AI spending hits $205 billion and spooks investors

Google’s revised spending forecast has given Wall Street a reason to question the economics of the AI boom. The company now expects capital spending of $195 billion to $205 billion, up from its previous projection of as much as $190 billion.

That $15 billion increase matters because it suggests Google cannot accurately predict the cost of its infrastructure buildout. The company is also spending more money than it is making, while facing competition from Chinese AI tools and pressure to keep model prices low.

AI infrastructure spending comes under scrutiny

The concern extends beyond Google. Meta, Amazon, and Microsoft are all due to report earnings this week, and investors expect they may also raise their data center spending forecasts.

Higher costs would be particularly uncomfortable in a market where companies are expected to hold prices steady or reduce them. That could leave businesses spending more for the same revenue—or spending more while bringing in less.

Several other developments are adding to the unease:

Recommended reading

Apple briefly crosses the $5 trillion mark

• SpaceX: Its shares are worth almost half as much as they were at their peak, according to the source.
• Oracle: Investors are worried about the debt financing its data center expansion. Oracle has become the public-market stand-in for OpenAI.
• Nvidia: The chipmaker has been involved in deal talks worth a combined three-quarters of a trillion dollars, placing it at the center of AI’s circular financing.

If Nvidia is putting more money into supporting the AI buildout, that could indicate demand is weaker than expected. Nvidia’s guarantee of $250 billion in OpenAI debt, in particular, may reflect financing pressure as much as customer demand.

“A reminder of funding strain in the AI build-out as it is a demand signal.”

Chinese models add pressure to chip demand

A new model from a Chinese startup has also revived investor anxiety. China is theoretically constrained by limited access to GPUs compared with US companies, yet Chinese AI systems remain competitive.

If that pattern continues, Nvidia and other chipmakers may eventually see their extraordinary cash generation weaken. It could also suggest that companies are building more data centers than the market will need.

Even investors who remain optimistic about AI expect some overbuilding. The author says people she has spoken with believe many AI companies will fail during an eventual correction, while the survivors could generate enough returns to offset those losses.

Investors are now watching for signs that the market has peaked. Earnings from other major technology companies could ease the concern, but the combination of rising infrastructure costs, uncertain demand, Chinese competition, and financing strain is already pushing some money elsewhere. The source points to Elon Musk and SpaceX’s public debut as a possible warning signal.