Sam Altman said startups with $10M were 'hopeless' competing with OpenAI

> "Look, the way this works is we're going to tell you it's totally hopeless to compete with us on training foundation models. You shouldn't try, and it's your job to try anyway, and I believe both of those things."

He’s only saying that he’s not incentivized to make a small and scrappy team compete with OpenAI. I don’t think you should read this as “Sam Altman says small companies will never produce valuable AI models”.

The DeepSeek v3 model had a net training cost of >$5m for the final training run, the paper lists over 100 authors[1], meaning highly-paid engineers. This is also one of a sequence of models (v1, v2, math, coder) trained in order to build the institutional knowledge necessary to get to the frontier , and this ends up still far above the $10m mark. It's hardly a "trio of super-smart engineers".

[1] https://arxiv.org/abs/2412.19437v1

It's popular to dunk on Sam, but I don't think he's wrong here. There's now hundreds of companies that have attempted to train a foundational models and almost every single one of them has failed to build a viable business around it (Even Mistral looks to be in rough shape).

Deepseek has done something remarkable, but they have the resources of a multi billion dollar quant fund at their disposal. Telling startups they have a chance is sending them to near certain death. There's way more promising niches to fill.

My personal opinion on society, is that many many businesses have massive inefficiencies and could be wiped off the map if people understood those weaknesses. But there is a culture of "that CEO is so smart, no chance you could compete". Reality is, they are just hiring random people with fancy degrees. I bet most OpenAI "ai engineers" have no clue how low level GPU CUDA programming even works. They are just tweaking pytorch configs, blowing billions on training.

In the past, tech got away with the above because capital meant if you hired enough people, you ended up with something valuable. But AI levels the playing field, reducing the value of capital and increasing the value of the individual contributor.

The fact that the human brain can still do better on certain types of problems than SOTA LLMs while using less energy than a nice LED lightbulb continues to bolster my belief that ultimately it all comes down to the right algorithm.

That’s not to say data isn’t necessary, but rather that the algorithm is currently the critical bottleneck of further AI progress, which a $10M startup absolutely has a chance of outcompeting big tech companies on. Heck, I wouldn’t even put it past an individual to discover an algorithm that blows existing approaches out of the water. We just have to hope that whoever achieves this first has a strong sense of morality...

Specifically he said

> Look the way this works is: we're gonna tell you 'its totally hopeless to compete with us on building foundational models' and its your job to try anyways

He's clearly being glib

Stratechery said that Altman's greatest crime is to seek regulatory capture. I think it's spot on. Altman portrays himself as a visionary leader, a messiah of the AI age. Yet when the company was so small and that the progress in AI just got started, his strategic move was to suffocate innovation in the name of AI safety. For that, I question his vision, motive, and leadership.

But Deepseek took some huge and hugely expensive models that others had paid $Ms to train (well those who didn't already only the hardware mostly used investor cloud credits, but still) and distilled them, rather than training from scratch?

He's right lol. Deepseek has likely spent hundreds of millions getting to this point where they can train a 6M~ dollar model, if that is even the truth.

In the end, he is a very good dealer. He already got his packet. From now on, it's OpenAI's investors' problem. not him.

If I had what OpenAI has I could imagine how to make it profitable tomorrow, because I could do that is why they HAVE to make it free without an account, to prevent anyone new from meaningfully entering the $0 to $20mth segment, If you look at their business strategy, it's top notch, anchor pricing on the $200, $20 sweet spot, probably costs them on average $5/mth to server the $20/mth customers, Take your $50m a year marketing budget and use it to buy servers, run a highly optimized "good enough" model that is basically just wikipedia in chatbot and you don't need to spend a dime on marketing if you don't want to, amazing top of funnel to the rest of your product line. They know what they are doing I suspect, they just need time to pull a platform together, and they have a long head start on that. Basically all of this reaction is to my mind, effectively presuming that "OpenAI" is a current gen chatbot? one would presume it's going to be a lot wider and deeper than that, and given what DeepSeek is pointing to, one with incredible unit economics over time.

I hope this is obvious but deepseek has more than $10 M. You can't spend all of your cash on a single training run.

Of course he's going to say that. He's incentivized to make everyone (especially investors) believe scaling is the only way so he keeps getting billions in investor money. If no one believes OpenAI can be caught, they win.

Scaling compute can only be taken so far. True advancement in AI comes from optimizing compression of human knowledge. Altman missed the first principle by a mile.

This is absurd for a number of reasons. Probably the biggest being that $10m need not be the only investment. Sam of... Y-Combinator is suggesting that startups do not get more funding after $10m? That $10m is insufficient to demonstrate capability that would then motivate more rounds of funding?

Honestly, I want to see more $10m startups. Stop this mega millions level funding into AI/ML and give out more small sums. Let people explore new ideas and scale. Give $10m to people doing non-LLM based stuff. Hell, give $10 to the Mamba team. Give it to the Ziming Liu (KAN). To Bruno Gavranović. To people doing Bayesian stuff. To a ton others. Let these people attempt to scale. If you really want to disrupt OpenAI the best way is to find something better than LLMs. You aren't going to beat the entrenched players at their game, but we know for certain that there's more efficient means of generating powerful learning systems. There's a whole community of "GPU Poor" people, which is just another name for "can't get their ideas through review because reviewers are too lazy to evaluate methods and just respond with 'not novel, needs more experiments.'" There's a metric ton of "unpublished" (still on arxiv) works that are more compute efficient and these people aren't just hungry, they're hungry and want to fight because they've been treated unfairly (I speak from experience).

I agree. It would be pretty hopeless if OpenAI had to compete with a model trained on less than $10M - can't they just buy DeepSeek and rebrand it?

I am beginning to think that the fate of the world might not rest on giving this man infinity dollars

I'm in the chip design industry. After reading Altman's comments in this article and thread last year I realized that he is not as smart as he thinks he is.

https://www.tomshardware.com/tech-industry/tsmc-execs-allege...

https://news.ycombinator.com/item?id=41668824

Ultimately another company finding a better cheaper way to do AI is good for everyone. We were reading how it will be hard for new companies to catch up but but the existing big companies are now going to have to catch up. We were seeing a lot of stories about fab capacity and new power plant construction just for AI. I've been waiting for the AI bubble to pop just like the dot com bubble popped when investors finally started demanding companies show profitability. AI is certainly useful but at some point companies have to start showing how to make a profit rather than just hype.

There are several weird aspects in this thread. I would like to clarify that I don't know Sam Altman nor don't care about him, and do not hold an opinion regarding whether he is a good or bad individual.

First, the title is inaccurate and misleading. It does not pertain to the hopelessness of competing with OpenAI; rather, it addresses the challenges of competing in the training of foundation models, which is not the same.

Moreover, what he stated is likely true. Training a model with a budget of $10 million will likely yield limited results. I appreciate his anwser.

I'm surprised no one has pointed out this seems like a classic example of innovator's dilemma - in the day's of HD's it took years for new technology to come around and upset existing companies. In this information age, everything happens quicker - so a new company, did something the entrenched leader said couldn't be done/didn't explore because their path was "the right path forward" - it created a blind spot. It isn't an exact usage, but I think the concept applies.

A long while back I'd read an article about Facebook's incentive to commoditize technology

It was posted here on HN (years ago) - I can't remember the specifics, but it touched on the subject of Facebook's ulterior motives in open sourcing a lot of their tech/models. At the time the conversation wasn't even about AI

Posting here to see if anyone remembers this article - been trying to figure out what it was for a long time

Deepseek blowing up like this really showed how tired everyone was with the AI founders acting like they’re saving the world.

this is hilarious.

hubris is a terrible thing. there is always bigger fish.

'sama is just maximising his upside in a completely rational way here. He knows he's the big fish, so he attempts to dissuade the little fish before they can even grow. I don't see a problem in him doing that, almost all VCs / CEOs would do the same.

Executive whose dominance is dependent on a moat (cost of training large models), points out the importance of that moat...

Of _course_ he wants to discourage competitors, but that is PR, not technical commentary.

Overconfidence is it's own category of system vulnerability.

He still right even with deepseek, they are way more than 10m

He's such an arrogant... character! If the future of AI is in his hands, the future is not bright, but pretty dark!

It really depends on how you spend your money and the shoulders of which giants you can and will stand on for free.

Creepy dude to be honest.

Commenters - watch the video - it's very short.

makes sense if you think about it - if OpenAI spent, what, $10B on its AI, then it makes sense to say anything less than $1B wont cut it.

You can barely buy a good car for that.

AI is the new bitcoin. Its a buzzword attractive to people looking for attractive investments but cannot be bothered to really map the practical utility back to practical application.

Let's be real. We have had autocomplete on the cheap for more than a decade before LLMs. How much greater utility do LLMs provide above autocomplete? Is that difference in value enough to justify the market value of something like OpenAI? I would say no. That is the definition of a bubble.

I suspect some of the misconception is that people envision LLMs will do things like replace developers. That is so misleading. Yes, AI can replace some developers that cannot do much on their on their own, but that draws entirely incorrect conclusions. Correlation does not imply causation. Just because your developers suck doesn't mean AI is magic. It just means your business couldn't select the right people, or train them, to write software. It also doesn't mean AI will provide something new or better than those people it replaces, which was likely just as true for a more intelligent autocomplete system.

Well, he was wrong about that.

its funny how this is great news but somehow the headlines are still negative

So much for Open-AI.

First they ignore you, then they laugh at you, then all of a sudden when your rival starts cozying up to the incoming Trump administration — boom.

I guess we delved too deep. Or something.

Sam Altman says a lot of things.

This is a bit absurd.

Why would he tell someone how to compete with him?

His answer is I'd say perhaps meant to be facetious considering the rather rude question posed.