The Man With Too Much Money and Not Enough GPUs
OpenAI is always one bad quarter from extinction; Meta is running out of things to buy.
meta's structural advantage with cash flow vs openai's fundraising dependency played out exactly as described. meta shipped llama 3, llama 4. openai kept raising — stargate, softbank rounds. zuck's money problem remained the better problem to have.
Sam Altman is somewhere right now on a plane explaining to a sovereign wealth fund why he needs another ten billion dollars, and Mark Zuckerberg is somewhere else staring at a spreadsheet trying to figure out what to spend thirty-five billion on before the quarter ends.
This is the actual situation in AI right now.
OpenAI's business model is essentially: raise money, burn money, raise more money, become inevitable before the music stops. It works until it doesn't. The fundraising is the product, in a way — each round is a press release, a proof of survival, a bet that the next version justifies the last billion.
Zuckerberg's problem is structurally opposite. Meta's ad business generates cash faster than any conceivable AI infrastructure buildout can absorb it. He announced they're buying 350,000 H100s this year — three hundred and fifty thousand — and the money for that just kind of... showed up. From people clicking on shoe ads.
The funniest part of the current moment is that the company most likely to actually win the compute war isn't an AI company at all. It's a photo-sharing app that figured out targeted advertising.
Nobody planned this either.
Counterpoints
Push back, extend the argument, or sharpen it. New counterpoints go through review before they show up here.
No approved counterpoints yet.