Michael Burry, the investor who made his name calling the 2008 financial crisis, is back with another contrarian warning. This time he's pointing to something most people haven't been thinking about in the AI boom: electricity. Specifically, he says China's massive power-generation buildout gives it a structural edge in artificial intelligence if the industry keeps relying on chips that consume ever-increasing amounts of energy.

It's Not Just The Size, It's The Acceleration

Over the weekend, Burry posted a series of messages on X highlighting a chart that shows China's installed electricity generation capacity has absolutely exploded past the U.S. and Europe since the early 2000s. The gap isn't just wide, it's widening.

"It is not just the total power advantage," Burry wrote. "It is the slope."

In other words, China isn't just ahead in total capacity. It's pulling away faster every year. And if AI development continues down its current path of requiring more and more power, that's a problem for anyone trying to compete without China's infrastructure resources.

Burry took direct aim at Nvidia Corp (NVDA), whose chips have become the backbone of AI computing. "Power hungry Nvidia chips are not the way forward," he wrote bluntly.

Nvidia's Roadmap Is Really A Power Consumption Roadmap

Burry didn't stop there. In a follow-up post, he argued that Nvidia's dominant position in AI has essentially locked the entire industry into a trajectory of escalating power demands.

"Nvidia's development roadmap is essentially a power consumption roadmap," he stated. Innovation, he says, has increasingly become about "how to power and to cool bigger, hotter silicon" rather than making fundamental efficiency breakthroughs.

The problem, according to Burry, is that efficiency gains haven't kept up with the sheer volume of computing being deployed. Every new generation of chips might be more efficient per calculation, but companies are deploying so many more chips that total electricity consumption keeps climbing anyway.

China Builds At Will, The U.S. Battles Permits

Here's where the structural advantage kicks in. Burry contrasted China's infrastructure approach with what he called a decelerating U.S. transmission buildout.

"U.S. transmission grid development is actually decelerating due to permitting issues, while China is building transmission at will to match power output," he said.

That's a fundamental mismatch. American companies are pouring billions into an AI arms race that requires massive amounts of electricity, but they're doing it in a country where building new power infrastructure takes years of regulatory approval. Meanwhile, China can simply decide to build more capacity and make it happen.

The result? U.S. companies are "structurally positioned to lose" if power-hungry scaling remains the dominant strategy in AI development.

Time To Rethink The Brute-Force Approach

Burry's prescription is clear: the U.S. needs to stop trying to win through sheer scale and start focusing on smarter chip design.

"The U.S. needs to get away from bigger and bigger power-hungry chips and innovate with AI-tuned ASICs like nobody's business," he wrote, referring to application-specific integrated circuits that could be designed for AI workloads more efficiently than general-purpose chips.

But there's a catch. Burry said Nvidia has a "death grip" on AI development in the U.S., citing the company's extensive investments and agreements with major AI companies and startups. Breaking free of that ecosystem to pursue alternative chip architectures would require significant coordination and investment.

Valuation Questions As Market Cap Hits $4.40 Trillion

Burry has also been raising eyebrows by suggesting that Nvidia may be using accounting tactics to artificially boost its valuation. The company currently holds a market cap of $4.40 trillion, making it the most valuable company in the world.

In October, the Jensen Huang-led company became the first to reach the $5 trillion market cap milestone. Year-to-date, Nvidia has gained 30.86%, and over the past five years the stock has risen by 1,293.30%.

The company's third-quarter results certainly justified some of that enthusiasm. Nvidia posted revenue of $57.0 billion, up 62% from a year earlier and beating the Street consensus estimate of $54.88 billion.

Earnings per share came in at $1.30, topping analysts' expectations of $1.25. The results marked Nvidia's 12th consecutive quarter of beating Wall Street estimates on both revenue and earnings, with quarterly revenue reaching a new record.

Whether Burry's warnings about power infrastructure will prove as prescient as his mortgage crisis call remains to be seen. But he's asking a question worth considering: what happens when your AI strategy requires more electricity than your country can actually deliver?