NVIDIA and the Case of the Missing $100B OpenAI Investment
When is $100B not $100B? Apparently when it's an amount touted in the highest profile way possible by two of the biggest companies in the world!
That's the part I don't understand about the push-back against the reports that NVIDIA is no longer planning to invest that amount of money into OpenAI. Again, they touted it in their own release! As did OpenAI! There was an interview on live TV and everything! And now it's not like Jensen Huang is flatly denying the new reporting about the change in scope, if not in heart, but he's clearly trying to gaslight his way out of the situation.
When asked about the shift this week, Huang would only say things like, "We will invest a great deal of money." But he wouldn't give an actual figure or even range beyond saying that it was "probably the largest investment we’ve ever made." That's nice, but again, it's not a denial that it's different from what the companies had originally intended! He's trying to sweep that right under the rug as a "nothing to see here" but again, we all saw it! And heard about it to no end! This is not that!
In fact, when the $100B number was brought up in relation to Huang's "largest investment" point, the answer was: "No, no, nothing like that."
Now, to be fair and clear, that entire deal seemed a bit slippery from the start. That's exactly why my headline about it was, "NVIDIA (Intends to) Invest (Up to) $100B in OpenAI (Over Time)". As I wrote back in September:
First and foremost, it's a letter of intent. What's up with OpenAI announcing those of late? (We'll get to that.) But the real key is in the second sentence: "NVIDIA intends to invest up to $100 billion in OpenAI...".
That's not one qualifier, it's two! We get "intends" and "up to". In other words, they may not invest $100B. Or they may not invest the full $100B. Or both. Or neither. And the sentence doesn't end there! "...as the new NVIDIA systems are deployed." In other words, this investment has stipulations.
My intent in that post was simply to point out that while many of the headlines (and push notifications) suggested this was a done deal, it seemed anything but. And well, here we are!
A few days ago, George Hammond of The Financial Times wrote the following, almost in passing, when reporting on OpenAI's latest fundraising efforts:
Nvidia last year struck a multiyear deal with OpenAI to invest $100bn in $10bn increments as the start-up brought tranches of data centre capacity online. However, that agreement has yet to be sealed.
A $20bn cheque from Nvidia for OpenAI’s funding round could be in addition to the $100bn deal, or could lead the two companies to adjust the terms, the people said.
Both reports note that this would be separate from NVIDIA's previous "up to" $100B commitment. But both also note that the previous deal still isn't actually done yet – and, as the FT notes, this new investment could end up in place of that other commitment. I wouldn't be shocked if that were the case – or at least the first tranche, with the rights to buy more later (as was always going to be the case with the original deal – hence, "up to")...
And now it's confirmed that any investment NVIDIA does here will be instead of that original deal. Which again, Jensen Huang would like us to believe is no big deal. But well, it was a big deal. A really fucking big deal. One of the biggest deals ever announced for anything, in fact!
With that in mind, the deflection is silly. Why not just say something like, "as you know, we had a tentative understanding with OpenAI about an investment, but in talking it through, and given OpenAI's needs at the moment, we both decided it was more prudent to invest a large amount alongside others to ensure OpenAI can get the capital they need right now."
Again, NVIDIA's original commitment was always stated to be over a long time period as OpenAI built out their capacity. Subsequent reporting stated it would be done in $10B increments (likely at the valuation of OpenAI at the time of those investments). It seems reasonable to suggest that OpenAI just needed more money upfront, hence the move to a more traditional fundraise.
Of course, they're probably not saying that because it's also probably not exactly the most honest way to frame the situation. As Berber Jin reported for The Wall Street Journal a few days ago:
NVIDIA's plan to invest up to $100 billion in OpenAI to help it train and run its latest artificial-intelligence models has stalled after some inside the chip giant expressed doubts about the deal, people familiar with the matter said.
"Some" is sort of funny there. As it sounds like "some" sure includes Jensen Huang himself:
Huang has privately emphasized to industry associates in recent months that the original $100 billion agreement was nonbinding and not finalized, people familiar with the matter said. He has also privately criticized what he has described as a lack of discipline in OpenAI’s business approach and expressed concern about the competition it faces from the likes of Google and Anthropic, some of the people said.
Again, NVIDIA and Huang are pushing back against that reporting, but not directly refuting it. Just suggesting that it's not accurate. When asked specifically about the reports that he was unhappy with OpenAI in some way, Huang snapped back, "that's nonsense". Okay, but then why has this deal drastically changed? You can't say that and give no reason for the change.1 It implies you simply don't want to admit there's at least some truth to that being the reason for the change.
Does it have something to do with the deal OpenAI struck with AMD – with OpenAI becoming a shareholder in AMD, rather than the other way around – shortly after the original NVIDIA announcement? Huang certainly seemed annoyed/confused about that deal with his chief rival at the time...
Is it related to OpenAI now clearly getting squeezed from Google from up top and Anthropic from below in terms of both product and certainly the business side?
Is it something to do with the notion that seemingly a big part of that original NVIDIA agreement was about leveraging the largest company in the world to help secure the debt OpenAI needed to build out their own AI infrastructure? Did this potential risk spook NVIDIA and/or their own investor base?
Was it Jensen acting fast on a trip with Altman (accompanying President Trump overseas) to ensure OpenAI didn't get closer to Google and their TPUs? We also know how much he cares about that by suggesting he doesn't care about that! Was it Altman agreeing but pushing to announce it immediately even without any real framework in place?
Something else? Nobody knows. Because neither NVIDIA nor OpenAI is saying beyond saying things as if to imply that anyone asking such questions is crazy. Again, this is going to be NVIDIA's largest investment ever! Well, "probably"! But also "probably" nowhere near the original $100B soft-circled. Because if the intent was still to get to that level eventually, Huang would probably say so, not "no, no, nothing like that."
Instead, it sounds more like this will be around a $20B - $30B investment in this round from NVIDIA. That's still massive, no doubt. But the context matters here! Especially with Amazon now reportedly angling to take $50B of the round! And if SoftBank wants another $30B... we're running out of math in even a $100B round! Maybe OpenAI raises more, or maybe NVIDIA puts in less. They've already clearly cut back from their original intent once.
M.G. Siegler
M.G. Siegler
M.G. Siegler
1 You'll also hear Jensen suggest that this is about OpenAI getting their round together, but again, that's not what was originally announced. And why would NVIDIA be waiting on that to get a far higher valuation than when they originallly announced the deal?! ↩