Summary
AI is becoming an electricity infrastructure story, and Doosan Enerbility sits at an increasingly important intersection of that transition. South Korea expects semiconductor expansion, AI data centers and broader electrification to create an additional 25–30GW of power requirements, while policymakers are considering how nuclear generation should fit into the country’s next long-term energy plan.
For Doosan Enerbility, the investment case is broader than a potential Korean nuclear buildout. The company is already converting rising power demand into commercial orders across gas generation, conventional nuclear equipment and SMRs. In March, it secured an order for seven 380MW gas turbines for a U.S. data-center project, bringing its contracted U.S. gas-turbine total to 12 units. It subsequently won its first North American steam-turbine order for another data-center project.
The underlying business is also beginning to show stronger order and margin visibility. The Enerbility division secured KRW 7.12 trillion of orders in the first half of 2026 and ended June with a KRW 26.35 trillion backlog. Its second-quarter operating margin improved to 4.4% from 3.0% in the previous quarter as higher-margin equipment contributed more heavily to revenue.
Nuclear provides another layer of optionality. In August, Doosan signed a manufacturing contract for key equipment for TerraPower’s first Natrium reactor in Wyoming, while potential large-scale U.S. nuclear investment involving Korean companies could create additional opportunities. Those projects, however, have not yet become contracted Doosan revenue and should remain outside the base case.
For global investors, Doosan Enerbility may therefore represent a second-order AI infrastructure opportunity: not through chips or data centers themselves, but through the generation equipment required to power them.
Key Takeaways
- ✓ AI-driven electricity demand is expanding the investment opportunity from semiconductors into power infrastructure, with Korea forecasting an additional 25–30GW of power requirements as semiconductor capacity, AI data centers and broader electrification expand.
- ✓ Doosan Enerbility already has direct commercial exposure to the theme. It has secured seven 380MW gas turbines and two 370MW-class steam turbines for U.S. data-center-related projects.
- ✓ The core Enerbility business is improving alongside the order cycle. Second-quarter operating margin rose to 4.4% from 3.0% in the previous quarter, while first-half orders reached KRW 7.12 trillion and backlog stood at KRW 26.35 trillion.
- ✓ Nuclear and SMRs provide additional upside, including the TerraPower Natrium manufacturing contract, but potential U.S. large-reactor projects remain scenarios rather than secured Doosan orders.
- ✓ Valuation sets a high bar for execution. At the September 8 close of KRW 89,500, the stock trades at roughly 179x 2026 consensus earnings, even though FnGuide’s KRW 129,000 consensus target still implies approximately 44% upside.
What Happened
Power infrastructure moved to the center of Korea’s AI investment narrative on September 8 after the government outlined the scale of electricity demand that could emerge from semiconductor capacity expansion, new AI data centers and broader electrification. Korea may require an additional 25–30GW of power, roughly equivalent to the output of around 20 nuclear reactors. The government is preparing a new long-term energy roadmap through 2040 and is considering whether additional nuclear generation should be included.
At the same time, Korean nuclear-related equities gained on reports that the U.S. side had proposed using part of Korea’s bilateral investment commitment for the possible construction of up to eight large reactors in the United States. The reported investment scale could be substantial, but the project structure, funding and participation of individual Korean companies have not been finalized. It should therefore be viewed as potential optionality rather than contracted revenue.
For Doosan Enerbility, the more important development is that rising electricity demand is already translating into equipment orders. In March, the company signed a contract to supply seven 380MW gas turbines to a U.S. customer’s data-center project, with deliveries scheduled to begin in May 2029. The contract brought its total U.S. gas-turbine deliveries under contract to 12 units.
Later that month, Doosan secured its first North American steam-turbine order, agreeing to supply two 370MW-class steam turbines and generators for another data-center-related combined-cycle power project. Together, the contracts provide tangible evidence that the AI power theme is moving beyond expectations and into the company’s commercial pipeline.
The domestic gas-power pipeline is also expanding. On September 1, Doosan signed a KRW 665.8 billion contract with Korea Southern Power for the 1GW Hadong combined-cycle power plant, scheduled for completion in 2029.
Nuclear provides a longer-duration growth layer. On August 14, Doosan signed a contract with TerraPower to manufacture reactor guard vessels, reactor support structures and reactor internal structures for the first Natrium project in Kemmerer, Wyoming.
The result is a broader investment story than a single nuclear-policy catalyst. Doosan is building order exposure across gas turbines, combined-cycle generation, conventional nuclear equipment and advanced reactors at the same time that AI is increasing the strategic value of reliable electricity supply.
Why It Matters
The AI investment cycle has so far been dominated by semiconductors, servers and data centers. But as computing capacity expands, electricity availability is becoming one of the most important physical constraints on further AI infrastructure deployment.
That changes the potential opportunity set for companies such as Doosan Enerbility. Gas turbines can provide relatively faster additions to generation capacity, while large nuclear plants offer large-scale baseload power over much longer operating lives. SMRs could eventually add another generation platform if commercialization accelerates. Doosan participates across all three areas.
This diversification is particularly relevant because electricity demand is becoming a global rather than purely Korean constraint. In the United States, data-center expansion is already driving large investment in generation infrastructure, and Doosan has secured gas-turbine orders directly connected to a U.S. data-center project. The company therefore has evidence of actual commercial exposure to the AI power theme rather than only theoretical positioning.
For investors, the next question is whether individual projects can accumulate into a sustained order cycle. Gas-turbine deliveries, long-term service contracts, large nuclear projects and SMR manufacturing could produce different revenue and margin profiles over different time horizons. That creates the possibility of a broader and more durable earnings pipeline, but it also means that headline announcements must be separated carefully from actual contracted backlog.
The reported U.S. nuclear discussions illustrate that distinction. Eight reactors would represent an enormous strategic opportunity for Korea’s nuclear supply chain if implemented, but the project remains under negotiation. Investors should therefore treat it as potential upside rather than incorporate the full reported scale into base-case earnings.
For global investors, the deeper issue is whether Doosan Enerbility is becoming part of the AI infrastructure value chain. If electricity scarcity continues driving orders across gas, nuclear and SMRs, the company could increasingly be valued not simply as a cyclical heavy-industry manufacturer, but as a supplier to a structural global power-capacity expansion.
Korea Alpha sees Doosan Enerbility’s emerging investment case as more important than a single nuclear-policy catalyst.
The strongest part of the thesis is that AI-related power demand is already producing commercial evidence. U.S. data-center projects have generated both gas- and steam-turbine orders, while domestic combined-cycle investment and nuclear equipment projects are adding separate sources of backlog.
That makes Doosan different from companies whose AI exposure depends primarily on expectations. The company does not need to become an AI company in the conventional sense. It needs to become one of the manufacturers repeatedly selected to solve the power-capacity constraints created by AI infrastructure.
The distinction between orders and expectations remains critical. Potential U.S. nuclear investment could materially expand the addressable opportunity, but it is not yet contracted revenue. Korea Alpha would therefore exclude the full value of those projects from the base case until specific projects, suppliers and contract values are confirmed.
The larger challenge is valuation. Doosan’s current share price already embeds substantial expectations for future nuclear, gas and SMR growth. That means additional headline orders alone may not be enough. The company increasingly needs to demonstrate backlog conversion, higher equipment mix and sustained margin expansion.
Korea Alpha therefore views Doosan as a credible second-order AI infrastructure play, but one where execution matters as much as the size of the opportunity. The bull case becomes significantly stronger if power demand produces not only more orders, but a structurally higher earnings base.