Sam Altman sure has been busy this year. His OpenAI signed approximately $1 trillion worth of deals this year, most notably $300B with Oracle, $22.4B with CoreWeave, and $10B with Broadcom. These commitments will secure the computing power to enhance the company's ability to run services such as ChatGPT and any future innovations.
While the goal of these deals is clear, how OpenAI will fulfil its promise remains a mystery. The company is currently valued at approximately $500 billion, having generated $4.3 billion in sales during the first half of the year, while also incurring $2.5 billion in cash burn during the same period.
All of this poses a question - where is all that money coming from, Sam?
Let's have a look at the most recent endeavours of the AI's leading players.
OpenAI signed a $300B deal with Oracle. As a result, Oracle's stock soared as much as 36% the same day, adding $244B in value. Now, Oracle runs on chips manufactured by Nvidia, which has also just invested $100B in OpenAI. Nvidia has a stake in CoreWeave- a company providing AI infrastructure to… you've guessed it, OpenAI. If all this feels oddly circular to you, it's because it is.
All these recently struck deals share a common thread, and as we're about to see, this entire ecosystem runs even deeper. OpenAI has also committed to purchasing a 10% stake in AMD, a deal that greatly increased the valuation of the chip giant. What could OpenAI do with all this increase in their investment? Purchase more AMD chips.
This growingly complex web of vendor financing relationships echoes the market activity of the "dot-com bubble," which, as we know, burst in the early 2000s, wiping out more than $5 trillion in market value from its peak. The AI Bubble is here and it's real, so much so that Jeff Bezos has called it a "good bubble," given the potential upside of the technology and the benefits to society as a whole. Good or not, one thing is sure: it will burst sooner or later.
The question remains: is (or will) there be enough money coming in from outside of the diagram? The cash flow required to finance the web of deals will eventually have to come from the customers, not investors. This back-and-forth passing of checks is what companies do in the absence of such cash flows.
Doing a little math, we can see that all deals combined, OpenAI would gain access to more than 20 gigawatts of computing capacity, roughly equivalent to the power from 20 nuclear reactors, over the next decade (Per FT). Each 1 GW of AI computing capacity costs about $50B to deploy (in today's prices), according to estimates by OpenAI executives, which rounds up the total cost to about $1 T—a substantial sum for a company that has yet to record a profitable year.
Asked how OpenAI will fund the deal with Nvidia, the CEO Jensen Huang responded: "They don't have the money yet." He added: "They're going to have to raise that money through, first of all, their revenues, which are growing exponentially, equity or debt." Huang also mentioned that after Nvidia previously invested in OpenAI, his "only regret is that we didn't invest more."
OpenAI "is now too big to fail for the sake of the (generative AI) data centre buildout," wrote Peter Boockvar, chief investment officer of wealth management firm OnePoint BFG Wealth Partners, in a note Tuesday. "For this whole massive experiment to work without causing large losses, OpenAI and its peers now have to generate huge revenues and profits to pay for all the obligations they are signing up for and, at the same time, provide a return to their investors."
While it is unclear how the company, which has never turned a profit and generates only around $13 billion in annual revenue, would finance its commitments, there are also reasons to be optimistic.
In a recent report, UBS has downplayed the sheer size of the circular deals, arguing that the way they are structured is performance-based, rather than speculative, as we had seen in the telecom bubble. On top of that, the companies architecting the deals are much healthier financially than the telecom ones back in the day. The "Big 4" tech firms are expected to generate $203B in free cash flow (after capital expenditures) in 2025, implying substantial investment capabilities without incurring debt.
Finally, the valuations themselves, UBS argues, are "more reasonable and earnings quality is higher." When examining the internet leaders of the late 90s, we see P/E ratios of around 60x, compared to 35x for today's AI big players. Combined with stronger balance sheets and more predictable earnings streams, this implies that the market is not being unrealistic and that the companies are well-equipped to face any potential market downturns.
The AI will undoubtedly change the way the world operates; what remains unclear, however, is how exactly and when this revolution will fully take place.
According to Bain & Company, investors shouldn't expect a return anytime soon. In its global tech report, Bain found that by the end of this decade, AI companies would require $2 trillion in combined annual revenue to fund the data centre buildouts they're projecting. Even taking into account the AI-related potential savings, the gap will still likely be too large—some $800 billion, according to Bain's analysis.
Only time will tell if recently struck deals will prove essential or turn out to be nothing more than just CEO's egos writing checks their technology can't cash.
Sources:
OpenAI generates $4.3 billion in revenue, Reuters
Oracle, OpenAI Sign $300 Billion Cloud Deal, WSJ
openai oracle sign 300 billion computing deal The INformation.com
Oracle stock gains 36% to post best day since 1992 CNBC
OpenAI Nvidia partnership OpenAI
Vebdor Financing Investopedia
Jeff Bezos hails AI boom as ‘good’ kind of bubble FT
Should recent AI financing deals be a cause for concern? UBS
Image: Istock
A writer and analyst with a keen interest in the intersection of science, philosophy, and culture.
Very nice wrap-up of the current AI brouhaha that is filling our ears and eyes.. And I see a number of "interesting" (as an editor, I hate that word) points. The industrial-scale wads of cash changing hands for a technology that depends on a plug and a socket is indeed quite unsettling. It reminds me of a Hägar the Horrible cartoon showing his son Hamlet nailing boards across a gorge without any structure. Hägar is arriving, his arms holding more boards and saying: "Don't ask how it's holding, just keep hammering..."
From what I understand, all those who have some serious cash invested are saying: Keep hammering, it's ok, this is not a bubble. Personally, as a regulard Joe Shmoe, I'd be very careful and keep observing from behind the parapet. Because right now You Tube, X and other "social" media are swamped with scammers pretending to be experts and enocouraging you to become a millionaire by sending them money... sort of like the televangelists... To double our money, you may want to fold it in two and put it back in your wallet (I wrote this in an editorial in August 2007, writing for a business mag)...
The whole idea of "too big to fail" is not a fact, it's wishful thinking. Credit Suisse was considered tbtf, but it vanished -- ok, it was swallowed by the UBS whale. But the revolution, when it comes, will come from out of the blind spot. It always does. And these super-billionaires toying with this like the feudal lords of old, think they are really all-powqerful bodhisattvas of the economic nirvana ... I'm not so sure. I think they are unscrupulous carpetbaggers and flimflamming influencers hyping a technology for their own benefit and ignoring the consequences.