You know that feeling when someone is overly inflating a balloon, and you can’t help but wince? It’s a sense of “be careful, that is going to hurt!” combined with bracing for the inevitable BANG. That is how I, and many others, have felt for years now regarding AI. We have been urging the world to brace for when this obvious bubble will burst for so long now, it feels like we are crying wolf. But over the past few months, even the finance bros have begun to wake up, especially with Nvidia’s $100 billion investment in OpenAI. This represents one of the most concerning red flags possible, and it is foreshadowing a catastrophic collapse. Let me explain.
If you need to know why the AI industry is a bubble, read my previous article here. Essentially, AI has been overhyped, and not only can it not deliver on its promises, but the limitations of the technology also mean that it cannot significantly improve beyond its current state. In other words, it isn’t profitable for people to use generative AI or to operate a generative AI company. This means that the hundreds of billions, if not trillions, of dollars that have been poured into this technology won’t generate a return. It is a bubble. And, once investors figure that out, they will try to exit quickly; the value of the entire industry will tank, taking anything associated with it, like indexes, banks, or the chip industry, down with it too.
Okay, so why is Nvidia’s investment such a red flag? Well, because it is circular financing.
Nvidia is the AI chip maker. Their chips are the most powerful and efficient on the market, so they power the majority of the data centres used by AI startups, specifically OpenAI. To give you an idea of how significant this is, Microsoft, the main backer of OpenAI, spent 47% of its 2024 expenditure on Nvidia chips to supply data centres to OpenAI, making Microsoft Nvidia’s largest customer.
Now, OpenAI is wildly unprofitable. It is set to post annual losses in the hundreds of billions of dollars for the foreseeable future (read more here). The company is also closely tied to Nvidia, with their plans to reduce dependency on Nvidia chips being half-hearted at best. So Nvidia is set to make an insane amount of money from OpenAI for years to come. It makes no sense for Nvidia to even consider investing in them.
Then why on Earth is Nvidia planning to invest $100 billion into OpenAI in return for equity in the company? Moreover, why is this $100 billion not actually a cash investment but instead allowing OpenAI to “lease” essential GPU chips for that equivalent value from Nvidia?
As a business move, it makes no sense. It won’t suddenly make OpenAI’s dismal outlook shine. It won’t increase Nvidia’s profits.
And let’s not forget $100 billion is a serious amount of money! That is enough to buy nearly three million Tesla Model 3s (which is only slightly less than how many have ever been built) and over 7.6 times what Microsoft has invested in OpenAI, which has bought them a 49% stake in the organisation! It is a stupidly large sum of money, and yet Nvidia seems to be content spaffing it away.
But they aren’t really — this is circular financing, and tech bros have been doing it since the ‘90s.
The dot-com boom of the ’90s was eerily similar to the AI boom. Internet companies were valued much higher than they could ever achieve, and the vast majority were operating at staggering losses. The only way they could stave off their inevitable collapse was to ensure their valuations kept soaring. So, co-dependent companies began to invest in each other. Company A would invest in Company B at an above-market price, and Company B would use that same investment to invest in Company A at an above-market price. Zero tangible work has been done, and no money has actually been exchanged, yet both companies are now magically worth more. This was such a common tactic during the dot-com bubble that some continued this circle jerk while buying fewer shares each time around in an attempt to perpetually inflate stock prices. This inflated the bubble even further, until eventually the legitimate investors got cold feet, and it all came tumbling down.
Nvidia’s deal with OpenAI looks almost identical to this scheme.
OpenAI’s valuation is based solely on the return on investment for equity investors. That is how they became the most valuable startup ever, despite being lightyears away from any profit and having a comparatively tiny turnover. So, if you want to boost their value, just buy a small fraction of the company for a huge amount. And, wouldn’t you know, that is what Nvidia is trying to do.
Now, OpenAI, via Microsoft, is likely Nvidia’s largest customer, accounting for 19% of their revenue in 2024, and that percentage is set to rise dramatically. Why does that matter? Well, unlike OpenAI, Nvidia’s valuation is based largely on its sales volumes and revenue, which have gone through the roof thanks to the AI boom. So, it makes no sense for OpenAI to reciprocally invest in Nvidia for equity, which would raise suspicion and possibly not even increase Nvidia’s valuation. Instead, using this $100 billion to buy or lease chips from Nvidia will increase their revenue by at least that much, sending Nvidia’s already sky-high valuation even higher.
The only way OpenAI can stave off bankruptcy (which it has already narrowly avoided once) is to keep its value shooting up. That way, more investors are willing to put money into covering the losses. Likewise with Nvidia, they can’t really afford to lose 19% of their revenue, as that would have catastrophic impacts on the company’s value and mess up its supply chain and planned production capacity. True, Nvidia might not go bankrupt like OpenAI, but it would be horrifically hobbled. So, for Nvidia to remain valuable and profitable, it needs OpenAI to stay in business, and for OpenAI to stay in business, its value must continue to soar.
Therefore, if Nvidia gave OpenAI the equivalent of $100 billion for a tiny stake, it would increase their value, enabling them to cover their substantial losses by attracting more investment or borrowing against their stock. What’s more, by giving it to them in GPU leases, it will in turn increase Nvidia’s revenue and valuation.
Yet again, no money has been exchanged, and no tangible work has been done, but the valuations of both companies have increased.
Guess what we call it when a financial market, sector, or industry lead’s valuation goes up for no tangible reason? A bubble.
With the public’s perception of AI quickly souring, AI development crawling to a halt, and dodgy backdoor deals like this beginning to emerge, major financial players are growing seriously worried. They are looking at the enormous sums of money they put behind this unicorn and are terrified they will never get it back. For now, despite the risk, they are all holding their nerve as valuations continue to tick upwards. It is a game of chicken. As soon as a major player has decided the risk is too much and sells off, crashing the value, it will trigger a landslide as everyone else rushes to get out and save what is left of their investment. That is what happened in the ’90s, and it is a matter of when, not if, it will happen with AI.
Thanks for reading! Don’t forget to check out my YouTube channel for more from me, or Subscribe. Oh, and don’t forget to hit the share button below to get the word out!
Sources: Bloomberg, CNBC, Will Lockett, Telegraph, Will Lockett, Motley Fool, Motley Fool, BI, TBIFGC
$100 billion here, $100 billion there, pretty soon you’re talking real (if illusory) money.
At least I predict AI will last longer than cryptocurrencies! Though watch the Orange Monster try to back the cryptocurrency crap with our gold at Fort Knox!!!