Korean Giants Ride Amazon's Nuclear Bet Into the SMR Boom

The partnership nobody expected, but everyone needed

Amazon is quietly becoming one of the most significant buyers of nuclear power on the planet. On October 19, the company announced it had signed a joint development agreement with X Energy and American Electric Power to study building up to a gigawatt of small modular reactors on the site of a retired coal plant in Sullivan County, Indiana.

The details matter. Twelve Xe-100 reactors — each 80 megawatts, four modules bundled together — could deliver 960 megawatts of firm, carbon-free electricity. That is enough to power a substantial slice of Amazon’s growing cloud and logistics empire, and it places the company squarely in the vanguard of a shift that will define the next decade of energy policy: hyperscalers going direct to nuclear.

Who wins when data centers knock on the reactor door

The immediate winners are clear. DL E&P and Doosan Enerbility, two South Korean industrial giants, are among X Energy’s earliest investors and the default suppliers for a project of this scale. They put up $25 million collectively in January 2023 — $20 million from DL E&P and $5 million from Doosan — and when X Energy went public in April, their equity stakes roughly quintupled in value. DL E&P now holds approximately 170 billion won in paper gains; Doosan’s portion sits around 40 billion won.

That is not the full picture, though. Both companies have since moved beyond passive investment. DL E&P signed a contract in March to help standardize X Energy’s reactor design — a move that positions it to lead the engineering, procurement and construction phase rather than simply act as a subcontractor. Doosan secured a binding contract the same month to manufacture the primary nuclear equipment, a role that typically commands the largest share of a reactor project’s budget.

The structure behind DL E&P’s play is elegant: its affiliate DL Energy handles project development and financing, DL E&P executes the construction, and DL Energy takes over operations once the plant is running. It is a vertically integrated model that captures margin at every stage.

The site itself tells the story

Sullivan County is not a random choice. The American Electric Power plant that once stood there was a coal facility — exactly the kind of stranded asset that a transition project like this is designed to resurrect. The closed plant already has grid interconnection in place. The surrounding workforce knows what it means to run a power station. The site carries none of the NIMBY resistance that greets new reactor construction in greenfield locations.

This is the template. A retired fossil fuel plant, an existing grid connection, a hyperscaler buying the output under a long-term power purchase agreement, and a Korean contractor doing the heavy lifting. It can be replicated dozens of times across the American Midwest and beyond.

The bigger bet: 60 reactors by 2039

The Indiana project is a feasibility study — a JDA, not a shovel-ready commitment. But it sits inside a much larger framework. Amazon and X Energy have a strategic agreement dating back to 2024 that targets the construction of at least 60 Xe-100 reactors by 2039, capable of supplying up to 5 gigawatts of power. They are also advancing a 12-reactor project near Richland, Washington, with groundbreaking targeted before 2030.

Fifty gigawatts is a staggering number if taken seriously. For perspective, the entire nuclear fleet in South Korea — the country’s backbone of low-carbon electricity — generates roughly 30 gigawatts. Amazon alone, under its current roadmap, would exceed Korea’s total nuclear output within two decades.

What this means for Korea’s industrial strategy

Korea has spent decades mastering large-scale nuclear construction, most notably through the OPR1000 and APR1400 reactor designs exported to the UAE and planned for Poland. Small modular reactors represent a different competitive landscape — one where design standardization, factory fabrication, and speed of deployment matter more than the muscle-memory of pouring massive concrete containments.

DL E&P and Doosan are betting that their existing relationships and supply-chain depth give them an edge in this new order. The evidence so far is encouraging. They hold binding contracts with X Energy. Their investment thesis has already paid off fivefold. The Indiana project, if it clears feasibility, will be the first Korean-engineered SMR deployment on American soil.

But the competition will intensify. Chinese and European SMR developers are moving fast, and American firms like NuScale and TerraPower are pursuing their own hyperscaler partnerships. The winners in this market will be the ones who can deliver reactors on time and on budget — something the nuclear industry has historically struggled to do at any scale.

The timeline that matters

Feasibility studies for the Indiana project will take some time. A final investment decision is unlikely before 2026 at the earliest. Construction, if it proceeds, would not begin until the late 2020s. First power is still years away.

For Korean industrial investors, the patience required is the point. SMR projects are not quick wins. They are decade-spanning commitments that reward the players willing to invest capital and reputation early — before the headlines arrive. DL E&P and Doosan made that bet in 2023. Indiana is the first public test of whether it was the right one.

The broader implication extends well beyond South Korea. If Amazon’s nuclear strategy works — and the Indiana JDA suggests it might be on track — the model will attract copycats. Microsoft, Google, and Oracle are all exploring similar arrangements. The data center boom is consuming electricity at a pace that solar and wind alone cannot satisfy without massive storage buildout. Nuclear is no longer a fringe option for tech companies. It is becoming a boardroom strategy.

And the contractors who built the supply chain for the last generation of large reactors are now positioning themselves for the next. That is how industrial transitions work — not with sudden leaps, but with players who were already in the room when the door opened.