Pearls From Vitaliy
An excerpt:
This brings me to the next future casualty of today’s hysteria, Nvidia, the most valuable company in the world, trading at 30 times 2026 earnings. That is not an insane valuation for a company growing very fast. But in the case of Nvidia, the E in the P/E will likely decline a lot in the future. Today, if you are building a data center, Nvidia is mostly the only game in town. Several things are likely to happen. There will be more competition. We are approaching peak demand. And capital markets won’t keep financing the customers who buy its chips. I could write a very long essay just about Nvidia.
Three of the biggies are going public: SpaceX (which owns xAI), Anthropic, and OpenAI. All of them are losing money. Together they spend hundreds of billions on data centers, the bulk of which went to Nvidia for AI chips. Meta and Google, cash cows in their own right, were subsidizing their AI spending from their core business. When that was not enough, they levered mildly and now, for the first time, they are actually issuing stock. Yes, they went from net buyers of their own stock to sellers.
You have two scenarios. In the first, these investments lead to such high returns that these companies end up swimming in cash and fund their future Nvidia chip purchases out of future cash flows. But please think about it. All of these companies collectively are throwing close to a trillion dollars, yes, a trillion, at AI. They feel like they are fighting for their survival, and that is exactly what creates a race to the bottom, where their future profitability gets competed away fiercely. And if these companies don’t spend a trillion dollars on data centers next year, Nvidia’s earnings will be a lot, lot less.
Nvidia and the semiconductor sector are a classic capital-cycle story: demand creates too much supply, and a boom leads to a bust. I have written about it many times before, from the railroads in Great Britain in the 1800s to telecom in 1999. I don’t know if it will be next week or next year, but this will predictably end in tears. Nvidia is unlikely to remain the most valuable company in the world, unless all the others become a lot less valuable. The signs of a bubble are everywhere. I am experiencing 1999 déjà vu.
Over the next few weeks the capital markets will be hit by three IPOs and two secondary offerings from Google and Meta, totaling multiple hundreds of billions of dollars. This alone will likely suck liquidity out of the capital markets and put a nice bow on the AI rally that drove the S&P 500 higher this year.
Position: None
