The Power Problem Nobody Wants to Talk About

Here's the thing about training AI models that can write poetry and diagnose diseases: they consume absolutely staggering amounts of electricity. And Nvidia Corp. (NVDA) CEO Jensen Huang just went on "The Joe Rogan Experience" to explain why that's actually a nuclear energy story.

In his Wednesday podcast appearance, Huang laid out the energy bottleneck facing the AI industry with unusual clarity. AI data centers—which he colorfully described as "gigawatt factories"—need so much concentrated power that plugging them into the existing public grid risks destabilizing the whole system. His solution? Dedicated, off-grid power generators. Specifically, small modular reactors.

Huang wasn't vague about the timeline either. He expects to see "a whole bunch of small nuclear reactors" in the hundreds of megawatts range powering data centers within six to seven years. That's the kind of specificity that makes investors sit up and pay attention.

Why November Was Brutal for SMR Stocks

The timing of Huang's comments is particularly interesting because small nuclear reactor developers just lived through an absolutely punishing November. Companies like NuScale Power Corp. (SMR), Nano Nuclear Energy, Inc. (NNE), and Oklo, Inc. (OKLO) watched their stock prices crater as investors confronted an uncomfortable question: How exactly do these companies—which currently generate minimal revenue—plan to finance and execute massively expensive, multi-year nuclear construction projects?

It's a fair question. Building nuclear reactors isn't cheap, and it isn't quick. The skepticism was rooted in fundamentals, not just market sentiment. But Huang's podcast appearance directly challenges that skepticism by reframing the entire investment thesis.

From Risk Story to Opportunity Story

What Huang did, intentionally or not, was validate demand from the most credible possible source. When the kingmaker of the AI boom says SMRs aren't just nice to have but strategically necessary for AI's continued growth, that changes the conversation. Suddenly the focus shifts from "Can these companies execute?" to "How big is the market opportunity they're chasing?"

There are two elements working in favor of the nuclear sector here. First, Huang's comments link SMRs directly to the exponential growth trajectory of AI, positioning them as essential infrastructure rather than speculative energy plays. Second, his six-to-seven-year timeline gives investors a concrete, credible horizon—the kind of roadmap that encourages patient capital to stick around after short-term traders have already bolted.

The Bottom Line

Huang just handed the nuclear sector something it desperately needed: a compelling demand story backed by someone who actually knows what he's talking about. The AI revolution isn't just possible without SMRs—according to one of its chief architects, it's not sustainable without them.

That doesn't mean NuScale, Nano Nuclear, and Oklo are suddenly risk-free investments. These companies still face enormous execution challenges, regulatory hurdles, and financing questions. But they now have a fundamental tailwind that's hard to argue with: the AI industry needs what they're building, and it needs it on a timeline that's actually investable. Whether they can deliver on that promise will determine if November's selloff becomes a buying opportunity or a warning sign investors should have heeded.