Tech giants are going all-in on artificial intelligence (AI). Microsoft (MSFT -1.90%) recently announced plans to spend $80 billion this year building out AI data centers to support training advanced AI models and deploying those models to the cloud. Meta Platforms is following suit, with up to $65 billion allocated for AI infrastructure in 2025.
Not to be outdone, Amazon has upped its capital spending plan to $100 billion for the year, although not all of this will go toward AI computing capacity. Lastly, Alphabet boosted its capital spending plan to $75 billion for 2025, with much of that related to AI.
On top of all this spending, the Trump Administration’s Stargate project involves another $500 billion in AI investments over the next four years, and the European Union is aiming to mobilize 200 billion euros ($209 billion) for AI investments. Long story short, an enormous amount of capital is rapidly being spent on AI data centers and the necessary equipment to train and run powerful AI models, particularly graphics processing units (GPUs) from Nvidia (NVDA -4.05%).
Raining on Nvidia’s parade
Demand for Nvidia’s data center GPUs will likely remain strong for the time being as companies and governments go wild on AI spending. What happens after that? It depends on whether demand for AI computing capacity rises to meet an exploding amount of supply.
In a recent interview, Microsoft CEO Satya Nadella said something curious that should give Nvidia investors pause. While Microsoft is investing heavily in AI computing capacity, Nadella essentially predicted an oversupply sometime in the next few years. The company plans to lease “a lot of capacity” in 2027 and 2028 rather than build out additional capacity of its own. Here’s the kicker:
“Because I look at the builds, and I’m saying, ‘This is fantastic.’ The only thing that’s going to happen with all the compute build is the prices are going to come down.”
One way a bubble can form is through overinvestment. Companies and governments are building out massive quantities of AI computing capacity independently without any real idea of what demand will look like a few years from now. It won’t take much of a miscalculation for supply to greatly exceed demand once this current build-out is complete.
The closest historical parallel to this AI spending bonanza is not the dot-com bubble but the railroad bubble of the late 1800s. Far too much capital was spent building out railroad capacity, leading to an enormous oversupply that sent railroads into bankruptcy and helped trigger a financial panic. The railroad, like AI, was a revolutionary technology that fundamentally changed the world, but that fact didn’t save investors.
In an oversupply scenario, prices for AI computing will plunge. That’s great for anyone renting AI computing capacity and terrible for anyone owning it. In theory, cheaper access to AI computing could drive up demand. However, there’s a limit to this effect. An oversupply of railroads didn’t lead to people taking more trips or shipping more freight to fully match that supply. Trees don’t grow to the sky, after all.
For Nvidia, AI oversupply is a double whammy. Not only would the pricing of its powerful data center GPUs come under pressure, but demand would also fall off a cliff. If there’s too much AI computing capacity, demand will come mostly from maintaining existing infrastructure rather than expansion. There’s also the complication that a growing number of tech giants are designing their own AI chips. OpenAI is the latest to reportedly jump on the bandwagon as it aims to lower its dependence on Nvidia.
An incredibly risky stock
The trillion-dollar question is whether demand for AI computing capacity can possibly rise enough to absorb the enormous supply coming online over the next few years. The fate of Nvidia stock hangs in the balance.
Microsoft is taking a smart approach to AI investment. The company is still building out plenty of capacity, but it’s taking care not to overdo it. By planning to lease a lot of capacity down the road, Microsoft is taking quite a bit of risk off the table.
If Nadella is right about AI oversupply, it’s hard to see how Nvidia’s revenue and profit can grow at a quick enough pace to justify a multitrillion-dollar valuation. The short term may look bright for the GPU leader as AI investments explode, but the longer-term picture isn’t as clear. If we truly are witnessing a massive AI overinvestment cycle, Nvidia stock will come under some serious pressure sooner or later.
John Mackey, former CEO of Whole Foods Market, an Amazon subsidiary, is a member of The Motley Fool’s board of directors. Randi Zuckerberg, a former director of market development and spokeswoman for Facebook and sister to Meta Platforms CEO Mark Zuckerberg, is a member of The Motley Fool’s board of directors. Suzanne Frey, an executive at Alphabet, is a member of The Motley Fool’s board of directors. Timothy Green has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Alphabet, Amazon, Meta Platforms, Microsoft, and Nvidia. The Motley Fool recommends the following options: long January 2026 $395 calls on Microsoft and short January 2026 $405 calls on Microsoft. The Motley Fool has a disclosure policy.
