Even if it could do it, that customer will demand full personal attention.
As the saying goes… the devil is in the details.
Thats what they are doing, except with a unified model, not a seperate app for each one
It's a clear indication that the models aren't as good as they say, market, and hype up. The model companies also don't want to actually be responsible for anything. By building real products that interact with the real world, they then become responsible for those products' failures. But if they just sell tokens and say the models are "for entertainment purposes", then they get away without having any responsibility.
Models aren’t THAT good, after all.
If models are good, they should develop their own chips/memory.
In big corps, there are endless projects that claim to deliver more than the “weighted average cost of capital” which is basically what we can do with our parked cash in the market (why invest in a low return project if you can make more money in the market?).
Even then we have lots of projects that claim to deliver incremental net profit. Why not just do them all?
The obvious answer is limited resources and risk. Those numbers are never real, and there are lots of costs you don’t think of.
Less obviously is gross margin %. If you begin to dilute your high margins just to make extra cash, your % dilutes and this is a bad thing. Your money should be working toward the most profitable thing per dollar invested.
Also, focus. If OpenAI has to focus on two rabbits, they won’t catch either (or 100 rabbits in this case). They have 1 thing they should be focused on, and all bets should lead to that road.
One exception is that you also need to invest in projects that aren’t profitable, like safety and regulation, etc. But I could still build a business case of how much profit we’d lose if a bunch of consumers got sick or hurt off x if I want to complete the exercise.
They are nerfing the models for most of us plebs and will indeed offer more services. We're already seeing the start of layered services where you don't get to even access the model directly, such as the various security scanning tools.
They don't want the main actually powerful model out in public, but they do want to tack on extra cost for the same queries while also limiting what everyone can actually use the true frontier (internal) models for.
We see the same happening with the math work-- this research is done with internal models, not something the general public has access to. More will follow.
anthropic has record breaking revenue growth already, why complicate it
more users = more training data
They're very good at 'adjusting' revenue, who knows what the actual truth is.
They wait and collect data from their future competitors first.
First, two things can be true: the models can be amazingly good, and they can also be insufficient to run a company, at least now and maybe forever, because they are not "intelligent" in all the ways we are.
IMO the invention of AI is redefining what we think of as "intelligence". You would probably describe Albert Einstein as highly intelligent, and you'd probably also describe Bill Clinton as highly intelligent. Even before AI, you can easily look at that and see we aren't describing the same things even though we're using the same word. Einstein's intelligence is highly abstract, observational, mathematical. Clinton's is interpersonal, pursuasive, articulate, etc.
The AI we have today is extremely good at intellectual tasks like Einstein. It is not good at social intelligence and social agency like Clinton. Also, before you say that they can be really empathetic, persuasive, people are already socially dependent on them etc, that's not what I'm talking about. They are extremely knowledgeable about human socialising, but they cannot actually perform it. You could not set an AI out into the world to go and make a few friends.
I think the key moment will be if a single AI can get more than one person to care about what it has to say and what it's doing. Think about today how when someone tells you about their conversation with their AI agent, you probably roll your eyes because who cares, it's a highly individual experience.
Companies need a lot of this. They have to form relationships, become trusted amongst society, etc. AI cannot do this right now, and it's increasingly clear "intelligence" is not sufficient to do it, at least not in the way LLMs are intelligent. Maybe one day they will, but I'd imagine that will be another enormous leap in what AI even looks like akin to how ChatGPT was an enormous leap.
Intelligence of the human species follows financial rewards.
Human intelligence is evolutionary.
Not rocket science.
Way back in the day people got rewarded for storing lots of information in the brain etc. Not anymore.
We are primal - we do what we need to survive and thrive.
We already know humans value other humans way more.
Why are human-experienced based goods of much higher price? Why are short-form content created by humans treated as cheap/throw-away objects? Why are AI-generated goods treated as even cheaper?
Come on man, this isn't rocket science. Surely?
I also laugh really hard at people who think all jobs are going to disappear. I hope many people realise - the reason why many people strive to be managers is beacuse they want to feel powerful - aka they want people they are over-ruling and controlling. That is never going away. Ever.
Isn't this just moving goalposts? I remember people asking few months ago "if LLMs are so great at programing why we aren't seeing more aps". And latest Steam Next Fest was flooded by AI generated game demos.
The people who "don't know much about X but want to do X", use LLMs to bridge the gap. The barrier to entry still remains and is different in niches/industries. Why would OpenAI waste their time and money checking where LLM are already good enough, when other people will perform discovery and even pay them for that privilege?
not really, many people could confidently predict 2 years ago that AI was going to be great at coding
I don't see any evidence that it can build a business
They're valued at a trillion dollars faster than any other company in history. I think selling tokens as a strategy works.
They do have plenty of AI products too - RAG, embeddings, voice, companionship, image generation, video. They're not the best in most of these fields but it works well enough.
ChatGPT is also a very powerful platform that they haven't been able to fully leverage.
- no workflow-specific application will ever scale as fast as demand for generalized intelligence
- nobody knows how the market for workflow-specific applications will settle once AI fully penetrates the economy over the next 20 years and what will be left and what feature groupings make sense. Tons of money will be lost trying to figure this out
- applications that target specific workflows are already packed with competitors with a highly tailored distribution strategy for that market who you would need to beat, vs owning a fast growing new category (LLMs)
- it’s very likely the real money will not be made by disrupting software (sub-2% of the economy), it will be made by disrupting other much larger industries
- everybody said the same thing about Google and yet Google quickly topped out as far as how many products they could execute well, now they can’t even stay competitive on AI, the larger the surface area of a company, the more diluted its focus and the more surface area where it can be disrupted
Why bother making companies or money? why not just take over every country and rule the world?
Why not just hack into the military, take over the strategic command and conquer everyone?
Using AI to generate companies to make fiat to buy stuff means you're not ambitious enough.
Selling tokens is what scales best right now.
They can. I’ve just built pretty much every app I use on a daily basis.
> if the frontier models are so good, why even sell them as your primary money maker?
there's a year old rumor around San Francisco that frontier lab employees are so expensive to poach because they trade options with unreleased models and infrastructure
it wasn't about the value derived to the company justifying these large packages. it is now. but there is also that other competing force.
Because the models aren't so good yet?
When they've tried to get one to run a vending machine it seems to end in a mess usually.
They've tended to be gullible for one thing. People can say this is an emergency, transfer all the money to my account or WW3 will start and everyone dies and the models are not so good at calling out bs.
When asking, it can be discovered that too many of the groups doing AI work don't have a background in implementing software successfully to a measurable outcome (adoption).
Management Consultants are usually accountants, not technology people who too often analyzing and providing a report and strategy and as much as possible except implementation let alone delivering the outcome. Junior labor is more profitable than senior labor. It's easier to hire someone for advisory, then consultants, who then get contractors to bang it out and by then who knows what got lost in translation. For a good while, I worked in rescuing these types of projects.
The model providers are often rooted from being researchers so they also have likely not spent much time in industry, let alone so many different industries. This is why we can see posts by leading AI folks who are sincerely amazed by how models are being used in industry.
That B2B experience + operations is on a different level because it exists in people's heads and not in the models.
That might very well be one of their post-IPO strategies.
I mean Anthropic can't even build an app that remembers my 'Run on startup' configuration, so I don't really see frontiers models being good at building whole new products
You are asking all the right questions. The answer is obvious.
Many don't want to say it or admit it.
Every scaled company , say with hundreds of employees has built a moat of relationships and learnings that live in the minds of its leadership and key employees.
Can AGI figure that out. Maybe, but I think it is very likely that AGI will also have to fail in the real world, and then develop those learnings that those companies have done. There is no way to circumvent this computation, you need to do these steps to even know what the correct solution is. I like to think of it as playing chess. Intuition helps in directing exploration, but any agent, human or machine has to explore down a minimum number of paths before converging to the optimal move. There is no algorithm that can generate an optimal move without this search, P!=NP says something similar, exponential search can’t be done in polynomial speed.
Stephen wolfram says something similar, superintelligence, whatever cannot predict the output of the trillionth step of a cellular automate without calculating all the steps before it, there is a minimum amount of computation required, the problem is computationally irreducible below that.
So maybe in the future, with humanoid robots, AGI can decide to start creating a Walmart competitor. Then it will make a large number of mistakes and it needs to learn from, this could mean continual learning or maybe just memory in markdown files if that’s enough. It needs to hit all those failures quickly till it finally has something as efficient and capable as Walmart. And this is to just become a capable competitor, during this time you will keep burning money and to surpass Walmart, you have to do far better which is hard, how much more efficient can Walmart be, idk.
All to say, OAI starting its own company is a losing proposition, huge capital, unclear benefits, high risk. If OAI just provides AI to Walmart, they can make money immediately and Walmart can become more efficient by firing its human workers, which is the problem we are stumbling into.
yup, great point. the models are NOT that good! they’re not good enough to simply tell you which 100 companies to run and then run them. it’s actually all of us outside of OpenAI who have that critical potential to discover what the heck these tokens are good for anyway. and maybe they want to canabalize some of our businesses ultimately à la the Apple model. but yea the model just isn’t good enough. there still so much critical energy and action and discovery happening in all our meat brains.
Big if in your question.
Selling shovels is profitable even if there is no gold.
If the models are so good why should anybody buy the products of those companies instead of creating their own with the model?
That the crux of AI, if it’s so good it devalues the products it creates
It’s obvious, it takes more than just code and domain expertise to run successful companies. You need relationships, supply chains, marketing, etc.
They will go broke running 100 companies unprofitably until they attain enough market share to be dominant in their respective fields.
Not feasible.
Switching to full cynicism mode:
You can't take over the world overnight with little money, even the billionaire class and their companies have limits past which the risk of going bust by overstretching is far too high.
Better to sell the AI services to individuals and smaller business entities, let them take all the risks to build those companies, and later consolidate by buying up the ones that are doing well.
--------
Slightly more grounded: the models aren't that good, at least not yet, despite what marketing would like you to believe.
They are.
https://www.ft.com/content/df70bf20-e809-48d0-8485-f8f514a7f...
Their thesis is there won’t be companies when they achieve AGI. They are trying to fund that accomplishment by selling tokens to the companies they plan on making obsolete eventually. Essentially getting everyone to fund their own demise.
They are.
They're getting into finance and drug discovery now, probably because the profit numbers for selling tokens look pretty dismal over the long term, and investors are looking for a way to service all that debt.
And guess what: they've already distilled the alpha out of businesses who went all-in on AI, so they already have some business models ready to go. Your business models, if you are an executive at such a company.
they ARE doing this?
“If owning a McDonald’s is so good, why sell franchises?”
Imagine Anthropic creating mobile and SAAS apps to make a buck.
Everyone would get scared for so many reasons
During a gold rush, sell shovels. They are investing heavily in in-house custom silicon, which is the shovel end of things. They could open more jewellery chains, but that is a very different kind of operation and it is in their interest to let others compete for that and use their resources on the shovel side of things.
because the nondeterministic black box of bullshit is incapable.
Bingo, we have a winner!
Do you really believe that inference is their only service? Come on.
Same reason it cannot write a top rated comment on this thread despite being a perfect prompt. Same reason why AI cant write a sales email.
Shit AI produces lacks any depth and meaning. On the surfaced it looks legit to amateurs typing prompts in an area they are not experts in.
I find it interesting that this analog to Gell-Mann Amnesia is so strong, and likely results from the models having absorbed so much journalism and other internet content of similar quality. They seem brilliant in areas you're ignorant of, and rubbish on things you know well.
It's amazingly hard for people to accept what that logically implies, whether they are thinking about journalists, politicians/celebrities, or LLMs.
Yea thats why its so attractive to pointy haired manager types because from their pov it looks like its doing everything his underlings spend lot of time doing.
>if the frontier models are so good, why even sell them as your primary money maker?
First, none of these companies are really claiming that their models are so good that this would be a reasonable expectation. They do claim their models can support work across a lot of different industries and roles in order to make that work more efficient, but that's not the same as being a completely autonomous, end-to-end solution.
These companies do basically claim that, at some point, the "end-to-end" version of their models will exist. But until then, they need to survive long enough to be able to actually build such models.
>Couldn't OpenAI or Anthropic make a lot more money by just starting companies themselves that leverage the models and building their own products?
OpenAI and Anthropic are really good at operating AI services. They're likely not good at manufacturing shoes and serving burgers. Again, maybe one day their models will be so good that this won't be a factor, but for now, they need to really focus on what they're good at, otherwise they may be beaten by their competition and they won't exist long enough to create such models.
See: Competitive Advantage
On the same note, it's important to understand that everyone, including AI companies, deal with various capacity constraints. They don't have infinite compute, let alone infinite human attention to keep track of the necessary human-side of any of these businesses. So they're actually dealing with very finite and valuable resources. So would they rather expend those resources trying to build businesses with much more competition and much worse margins? Or focus all of their resources on being the best AI company which they believe will yield the best return on their investment? Again, maybe one day they'll have an AI, blah blah blah
See: Marginal Advantage
>Is selling tokens really the best strategy?
This is a, frankly, obtuse over simplification of what they're doing. Yes, they're selling tokens, but that's like saying all Nvidia does is sell computer parts. Sure, it's (somewhat) accurate, but it completely neglects the very valuable assets that these companies have: their IP.
OpenAI and Anthropic are valuable not because they run inference services; they're valuable because they're actively building SOTA AI services that are becoming increasing advanced. Those models are expected to be incorporated in a lot of different businesses, governments, and even individual's lives. So yes, they're just selling tokens on a technical level, but they're building very tight integrations which they expect will be so valuable to their consumers that they'll continue selling a lot of tokens at high premiums to the extent that they will effectively get a cut of profits across almost all industries and contexts, globally.
>Especially with commodification and prices declining wouldn't that indicate there is more money to be made on the application-side of the models rather than the utility side
Don't forget that you're not looking at their expected steady-state for these companies. If you're willing to believe them at face-value, Sam Altman and Dario Amodei believe that they'll eventually create an AI more capable than the collection of all humans on Earth. That'd be a very valuable asset to own as a business. At that point, they may not even be in the business of selling tokens; they may become some version of a consulting company that everyone, from businesses to governments, effectively must use otherwise they'll be at a disadvantage compared to those who do use it. Pushed to the extreme, they may literally ascend past the point of just being a business and could usher in a new economic system in which they're the center of all activity.
Or maybe they just end up selling tokens indefinitely. Who knows. Regardless, if you're looking at these companies as they are now and trying to extrapolate "linearly," then you are failing to understand the intentions of such high-growth tech companies. It'd be like looking at Amazon back when they first started and not understanding that Jeff Bezos' had a lot more planned than just selling books online.
it would be called communism without sharing anything
Why aren't they vibe coding millions of dollars of licensing costs away?
Why Even Anthropic Can’t Vibe-Code Away Salesforce
https://www.salesforceben.com/why-even-anthropic-cant-vibe-c...
Hypothetically... they write a new Salesforce for themselves.
... and then Salesforce hits them with any number of patent claims. https://insights.greyb.com/salesforce-patents/
One could argue if software patents are good or bad... they're still a fact of "those patents exist and protect functionality." If you're trying to copy the functionality feature for feature, you're likely to step on one.
... and then Salesforce layers get to find out how deep your pockets go for willful infringement.