Tech builders love a good feedback loop, and Nvidia Corp. and OpenAI have created a $100 billion one this week.
Nvidia is investing the sum in OpenAI as part of a build-out of data centers, potentially nabbing 2% of the company through the first $10 billion tranche. The circle continues with OpenAI pledging earlier this month to spend $300 billion on cloud compute from Oracle Corp., which also buys chips from Nvidia. Meanwhile, Nvidia is pouring $5 billion into Intel Corp., while OpenAI is formalizing a new corporate structure with its most important investor, Microsoft Corp.
The high cost of developing AI systems has meant that the fortunes of the world’s biggest tech companies are becoming deeply entangled, bordering on incestuous. For some, that’s a clear signal that we are in bubble territory. But that misses the point. This also shows troubling industrial consolidation almost certain to keep new entrants to the market locked out as policy concerns fall further into the background. Nvidia is now the world’s biggest company at a market valuation of $4.5 trillion, while OpenAI is the world’s biggest private tech firm, and in the act of propping one another up, they are pushing potential long-term costs down the line, from concentrating resources to stifling competition and shunting policy efforts.
Nvidia’s shareholders sent its stocks up 4% on the news of its new partnership, knowing that such deals are helping cement their dominance of the AI boom. It’s easy to assume all this exuberance shows a lack of due care to the prospect of a bubble. But Azeem Azhar, an independent AI analyst, recently pointed to several reasons why the markets should be fine for the next few years at least.
While AI spending is gargantuan, for instance, it isn’t distorting wide swathes of the economy as with previous bubbles. Much of the funding is coming from well-capitalized hyperscalers – firms like Microsoft and Alphabet Inc. — rather than from risky debt. And AI companies’ revenue is real — and growing. OpenAI expects sales to triple to $12.7 billion this year (and ChatGPT could well hit a billion users before the end of the year) while its rival Anthropic has annual revenue of more than $5 billion.
Of course, these numbers are a drop in the ocean when you look at how much AI is expected to cost in the coming years. Morgan Stanley has estimated that total spending on global data centers will hit $2.9 trillion, or roughly the gross domestic product of France, between 2025 and 2028, and a recent report from Bain & Co., which cited a similar ballpark figure for computing power costs, says AI company revenues will fall hundreds of billions of dollars short of covering them.
Then again, others expect an acceleration of genAI revenue. Citigroup Inc. analysts estimate AI model makers will see revenue grow 483% to $780 billion in 2030 from roughly $43 billion. Morgan Stanley says that number will reach $1.1 trillion in 2028. To help cover the rest of those computing costs, another $1.5 trillion according to Morgan Stanley, banks will likely step in with trades in private credit and asset-backed securities. That is the idea, anyway.
If you believe today’s financial system is robust enough to cover those costs as generative AI revenues catch up, then you can take the view, like Azhar, that a bursting bubble is not on the horizon, and that living in an age of unprecedented tech profits and capital gives ample cushion for the galactic ambitions of people like Sam Altman, Mark Zuckerberg and Jensen Huang, who himself expects that global investment in AI infrastructure will sit somewhere between $3 trillion and $4 trillion by 2030.
The darkest consequence is harder for the market to quantify, but one that entrepreneurs and policy makers will sense acutely, as these overlapping deals fortify a handful of tech giants in a position of dominance. The generative AI boom doesn't seem to have fostered a vibrant ecosystem flush with startups or midsize firms driving innovation, as much as it has brought the biggest developments behind a wall of gargantuan financing that almost no one else can match.
And as Nvidia gains greater leverage over OpenAI (perhaps even pushing for a board seat), Altman is binding OpenAI to ever more complex financial commitments, prioritizing massive infrastructure expansion at the expense of the public-interest vision he set out for the company at its inception. OpenAI’s latest pledges on child safety and parental controls for ChatGPT risk getting neglected as pressure mounts from more powerful investors to grow revenue, very likely by introducing ads to ChatGPT.
Altman has admitted we’re already in an AI bubble of sorts, saying (rightly) that such things happen when “smart people get overexcited about a kernel of truth.” But huge investors and deep reserves mean that OpenAI itself looks like it would likely be able to ride out any eventual crash. That may comfort big tech shareholders, but for startups and ordinary users it means even more power and opportunity are now concentrated in the hands of a few giants.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Parmy Olson
