Gas Turbines vs Small Nuclear: Which Actually Powers AI Right Now?

The two answers to the same question

Both gas and nuclear will power AI. But only one can do it this year — and that timing gap is the entire investment case for gas right now.

Every data center racing to come online in 2026 faces the same question: where does the power come from? The mandate sweeping through state legislatures says "build your own," and that leaves two serious answers on the table — a natural gas turbine plant, or a small modular nuclear reactor. They're often discussed as rivals. In practice, they're running on completely different clocks, and understanding that gap is how I'm sizing the trade.

Option A: Gas turbines — the answer for right now

Gas wins on one dimension that beats every other: speed. A gas turbine plant can be up and running in about a year, and a larger one in two to three. That's why, for almost everyone who needs power today, gas is the only realistic answer. You simply can't wait years for a first-of-its-kind plant when you're trying to bring AI capacity online this quarter.

The catch is that the ability to build these at scale is concentrated. GE Vernova is one of only a handful of Western companies that can make large-frame gas turbines, and its factories are capped near 10 gigawatts a year — with the order book already sold out through 2030. Baker Hughes covers the lighter, faster aeroderivative turbines a campus grabs when it can't wait. Behind them sits Howmet, casting the single-crystal blades that let those turbines run hotter than the metal's melting point. That's a short supply chain with real pricing power.

Option B: Small nuclear — the answer for later

Nuclear is coming, and over a long horizon I'm genuinely bullish on it. Small modular reactors promise clean, dense, always-on power that's tailor-made for a 24/7 data center. I've laid out that case in detail in my nuclear power bottleneck thesis, and the wave of hyperscaler nuclear contracts is real — I covered it in the PPA wave here.

But the timeline is the whole problem. Most people don't realize the first small reactors in the US aren't really expected until around 2030. For a data center that needs power now, "2030" might as well be "never." That's why nuclear can't capture this first wave of demand no matter how compelling the technology is.

Head to head

	Gas turbines	Small nuclear (SMR)
Time to power	~1 year (2–3 for large)	~2030 for first US units
Best for	Demand that can't wait	Long-term, always-on base load
Supply bottleneck	Turbines (GEV sold out to 2030)	First-of-a-kind, unproven at scale
Key names	GE Vernova, Baker Hughes, Howmet	GE Vernova, and the nuclear stack
Fuel	Natural gas (EQT, ET, WMB)	Enriched uranium

The one that wins either way

Here's what I keep landing on: the question isn't purely gas or nuclear — it's timing. Gas captures the demand that can't wait, and nuclear captures the demand that can wait for something cleaner. The most interesting position, then, is the company standing on both sides.

GE Vernova is exactly that. It's the gas turbine leader — sold out through 2030 with 10–20% pricing power — and it's also building one of the first small reactors in the Western world. So when nuclear finally kicks in around the end of the decade, GE Vernova is already on that side of the trade. For the near term I'd rather own the speed; for the long term, I want exposure to both clocks. That's the practical way I read this build-out.

This is educational, not financial advice — do your own due diligence.