Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Artificial intelligence is a team effort. For example: OpenAI needs microchips, and lots of them; Nvidia makes the best chips on the market. So it is intuitive and inevitable that the two tech companies would find common ground. Nonetheless, the enormous $100bn tie-up they announced on Monday looks a lot like financial theatre.
The crux of the deal is simple. Nvidia is going to supply chips to help OpenAI build enormous data centres, in which it will train and then host AI models such as the recently launched GPT-5. With it, though, comes a twist: Nvidia will also buy $100bn of OpenAI’s unlisted stock over time, adding to the small stake it already has. That’s comfortably more than the $72bn OpenAI has raised over its 10-year life, according to Crunchbase.
At first glance, this you-scratch-my-back, I-scratch-yours arrangement sounds a bit like the kind of vendor financing common during the early 2000s telecom frenzy. Companies such as Nortel, Lucent and Motorola bunged money to their customers in order to keep their revenue growing briskly, and were saddled with bad debt when the mania ended.
Yet while the AI era already bears unmistakable signs of excess, there are also some important differences. Nvidia is offering equity, so there’s no obligation for OpenAI to pay Nvidia back. True, Nvidia’s investment could fall in value, or OpenAI’s orders could fail to materialise. But given Nvidia has free cash flow of about $100bn a year and a market capitalisation of $4.5tn, founder Jensen Huang is hardly taking on an existential risk.
The mystery is why all of this is necessary. Nvidia is still pretty much the only game in town when it comes to advanced AI chips in large quantities, even if others —including OpenAI itself — are working on alternatives. It’s not, therefore, like Huang needs to lock in such promises. Monday’s accord added about $180bn to Nvidia’s market capitalisation, but even that is a drop in the bucket for a company so large.
As for OpenAI, it’s true that it needs the money — or at least the visible evidence that money is available. Sam Altman’s company is roughly on track to make $12bn in revenue this year, so there’s no way it could fund hundreds of billions of dollars in capital expenditure without help. That said, if OpenAI really is the leader in its field, raising money in $10bn chunks from the market over time ought not to be problematic.
But if Monday’s deal seems performative, perhaps that’s the point. Both sides have something to gain from the impression that the AI race is stepping up a gear. OpenAI is questing towards achieving AI that can outgun human intelligence; its $500bn valuation is driven by investors’ believe that it has a fighting chance. Plans to roll out yet more mammoth data centres should keep that aspirational flywheel spinning.
As for Nvidia, the belief that AI will be a death match between OpenAI and its peers is nothing but a boon. Alphabet, Meta Platforms and the privately held Anthropic are also hurtling towards superintelligence, driven by the belief that those who under-invest will be history’s losers. They now have good reason to try and snag megadeals with Nvidia too, if they can. Huang has secured his status as AI influencer par excellence.