OpenAI resets spending expectations, from $1.4T to $600B | Hacker News Reader

OpenAI resets spending expectations, from $1.4T to $600B

cnbc.com

225 pointsrandycupertino191 comments

> OpenAI is projecting that its total revenue for 2030 will be more than $280 billion

For context, that is more than the annual revenue of all but 3 tech companies in the world (Nvidia, Apple, Google), and about the same as Microsoft.

OpenAI meanwhile is projected to make $20 billion in 2026. So a casual 1300% revenue growth in under 4 years for a company that is already valued in the hundreds of billions.

Must be nice to pull numbers out of one's ass with zero consequence.

I like the little blurb at the end which said that Codex had 1.5 million users. So, if you can get each of them to pony up a mere $186k a piece, they can hit those revenue numbers.

Don't forget that Codex is free until March so the numbers are heavily inflated.

> Codex had 1.5 million users

I'm three of them and I never spent a cent on any llms, I doubt I'm the only one

Its a circular economy...He is talking about the money moving from Nvidia to OpenAI and back to Nvidia. You got to go with the flow...

He is counting on hundreds of husbands: https://xkcd.com/605/

How will Nvidia give revenue to OpenAI?

Nvidia gives money to OpenAI so they can buy GPUs that don't exist yet with memory that doesn't exist yet so they can plug them into their datacenters that don't exist yet powered by infrastructure that doesn't exist yet so they can all make profit that is mathematically impossible at this point - Stolen from someone else.

There are other forms of money transfer than revenue.

Yes but we are talking about OpenAI's projected revenue here, not funding.

https://www.bloomberg.com/graphics/2026-ai-circular-deals/

> Its a circular economy

Garbage in, garbage out, same as before.

1.4T was estimate by gpt4/5, 600b by gpt5.3?

they'll probably fix it just like they did fix strawberry

their estimates will drop by ~20x which will be their max

as underdog in the race they'll grab fraction of even that

where are they planning to get that much money from? by showing adverts for 14h before you can prompt?

> a casual 1300% revenue growth in under 4 years for a company that is already valued in the hundreds of billions.

Such a weird sentence. The correct causality should be: It's valued in the hundreds of billions because the investors expect a 1300% revenue growth.

And if we all buy umbrellas, then it will start to rain??

The metaphor for the original post was more like "You're already wearing a raincoat and umbrella, and you're forecasting a flood warning?" So, the flood warning (project revenue) may be completely incorrect, but it's not incongruous with the fact that I'm wearing a raincoat and umbrella (current investor valuation). :-)

I mean, if you go outside and everyone else is carrying an umbrella, it's probably going to rain.

Or the town has been hoodwinked by a smooth talking umbrella salesman.

Again!?!!

Mono! Doh!

If you go outside and they are burning witches, it's best to go along with it.

If you go outside and see people buying tulips, it doesn't mean that tulips are great investments.

Another example is how Isaac Newton lost money on some other bubble as well: https://www.smithsonianmag.com/smart-news/market-crash-cost-... [ The market crash which cost newton fortune]

So even if NEWTON, the legendary ISAAC NEWTON could lose money in bubble and was left holding umbrellas when there was no rain.

From the book Intelligent investor, I want to get a quote so here it goes (opened the book from my shelf, the page number is 13)

The great physicist muttered that he "could calculate the motions of the heavenly bodies, but not hte madness of the people"

This quote seems soo applicable in today's world, I am gonna create a parent comment about it as well.

Also, For the rest of Newton's life, he forbade anyone to speak the words "South Sea" in his pressence.

Newton lost more than $3 Million in today's money because of the south sea company bubble.

People often use that example, but Newton, for all he was unquestionably a giant of physics, was a bit of a weird dude and not 100% rationalist[1]. Additionally, just because he was a great physicist doesn't mean he knew anything at all about investment. You can be an expert in one field and pretty dumb in others. Linus Pauling (a giant in chemistry) had beliefs in terms of medicine that were basically pseudoscience.

Intelligent investor is a great book though.

[1] eg he wrote more than a million words on alchemy during his lifetime https://webapp1.dlib.indiana.edu/newton/project/about.do

> ...was a bit of a weird dude and not 100% rationalist...

That covers everyone. Especially and including the rationalists. Part of being highly intelligent is being a bit weird because the habits and beliefs of ordinary people are those you'd expect of people with ordinary intelligence.

Anyone involved in small-time investing should be considering that they aren't rational when setting their strategy. Larger investment houses do what they can but even then every so often will suffer from group-think episodes.

> Newton, for all he was unquestionably a giant of physics, was a bit of a weird dude and not 100% rationalist

The norms of "rational" science hadn't really been established yet. There wasn't really a clear line drawn between alchemy and what we would consider chemistry today.

That is what I used to think, but if you dig a little deeper I'm not sure it's quite that simple. If you read the link I posted, all that work on alchemy was not printed after his death because people examined it and deemed it "not fit to print". So it definitely seems that even at the time, there may not have been a clear line, but people felt that his alchemical writings were on the wrong side of whatever line might in future be drawn.

Newton was also definitely in favour of an empirical/axiomatic basis for science in general. If you read principia he proves almost everything[1] and of course he famously deformed his own eyeballs with wooden gadgets to do his experiments in optics.

[1] In fact pretty much the one thing he doesn't prove is the calculus, which Alex Kontorovich once said in a lecture on youtube that he has a pet theory that the reason that Newton never published the calculus was not the one everyone says about his rivalry with Hooke etc but that he wanted a rigorous proof first (which of course didn't come about until much later with Cauchy, Weierstrass, Dedekind etc for normal calculus and the 1960s for non-standard analysis to prove Newton's fluxions rigorously).

Newton knew a lot about investing for the time. He was a master of the mint for much of his adult life.

As I understand it, Master of the mint was more about knowing enough metallurgy to not be ripped off by people using weak alloys to smelt coins. It wasn’t like a modern central banker or anything like that.

> So even if NEWTON, the legendary ISAAC NEWTON could lose money in bubble and was left holding umbrellas when there was no rain.

The moral of that story is that being a legend or smart doesn’t count for much in investing.

perhaps tis not the rain but the sun they fear.

That would be compatible with them carrying umbrellas; https://www.etymonline.com/word/umbrella

Stray from the light long enough and they'll start calling blindness "market strategy"

This greatly overestimates the rationality of markets.

Hint: its a classic LLM misunderstanding; theyre protecting from sunshine. Theyre parasols.

They said casual, not causal.

I didn't read it wrong. And the illogical part isn't 'casual.' It's the whole sentence, especially 'already.'

How much money was WeWork supposed to bring in when they were valued at $50 billion and it dropped to $10b when they put out their S-1 and faced some public scrutiny for the first time? This happened before covid and the switch to WFH. Were their investors unaware of their actual finances?

This is important. 20 billion is just like a rumor? would love to see the breakdown in the form of audited financials.

Investors are valuing it at ~$500B, which already projects massive revenue growth. OpenAI is saying "actually we are going to grow 10x faster than that". And all of this is without bringing up the “profit” word.

Oracle said something very similar, a short while ago. Besides a short lived peak, it didn’t do any good to their stock thus market valuation.

>investors

The marginal investor does.

If investors expect Microsoft profitability that means their stock is worth 1/6th today what it will be in 5 years.

That is a cost of capital estimate of 40%.

Which points to investors not believing the company will be that profitable.

I am not saying investors don't think they will be profitable just they certainly don't believe that profitable.

Consequences come later friendo.

> and about the same as Microsoft

> Must be nice to pull numbers out of one's ass with zero consequence.

Seems accurate?

What they are saying is if Microsoft ends up buying the rest of their shares then i.e. Microsoft's total revenue by 2030 will be more than $280 billion.

I was a paying customer ($20 a month) until AI prompted a layoff in my dying field that is web design and front end design coding. Now everytime chatGPT yells at me about memory i tell it fine Im just gonna use Gemini! I bet a lot of ppl are doing the same thing as both sit at the top of the iPhone charts.

Today I got a feature request from another team in a call. I typed into our slack channel as a note. Someone typed @cursor and moments later the feature was implemented (correctly) and ready to merge.

The tools are good! The main bottleneck right now is better scaffolding so that they can be thoroughly adopted and so that the agents can QA their own work.

I see no particular reason not to think that software engineering as we know it will be massively disrupted in the next few years, and probably other industries close behind.

It really doesn't matter how "good" these tools feel, or whatever vague metric you want - they hemorrhage cash at a rate perhaps not seen in human history. In other words, that usage you like is costing them tons of money - the bet is that energy/compute will become vastly cheaper in a matter of a couple of years (extremely unlikely), or they find other ways to monetize that don't absolutely destroy the utility of their product (ads, an area we have seen google flop in spectacularly).

And even say the latter strategy works - ads are driven by consumption. If you believe 100% openAI's vision of these tools replacing huge swaths of the workforce reasonably quickly, who will be left to consume? It's all nonsense, and the numbers are nonsense if you spend any real time considering it. The fact SoftBank is a major investor should be a dead giveaway.

> In other words, that usage you like is costing them tons of money

Evidence? I’m sure someone will argue, but I think it’s generally accepted that inference can be done profitably at this point. The cost for equivalent capability is also plummeting.

I didn't think there would need to be more evidence than the fact they are saying they need to spend $600 billion in 4 years on $13bn revenue currently, but here we are.

Here you go: https://www.wsj.com/livecoverage/stock-market-today-dow-sp-5...

Right, but if OpenAI wanted to stop doing research and just monetize its current models, all indications are that it would be profitable. If not, various adjustments to pricing/ads/ etc could get it there. However, it has no reason to do this, and like all the other labs is going insanely into debt to develop more models. I'm not saying that it's necessarily going to work out, but they're far from the first company to prioritize growth over profitability

Nope. The only "all indications" are that they say so. They may be making a profit on API usage, but even that is very suspect - compare against how much it actually costs to rent a rack of B200s from Microsoft. But for the millions of people using Codex/Claude Code/Copilot, the costs of $20-$30-$200 clearly don't compare to the actual cost of inference.

This meme needs to go in the bin. Loss making companies love inventing strange new accounting metrics, which is one reason public companies are forced to report in standardized ways.

There's no such thing as "profitable inference". A company is either profitable or it isn't.

Let's for a second assume all the labs somehow manage to form a secret OPEC-style cartel that agrees to slow training to a halt, and nobody notices or investigates. This is already hard to imagine with the amount of scrutiny they're under and given that China views this as a military priority. But let's pretend they manage it. These firms also have lots of other costs:

• Staffing and comp! That's huge!

• User subsidies to allow flat rate plans

• Support (including abuse control and handling the escalations from their support bots)

• Marketing

• Legal fees and data licensing

• Corporate/enterprise sales, which is expensive as hell even though it's often worth it

• Debt servicing (!!)

• Generating returns for investors

Inferencing margins have to cover all of those, even if progress stops tomorrow and the RoI to investors has to be likewise very large, so margins can't be trivial. Yet what these firms have said about their margins is very ambiguous. As they're arriving at this statement by excluding major cost components like training, it's not clear what they think the cost of inferencing actually is. Are they excluding other things too like hw depreciation and upgrades? Are they excluding the cost of the corporate sales/support infrastructure around the inferencing?

To be clear, it's absolutely impossible for OpenAI and the others to stop. The valuation and honestly the global markets depend on them staying leveraged to the hilt. So they're not going to stop. However, the point is that the models are genuinely useful and people pay for them, and if we reset the timeline with a company that has just the current proprietary models, they could turn a profit. That might involve charging more than they do now, etc. But this is much different than OpenAI, specifically, trying to turn a profit today, which wouldn't work for many reasons.

But also, "profitable inference" IS a thing! "Gross margin" is important and meaningful, even if a company has other obligations that mean it's overall not profitable.

"profitable on inference" means "marginal costs of inference are lower than revenue". It is very common to distinguish between upfront costs vs. marginal costs when judging the economic viability of a business.

You mention "debt servicing", but OpenAI has no debt. All the money they have raised is equity not debt.

Indeed. Many of the posts I see on here are hilarious.

Have any of you tried re-producing an identical output, given an identical set of inputs? It simply doesn't happen. Its like a lottery.

This lack of reproducibility is a huge problem and limits how far the thing can go.

Determinism in agents is a complex topic because there are several different layers of abstraction, each of which may introduce its own non-determinism. But yeah, it is going to be difficult to induce determinism in a commercial coding agent, for reasons discussed below.

However, we can start by claiming that non-determinism is not necessarily a bad thing - non-greedy token sampling helps prevent certain degenerate/repetitive states and tends to produce overall higher quality responses [0]. I would also observe that part of the yin-yang of working with the agents is letting go of the idea that one is working with a "compiler" and thinking of it more as a promising but fallible collaborator.

With that out of the way, what leads to non-determinism? The classic explanation is the sampling strategy used to select the next token from the LLM. As mentioned above, there are incentives to use a non-zero temperature for this, which means that most LLM APIs are intentionally non-deterministic by default. And, even at temperature zero LLMs are not 100% deterministic [1]. But it's usually pretty close; I am running a local LLM as we speak with greedy sampling and the result is predictably the same each time.

Proprietary reasoning models are another layer of abstraction that may not even offer temperature as knob anymore[2]. I think Claude still offers it, but it doesn't guarantee 100% determinism at temperature 0 either. [3]

Finally, an agentic tool loop may encounter different results from run to run via tool calls -- it's pretty hard to force a truly reproducible environment from run to run.

So, yeah, at best you could get something that is "mostly" deterministic if you coded up your own coding agent that focused on using models that support temperature and always forced it to zero, while carefully ensuring that your environment has not changed from run to run. And this would, unfortunately, probably produce worse output than a non-deterministic model.

[0] https://arxiv.org/abs/2007.14966 [1] https://thinkingmachines.ai/blog/defeating-nondeterminism-in... [2] https://learn.microsoft.com/en-us/azure/ai-foundry/openai/ho... [3] https://platform.claude.com/docs/en/about-claude/glossary

Appreciate the response. I agree that non-determinism isnt a bad thing. However LLMs are being pushed as the thing to replace much of the deterministic things that exist in the world - and anyone seen to be thinking otherwise gets punished e.g. in the stock market.

This world of extremes is annoying for people who have the ability to think more broadly and see a world where deterministic systems and non-deterministic systems can work together, where it makes sense.

Yeah, I think you're right that LLMs are overused. In most cases where a deterministic system is feasible and desirable, it's also much faster and cheaper than using an LLM, too..

LLMs have randomness baked into every single token it generates. You can try running LLMs locally and set the temperature to low and it immediately feels boring to always have the same reply every time. It's the randomness that makes them feel "smart". Put it another way, randomness is required for the illusion of intelligence.

Im fully aware of that. However, this illusion is a dangerous mirage. It doesnt equate to reality. In some cases thats OK. But in most cases its not, especially so in the context of business operations.

What was the feature and what was the note?

It was a modest update to a UX ... certainly nothing world-changing. (It's also had success with some backend performance refactors, but this particular change was all frontend.) The note was basically just a transcription of what I was asked to do, and did not provide any technical hints as to how to go about the work. The agent figured out what codebase, application, and file to modify and made the correct edit.

That's pretty neat! Thanks for elaborating.

Yeah but was Cursor using Claude? What's the moat that any of these companies have that prevents me from using another LLM?

I have used AI a bit, like it for a bunch of use cases. But god damn, these numbers are so big. Gotta wonder, are the returns even worth it? RAM prices up, electricity prices up, hard disk prices up… Maybe this is the price to pay for “progress”, but it sure is wild

Simple - they returns are not worth it. :-)

You're missing one point: they are just talking about revenue. Nobody said something about making profit.

I think he meant for Anthropic?

OpenAI is a bet on LLMs replacing a large chunk of the labour force in whatever sector it’s best at replacing. It’s essentially looking to get companies to pay $5k-$10k a month to have coding agents replace the output of a single software engineer.

If the S-curve levels off below that level OpenAI will be an unsuccessful company.

I, too, can make $280B in revenue by 2030 (by selling $10 bills for $5 (as long as I bamboozle enough investors into giving me sufficient capital, of course)).

I honestly don't think that sounds terribly outrageous.

OpenAI and Anthropic aren't building companies that aim to be API endpoints or chatbots forever, their vision is clearly: you will do everything through them.

The gamble is that this change is going to reach deeper into every business it touches than Microsoft Office ever did, and that this will happen extremely quickly. The way things are headed I increasingly think that's not a terrible bet.

Doubling every year gets you to 1300% in 4 years. They have exceeded that for the past 5 years.

Don't forget that Anthropic also predicts similar revenue in in the next 4 years.

If all the AI revenue projections were correct, then 1% of worldwide GDP would end up at AI companies. Or said differently: you buy a sandwich for $5 and somehow AI gets $0.05 out of that transaction.

This is basically what happens with the advertising/social media giants (Facebook, Google, etc) because everyone needs msrketing, and mobile companies (Apple, Google) because they handle payments.

Honestly seems reasonable to me. ChatGPT alone will be be at least as valuable as Google, plus all the enterprise adoption around API use cases.

I don’t believe one word of what they say publicly.

I too have reset my spending expectations down from $1.4T.

best comment

A wise move.

ugh lower the interest rates Jerome, I'll do anything to tank the economy until you finally do.

I have done my fair share of misunderestimating before, but I've never been off by that much.

and that's just my doordash order..

The 1.4T commitments was over 8 years, not by 2030.

https://x.com/sama/status/1986514377470845007

the tweet has capital letters and punctuation - I doubt the authenticity of this post

It’s interesting that they felt the need to leak this to the press.[0] Some investors or partners (or LPs, board members, etc. of those) are getting spooked by the spending plans and rightfully questioning if the return is there. Putting it in public my feel like a stronger commitment (though I doubt it.)

Even with the revised numbers, I cannot believe that they’ll have $280bn in revenue by 2030.

[0]: You can tell by the reason the sources are granted anonymity: because the information is private, not because they aren’t authorized to speak on the matter

first bullet from the link

> After previously boasting $1.4 trillion in infrastructure commitments, OpenAI is now telling investors that it plans to spend $600 billion by 2030.

does the word "commitment" have a different meaning in this context? How do you cut a commitment >50%? OpenAI's partners are making decisions based on the previous commitment because.. OpenAI committed to it. I must be completely wrong because how does this not set off a severe chain reaction?

edit: as others have pointed out, the article is misleading. $1.4T was over 8 years or by 2034. 2030 is halfway to 2034 and $600B is not too far from half of $1.4T.

I think TSMC laughed them out of the room when they announced the original numbers. So maybe there’s no reaction now because everyone already knew not to trust OpenAI’s promises.

These were more like infrastructure suggestions.

> how does this not set off a severe chain reaction?

Just like you and me, Sam Altman can say anything he likes to say. To pump the investors' confidence, to make the US administration believe he's serious about AGI, or just to make himself feel good. It's not legally binding in any way.

You should never read it as "OpenAI committed to..." but as "Altman said these words..." and words mean very little today.

I don't get it.

A trillion here, a trillion there and all the AI companies are also telling us they're planning on wiping out 2/3 of jobs in the next 10 years? Nothing about the economics of the AI boom makes any sense.

I'm not saying it's not possible, but if we wipe out 2/3 of jobs with AI, who is going to be buying *all the stuff*?

Unemployed people aren't much of a demographic, and you can't just say UBI because that doesn't make sense either. You think the billionaires are going to allow themselves to be taxed heavily enough to support UBI just so that there's a market for people to buy stuff from them? That's nonsense.

Not trying to creep anybody out, but I just don't see a stable outcome for a society that doesn't need 2/3 of the population.

> You think the billionaires are going to allow themselves to be taxed heavily enough to support UBI

They will have no choice. Proletariat must not be hungry and agitated. Free legal MJ for everyone!

A gramme a day...

Keeps the doctor away?

It’s from a book called Brave New World.

If you want some light reading: 1984, brave new world, and atlas shrugged will mostly get you caught up on current events.

In fairness it's mostly Anthropic that is constantly banging the were taking your jobs drum.

Everyone else has been less explicit, likely because it's just not politically a good idea to keep pronouncing it.

It's part of Anthropics marketing though. Maybe to push the idea you can't beat us so join us?

Anthropic is running a similar marketing campaign as AWS/Devops tools which were trying to replace in-house IT. Pitch to the few that you can be 10/100x as productive and valuable on the hopes that they will push their organizations in this direction.

> Nothing about the economics of the AI boom makes any sense.

what if… MBAs turned from economics to a religion and no one noticed?

Unbelievable! Next you’ll tell us that Elon‘s self driving car promises were all just hype for the cult…

Econ has always been a bit faith based as it is.

Depends. The basics are testable. An explanation of scarcity is available in Basic Economics and should be required reading (Sowell)… but whatever this VC nightmare thing is… I agree.

The viability of a currency is nothing more than faith.

The singularity is nigh

>I'm not saying it's not possible, but if we wipe out 2/3 of jobs with AI, who is going to be buying all the stuff?

Money is just a proxy for access to resources. If a machine that is capable of replacing almost all jobs is really created then money will matter much less than access to said machine. Taken to the extreme to make the point, if you had a genie that could grant your every wish, what would you need money for ?

Yeah but what if that genie charges money for wishes.

> If you had a genie that could grant your every wish, what would you need money for ?

The things that a magic AI Genie will never be able to give you no matter how far into the AGI/Singularity things get. Such as Land, Energy, Precious Metals, Political and Social Capital, etc.

Georgism is the only way forward. Tax land, energy, metals, and other constrained natural resources, not labour.

Yep. Tax the resources that capital needs to produce the stuff. This is just a simple way to think about how we think about tax regimes etc can evolve.

'Political and Social Capital" don't belong on that list.

All of these things can be easily obtained with control of a machine far enough into 'AGI/Singularity'.

Energy, Precious Metals etc are not obtained with Money. They are obtained with human work and effort, all of which we are now saying is doable by the machine.

This is what a lot of people don't get. The magic genie that lets you wish for more wishes isn't a a rack of GPUs in a DC somewhere.

It's a domestic robot that can do full maintenance on another domestic robot.

Self replicating machines are the genie that grants you more wishes. They are the genie that can turn that land, energy, and precious metals into copies of themselves.

What happens when the genies realize that the meatbags that require 3000 kilocalories a day are just a net drag on the whole system?

Altman was just this week quoted as saying things in that direction...

Money is a proxy for the value of peoples' time. Since AIs are not people, they cannot create value.

Couldnt they create value by saving people's time? Like a shovel, or a bulldozer.

OpenAI is not going to pay off my mortgage, it’s not going to replace my roof, it’s not going to fix my car, and so on. Money is still going to be very necessary for goods and services.

I don't see what point you are making here. I responded to OP asking about "who is going to buy all the stuff". The people who would be concerned with that are by and large not stressed about paying house mortgages, replacing roofs or fixing cars.

And if they were, then the machine will just do all that for them. That's the point. The things you mentioned don't need intrinsically need money. The machine can fix or create whatever car, replace whatever roof, and build whatever house.

> The people who would be concerned with that are by and large not stressed about paying house mortgages, replacing roofs or fixing cars.

Well they should be, because actually putting 2/3rds of the workforce out of work in a short, sudden fashion is probably not going to end well for them.

> The things you mentioned don't need intrinsically need money. The machine can fix or create whatever car, replace whatever roof, and build whatever house.

What machine is this? It certainly doesn’t exist and won’t in the short timeframe these AI companies are predicting everyone is gonna be laid off. Maybe, maybe if the timeframe for “no one has a job anymore” happens over say, 100 years, things might go slowly. Over two or three years? Heads will roll.

>Well they should be, because actually putting 2/3rds of the workforce out of work in a short, sudden fashion is probably not going to end well for them.

Maybe. The ruling elite being a small fraction compared to the downtrodden masses is hardly a new manifestation. Regardless, money won't be the primary issue. Again it's just (intrinsically) worthless paper. All of its current value is a social construction and new ones could take its place if necessary.

If you are one of the 2/3 replaced by a bot, will you be able to pay for those things?

UBI is a more of a convenient trick we use to suppress the part of our conscious that tells us "wiping out 2/3 of American jobs is Bad".

Companies will save billions on those pesky health insurance premiums and payroll taxes...by paying for OAI tokens instead.

Then when the labor market is nice and hollowed out, the tokens will go up in price several-fold.

We're already there. Most of us have jobs that are just made up to fill the gaps after steam power and automation. In the future, we'll have jobs that fill up the AI gap. It's UBI, but more arbitrary so we can tell ourselves we're useful while group X is not.

Hmmm. I wonder if you get to choose between window maker or window breaker.

I’ve seen no discussion about what the social consequences are going to be if a predicted 2/3rds of people lose their jobs.

I don’t imagine they’re pretty.

"It may be difficult to know what role money will play in a post-AGI world."

Tripling productivity is not the same as thirding jobs. Demand will grow. People enjoy products and services.

Tripling productivity? Where? You can say this but where is this measurement being sourced. Every time I ask how LLMs can simply replace a real front desk assistant I get responses like: well that use case isn't viable because <enter excuse here>.

> "People enjoy products and services." ???

WTF does that even mean? Folks are so deluded with all of these "right around the corner" solves that AI has in store that they fail to realize how out of whack the numbers game has played out. In any other reality people would be scrutinizing Sam Altman at every angle. But because of some magical AI sauce the incomprehensible numbers now magically make sense.

But for a lot of us: it doesn't. If you're going to claim hundreds of billions in revenue, just a few short years from now, you better have a really fucking great product today. Not in 6 months, not in a year - but right now.

SaaS has not been displaced. Workers have not been displaced (other than shifting their salaries to AI spend which does not equate to worker replacement). Where does madness end? The only thing that makes sense is an implosion that will ripple all the way through many other markets which will now take years to fix.

In the ex-USSR internet segment there was a saying on nerd forums - "Don't know matan(1), will convert you to methane". Just saying :) . It's not like billionaire sociopaths had ever any issues with "useless" humans. Peter Thiel even follows a modern neo-religion along those lines.

(1) matan - mathematical analysis, as a reference to a widely known and hard to learn university course.

Welcome to late stage capitalism. Where non of the incentives have anything to do with helping people and reducing costs for things people care about - energy, food, healthcare, basic needs.

That's the wealthy outlook: solve climate change and that pesky employment issue with those oh so annoying plebs in one smooth go.

"who is going to be buying all the stuff?"

Easy - a greater portion of the world's resources can go toward the luxury market for the wealthy. This is already the trend.

It's dark but certainly not impossible to have a smaller and smaller group doing all the spending and keep spending the same, and to keep stability by force using technology.

I want no part of it.

Luxury products can only exist at the margin of a broader economy. You can have luxury sports cars only if you have a market and an entire supply chain for millions of non-luxury cars.

As far as an AI no jobs catastrophe, I doubt they have any idea / plan any better than any rando person out there. They just think being first puts them in a good spot.

I think the optimistic scenario is AI can do the jobs but humans don't become unemployed so the workforce is 1 lot of humans +2/3 that in AI. The humans are wealthier and can buy the stuff.

Like rather than Dilbert writing code, he gets promoted to pointy haired boss and manages an AI which writes the code.

The market is spooked by capex projections generally. Interesting that Microsoft, despite some apparent hesitation in 2025, seems to be still going all in on AI spend over the next several years according to the most recent earnings call.

MS, GOOG bonds being sold to fund capex still put them green $/employee, they will survive of not thrive.

OpenAI...not so sure, they need an IPO soon while public still is high off the double bull run post 2020

These numbers were always out of line with basic infrastructure constraints. People were talking like the US would build 50 new nuclear power plants in 10 years. And I believe we will not see $600B either, there are basic infrastructure, permitting, and power delivery limits.

However, we are all going to be paying higher energy costs for these ridiculous infrastructure claims. Utilities typically price out energy three years in advance. If they were protecting for twice as many energy sinks, that represents an enormous amount of generation capacity which needs to be accounted for in projections.

I saw a report that previous capacity pricing was $28/MWh/day. Latest numbers have shot up to $300.

Absolutely, and that's why we should be applying higher infrastructure fees to the permitting of data centers. The problem is that local governments want the tax revenue and are willing to screw over their constituents. This also goes in line with the decline of local newspapers, there is an epidemic of fraud and abuse of power happening in local governments across the country.

It's not unforeseeable that the US demarcates Special Economic Zones without environmental oversight or labor regulations to speed up the construction.

Nothing like superfund sites and The Jungle to feel like a true patriot again.

That's why Elon is betting on datacenters in sun-synchronous orbit.

OpenAI doesnt have a single model in top 10 models being used on openrouter.ai

Thats a weekly metric on https://openrouter.ai/rankings flagship chatgpt 5.2 model is at #16

PMF is now evolving when competitor models are either smarter or cheaper.

I hadn't heard of openrouter.ai. I have heard of OpenAI.

Is this like Windows and MacOS not being in the top 10 of distrowatch.com?

OpenRouter is the leading place to go to to get general purpose models of all sorts. It's fairly popular, and processes tens of trillions of tokens a year.

I think the primary issue is that the metric is tokens and not dollars spent.

OpenRouter is valued at >$500m and processes >$100m/year, 5% of which goes to them. Not that large compared to e.g. OpenAI, but it's the largest that doesn't produce its own models & with the largest selection I'm aware of.

Could also just be a bias in the audience

OpenAI, frankly, benefits from the "nobody ever got fired for buying IBM" phenomenon.

And there's a reason that OpenRouter has an OpenAI compatible layer highlighted not deep in docs, but on their Quickstart page: https://openrouter.ai/docs/quickstart#using-the-openai-sdk

The number of projects accessing OpenAI directly, who might only reach for OpenRouter once an alternative is desired, is unknowable (since OpenAI doesn't share usage statistics), but likely meaningful.

Well also because OpenRouter launched when OpenAI’s api was the only real game in town.

The number of tokens seen per model on OpenRouter is not a good measure of quality.

There are so many plausible explanations for why a particular model is or is not ranked in the top 10 by this metric.

Maybe people using OpenAI models are so happy that they don't care about other models and have no need for OpenRouter. Maybe OpenAI models produce fewer tokens, or are more expensive per token.

Your conclusion might be correct, but citing the number of tokens seen by OpenRouter is not very strong evidence.

If ChatGPT 5.2 were actually superior, developers wouldn't be overwhelmingly routing traffic to Gemini 3.1 Pro just 6 days after release.

I use openrouter.ai as the benchmark because it's the foundational API layer for innovator apps that are always the quickest to adopt new tech.

ChatGPT has 100x more interest on google trends than Gemini and OpenRouter combined, which in the context of this article is a much more relevant "popularity score".

But I don't think either are very meaningful when there are actual benchmarks to measure the quality of models on specific tasks.

So handwavy... 1.4T.. 600B. Pure marketing fluff to keep the hype machine going.

So are all the RAM, GPU and HD manufacturers going to honour their purchasing commitments?

Didn't oracle take out real loans and spend real dollars based on this commitment?

Wasn't it always an expectation, not a commitment?

If they didn't appropriately account for risk that the expectation would not pan out, well, that's on them.

How do you account for the risk of something that has ever happened before?

In the standard risk assessment and mitigation manner: imperfectly, but with the best information available at the time.

I dunno, but if someone is saying they expect to spend a vast, unprecedented sum of money acquiring an interdependent set of resources that current production could not come anywhere close to accommodating in multiple dimensions, and which money they don’t have and aren’t making in their current operations so that it would also require unprecedented fundraising on top of the other issues, you’d probably want to do some work to verify the plausibility before putting your own resources at risk for the chance of profiting from that spending spree.

Extraordinary claims and all.

The bankruptcy couldn't happen to nicer company.

What do we think? Is this possible without AGI level breakthroughs?

If we see a continuation or even a slowdown of the current trend, the technology overhang, lagging productization, and catch up from the slow adoption of AI by businesses probably gets them part of the way there, but I don't know about 1000% growth at this point... Seems kinda like they're banking on another breakthrough no? And if they don't get the breakthrough, the downside risks such as a competitor of some sort destroying their margin can't exactly be ignored...

This article is bad. It is mixing up capex and opex. OpenAI is projecting more spending on compute through their income statement now than they were 6 months ago.

I'm not convinced that companies venturing into the unknown really know more than anyone else, they just survive or don't. I've no idea what OpenAI is up to and honestly the public actions of Sam & Co seem like they feel kinda insecure about their position... whatever that position is.

Wasn't most of this spending going to ORCL?

https://www.cnbc.com/quotes/ORCL

Remember this press conference?

https://www.youtube.com/watch?v=IYUoANr3cMo

It is insane that they have this little of a handle on their buildout. It makes the $600B feel even more empheral.

This is more complicated than just hand wavy spending expectation resets. Other companies were taking these “commitments” and gearing up for capital investments to meet all that demand which is now vaporizing. That creates a big mess as the hype AI hype machine starts to unravel.

This looks very much like a careful move to deflate the bubble without popping it, but we’ve likely passed that point.

We are at the end of the exponential!

90% chance in 6-12 months spending expectations drop to $0.

Based on what? Each model so far has been noticeably better than the last, so I don’t see why the next wouldn’t be too?

"Can’t you just draw an exponential line on the curve?" - Dario Amodei, February 13th 2026.

But this time draw it for spending expectations.

Are you predicting that OpenAI is out of business in 6-12 months?

I'm not an AI booster, and I don't see "sustainable" in the current markets, but I'd take the other side of that bet!

Even if that were true, it wouldn’t make the growth exponential. There are many other trajectories out there.

Even if it’s not exponential, I still can’t see logic in it going to zero.

So does that mean they aren’t buying up all storage production capacity for the foreseeable future now?

They sure are humble for people who say they are building God.

From another comment I wrote here but I am gonna paste a quote I found from Intelligent Investor (page 13) from Isaac Newton during the hottest stock of his time in his country, South Sea company.

The great physicist muttered that he "could calculate the motions of the heavenly bodies, but not the madness of the people"

There seems to be a lot of madness happening in the world again as well. A lot of OpenAI claims make no sense except if we consider the world to have gone mad.

The bubbly nature of openAI and just doing whatever they think like doing with 0 regards to anything or everything including financials is a form of madness.

I was reading another comment and actually opened up the Intelligent Investor book to read the quote from there. I highly recommend that book although truth be told that I haven't read more than the first 50-100 pages as I quickly felt like passive investment is a great vehicle personally.

Am I the only one here who was amazed by the speed of improvement between 5.2-codex and 5.3-codex?

I feel that Sam is saying what investors want to hear, but the coding work it is capable of and how it improved with using the terminal (TerminalBench) in such a short time is something that I'm sure can't be seen by short term revenue projections. I'm sure the other AI companies are having the same speedups, but it's real.

The usual limit is of course the slop output that is not well modularized that makes it hard to do bigger things, and codex is terrible at refactoring into the right direction (it has no taste).

3x YoY growth in revenue is just not hard to imagine with this kinds of models, I think they have to get out with more expensive parallel working agents and higher-than-pro subscriptions, but it is coming I'm sure.

Does this mean RAM is going to get cheaper?

ClosedAI isn't worth even a million at the end of the day.

Open source models catch up quickly and eventually even large models could be run locally.

Has AI transformed the economy radically? Yes.

Will it continue to transform the economy radically? Yes.

Will that translate to the model-makers somehow capturing the entire value of the transformed economy? No.

There were a few key moments that revealed this. When OpenAI initially declared "there is no moat," I wasn't sure whether to believe them. GPT 3.5 and 4 were so much better than the competition, it felt like them saying that they had no moat was some sort of attempt to avoid regulation or scrutiny. But then, lo and behold, Claude and Gemini caught up; there really was no moat.

But up until then, while it was clear that there was no moat around OpenAI, it was unclear if there was a moat around big tech. Mistral was meh. Even Meta's were meh. We also had no idea how much these models actually cost to run. It wasn't until the "DeepSeek moment," and especially once these open source models actually started being hosted on third party services, that it became clear that this was actually a competitive landscape.

And as has already been demonstrated, because the interface for all of these models is just plain language, the cost of switching models is basically non-existent.

"there is no moat" usually mean "we have no moat" or "we want you to believe we have no moat". There are always moats, like being directly in front of eyes and thumbs (Apple) or having extensive data (Google) along hardware production capabilities, datacenters, and tons of money.

Has AI transformed the economy radically? Yes.

AI made “basically zero” difference in U.S. economic growth last year. https://www.youtube.com/watch?v=zZHN0-ZNe_4&t=399s

It doesn't matter.

Both numbers are fictional. No one really expects any of this to be true.

The people who claim to believe this are simply lying.

Ponzi scheme. Scam. I don't trust OpenAI a tiny bit.

you mean Altman scheme.

as I've said previously, OpenAI will be bailed out by US taxpayers. This isnt just another bubble, its a bubble within a bubble.

Seeing the same setup in 2008 and now. Enjoy your subsidized $200/month codex because its going to go up in the future.

https://news.ycombinator.com/item?id=46439545

Reply on news.ycombinator.com