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.
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’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?
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
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.
Came for this comment
yup that what claude does too
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!
In addition to this, most of the VC funds investing in OpenAI do have industry vertical plays.
The Firm is confederated.
> 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.
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?
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.
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.
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 real answer is that modern businesses depend very strongly on who owns the portal to a business (gatekeeping), and who owns brands, etc.
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.
Best response I've ever seen to this. Well said.
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.
Maybe companies should hire McKinsey to know what NOT to do
> 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.
>AI is brains
Is they though?
The pyramid needs a top!
> 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?
Same reason why Apple hasn't Sherlocked any applications sold for their platform!
Oh wait.
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.
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.
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.
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 .
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.
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.
>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.
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.
out source to India..Gone.
> 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.
give a list of industries killed please. that is a huge claim without any evidence provided (by you).
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.
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?
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?
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
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
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.
> 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.
Teams is dog water, but it's dog water that works.
> 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...
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.
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.
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.
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.
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
> 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.
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....
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.
>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
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.
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.
"On paper low debt" actually means.... low debt.
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.
> Investment isn't debt.
...yet.
SoftBank is very leveraged with …… debt.
One of the biggest AI investors.
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.
With investors like Softbank, who needs competition?
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.
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.
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.
The 10s were totally fake due to ZIRP.
> 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.
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).
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.
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
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.
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.
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.
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?
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.
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.
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.
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 merginal 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.
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.
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
I doubt if we will ever get AGI.
I asked my AI girlfriend whether AGI was inevitable and she said my question was brilliant and insightful.
Ever is such a strong 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.
And AGI is such an undefined word.
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.
Ten years ago people would have unambiguously identified what we had today as AGI.
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/
What does that even mean? They would have tried the models, noticed their limitations, and decide it's not AGI. Same as today.
Unambiguously?
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.
"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.
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.
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.
the real shovels are gpus
the real shovels for the real shovels is energy
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...
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.
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
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.
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.
> 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.
"sharing" their knowledgebase?
Nobody "shared".
> including the ops
Citation needed.
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.
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!
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.
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
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".
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 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.
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.
Forward Deployed Entrepreneurs.-
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.
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.
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.
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.
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.
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?
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
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 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 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.
I thought the agents were largely autonomous, and should replace workers?
If labor is not your constraint, you wouldn't be diverting your specialization.
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
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.
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.
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.
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)
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.
The models simply are not capable enough. (Yet). https://news.ycombinator.com/item?id=49601338
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.
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.
[dead]
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.
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.
[dead]
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
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.
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.
Imagine Anthropic creating mobile and SAAS apps to make a buck.
Everyone would get scared for so many reasons
frontier models are good because they can make stuff that is believable, but that doesn't mean the stuff they make is valuable.
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.
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.
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.
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.
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
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.
I'd say the bitter lesson is that OpenAI will just absorb all of the services of 1000 companies in one ChatGPT app.
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.
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.
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.
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.
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 ?
"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?
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.
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)
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..
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.
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.
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.
That'd be great CEO's, however.
> 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.
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 does the farmer not start a bakery if their grains are so good? And so on. Silly argument.
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.
Because a farmer is not backed by basically unlimited compute and can not one-shot prompt a bread to be baked automatically.
Actually, farmers owning other businesses that use farm produce as input is not super uncommon.
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.
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.
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.
Software might be solved, but not everything is solvable by software.
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.
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
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
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.
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.
When you think about it, who's to say they did not?
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.
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.
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
who will be in the bread line if it blows
It detracts from their core business and they’re struggling with money as it is.
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
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
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”.
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?
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.
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.
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.
Just like the people who tell you how to get rich in YouTube ads aren't actually rich.
They're selling the shovels to the gold miners.
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.
Can you elaborate? I've heard mixed opinions on pi/omp.
Try it with Qwen2.5 then.
As the saying goes… the devil is in the details.
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.
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.
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.
Models aren’t THAT good, after all.
Not everyone trust them.
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.
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.
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.
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.
Wasn't spacex/x trying to do exactly that mid September?
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)
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.
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.
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.
- 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
Good point, instead of wasting their compute in solving 100 years old math problems.
Spoiler alert: it's a sarcasm
they ARE doing this?
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
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
Why don't they simply run a quantitative hedge fund on the side?
Leopold Aschenbrenner tried this and got eaten by the wolves as far as I can tell...
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.
Starting a company is riskier than selling to companies who survived the “it’s hard to start a company” gauntlet.
Better to sell spades than mine for gold...
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.
Thats what they are doing, except with a unified model, not a seperate app for each one
If models are good, they should develop their own chips/memory.
They are.
https://www.ft.com/content/df70bf20-e809-48d0-8485-f8f514a7f...
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.
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.
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.
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.
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.
[dead]
Selling tokens is what scales best right now.
Big if in your question.
“If owning a McDonald’s is so good, why sell franchises?”
it would be called communism without sharing anything
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.
[dead]
[flagged]
[flagged]
[dead]
[flagged]
[flagged]
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...
[dead]
[flagged]
[dead]
[dead]
[dead]
[dead]
[dead]