Decision Tree: Structuring SMR–Data Center Financing
- Jun 1
- 5 min read
Earth Venture Capital's SMR–Data Center Financing Decision Tree routes a project to one of seven financing and ownership structures — Nuclear-as-a-Service, Long-Term PPA, Consortium Offtake, Government Anchor Offtake, Developer–Utility Joint Siting, Behind-the-Meter Deployment, BuildCo/OpCo separation, or Single-Entity Ownership — based on four sequential questions about risk tolerance, offtake concentration, public-sector participation and grid access. The framework exists to solve a structural mismatch: nuclear assets are built for multi-decade operation, while data centers run on far shorter investment cycles.
The mismatch at the root of the tree
Nuclear power projects are capital-intensive, slow to amortise and designed for decades of operation. Data centers, by contrast, operate on shorter cycles tied to rapid campus expansion, server refresh timelines and evolving AI workloads. Left unaddressed, that mismatch stops otherwise viable projects.
SMRs narrow the gap.
As Duangkamon Suttipat of Xplor Ventures observed in our research: "SMR fits perfectly well with data centers given its shorter construction period timelines. CAPEX is lower and it takes a shorter time to build, making it easier for hyperscalers to align energy procurement with rapid expansion."
Narrowing is not closing. Financing structures still have to do the remaining work — and which structure fits depends entirely on how a small number of questions are answered.
Question 1 — Does the data center want to avoid SMR construction and asset-ownership risk?
This is the root of the tree, and for most hyperscale operators the answer is yes. Building data centers is their business; licensing and constructing nuclear reactors is not.
If YES → Nuclear-as-a-Service
The SMR developer retains asset ownership and sells power under a long-term contract. The data center takes no construction or ownership risk and instead secures predictable long-term pricing for 24/7 power. This addresses the core mismatch directly while retaining private-sector flexibility.
From here, the tree branches on who the offtaker actually is.
If NO → proceed to Question 4 (ownership structuring)
Question 2 — Does a single data center serve as the primary anchor offtaker?
If YES → Long-Term PPA
The cleanest structure available. A multi-decade power purchase agreement with a single creditworthy offtaker provides the predictable revenue stream that lenders require. This is the single strongest de-risking tool for an SMR developer — without bankable offtake contracts, developers cannot raise construction financing regardless of technical capability.
If NO → Can multiple data centers jointly commit to pooled offtake?
If YES → Consortium Offtake. Multiple data centers jointly offtake a single SMR's output, reducing individual exposure and improving utilisation. This diversifies revenue risk for the developer and lets data centers commit in smaller, more manageable tranches — particularly relevant in markets where no single operator's load justifies a reactor.
If NO → proceed to Question 3.
Question 3 — Can a government or state-owned utility step in as an anchor offtaker?
If YES → Government Anchor Offtake
During the early phase when a data center campus is still ramping, unused capacity undermines project economics. A state utility absorbing excess supply provides a revenue floor, lowering the cost of capital and accelerating financial close.
This is where public-sector leadership becomes decisive rather than merely helpful. Government leadership emerged in our research as the single most important determinant of whether nuclear projects become investable in Southeast Asia — the constraints being regulatory readiness, institutional capacity and policy certainty rather than technology.
If NO → Is grid access controlled by a state or monopoly utility?
If YES → Developer–Utility Joint Siting. The SMR developer delivers the nuclear asset and power output to data center loads, allowing utilities to support system stability without assuming nuclear construction risk, while the developer retains control over the plant and commercial arrangements. This route matters in markets like Vietnam, where EVN's monopoly restricts direct low-carbon procurement, and Indonesia, where the absence of a fully operational DPPA framework keeps hyperscalers dependent on PLN.
If NO → Behind-the-Meter Deployment. Power is generated and consumed on site, bypassing the grid entirely. Matt Loszak of Aalo Atomics argues that tariff structures should reward firm, high-availability power delivered behind the meter, rather than forcing SMRs to compete on a merchant basis against intermittent renewables — in his view, structuring tariffs to reflect the reliability premium is as important as equipment cost in closing the LCOE gap.
Question 4 — Does the project want to phase risk across investor types?
This branch applies where the data center is willing to take ownership exposure.
If YES → BuildCo/OpCo separation
This mirrors the structure that enabled wind and solar to scale globally. BuildCo develops and constructs the SMR, absorbing licensing and construction risk. OpCo — typically an infrastructure fund — acquires the plant once operational and earns steady, long-term returns from contracted power sales.
The model channels different investor types into the project at the stage matching their risk appetite.
Carunpol Songkiatsri of Xplor Ventures described that progression: "Each stage of the technology development will attract different investor profiles. Early-stage nuclear projects are highly risky, so they usually attract venture capital and strategic government support, such as the U.S. DOE. Once feasibility is proven, large-scale funding from later-stage VCs and corporates becomes essential for expansion as the world moves toward net zero."
A realistic caveat: infrastructure funds in Southeast Asia are not yet fully prepared to underwrite nuclear OpCo assets. As regulatory certainty grows, this model could become the backbone for fleet deployment rather than isolated projects — but that is a forward-looking structure, not a present-day default.
If NO → Single-Entity Ownership
One party carries development, construction and operation. Simplest to govern, hardest to finance — generally viable only where a state entity or an exceptionally well-capitalised operator can absorb the full risk profile.
Reading the tree correctly for SMR data center financing
Three things are worth stating plainly, because the diagram alone can mislead.
First, the destination is not a ranking. Behind-the-Meter Deployment is not "better" than Government Anchor Offtake. Each terminal structure is the correct answer to a specific set of market conditions, and those conditions differ sharply across the region.
Second, the answers change over time. A market where grid access is monopoly-controlled today may liberalise. Vietnam has already expanded Direct Power Purchase Agreements, allowing buyers in industrial parks, high-tech agricultural zones and commercial developments to contract power directly from generators, with prices negotiated bilaterally — a reform directly relevant to SMR-powered data centers, where long-term offtake certainty is essential for financing first-of-a-kind projects. As
Bancha Yathip of 89 Plus Energy put it: "The development of data centers is expected to accelerate once the DPPA is approved."
Third, no structure rescues a project that fails on the fundamentals. The tree assumes regulatory readiness and a credible technology partner. It routes capital; it does not create licensing frameworks. Camille Zivré of Exa Ventures LLC framed the boundary precisely: "The remaining challenge, though, is still to create the right political, contractual and financing frameworks to finance beyond, and reach the first successful commercial NOAK."
Why Earth VC built this
Earth Venture Capital is a deep-tech venture firm investing in the science and engineering behind the climate transition. We have backed next-generation nuclear directly since 2024, with positions in Aalo Atomics (United States) and Blykalla (Sweden).
Early-stage nuclear capital sits at the riskiest node of every branch in this tree. Building the framework was partly self-interest: we needed a clear map of how our companies' projects reach financial close in markets where the financing playbook has not yet been written.
Comments