The Week That Clarified the Model
Two 20-year nuclear contracts landed in the space of a week. The first, announced September 30, 2026, paired a major cloud and e-commerce operator with a leading U.S. nuclear utility, underpinning a 190-megawatt expansion of a Maryland plant. The second, announced October 6, had a separate technology company backing a $4.3 billion upgrade program spanning 11 nuclear units across three states, all under a 20-year power purchase agreement. Taken together, the two deals add roughly 1.1 gigawatts of incremental generation capacity to the PJM grid, per Utility Dive.1
These transactions are worth examining not for their novelty but for what they confirm about the financing structure now governing nuclear expansion. The offtake agreement has displaced the rate base and the government loan guarantee as the primary instrument underwriting new capacity. Technology companies, not utilities or regulators, are now setting the terms.
What the Contract Structure Actually Does
A 20-year power purchase agreement does more than secure electricity. One of the deals channels $3 billion of investment into an existing plant and explicitly provides the revenue certainty required to relicense that facility for another 20 years and pursue additional development at the site. The offtake agreement, in other words, functions as project finance: it converts a merchant asset with uncertain cash flows into something closer to a contracted infrastructure bond, with a creditworthy technology company sitting in the position normally occupied by a sovereign or a regulated utility.
New nuclear projects can take decades and tens of billions of dollars to build, while extensions and restarts of aging reactors take far less time and money. The nuclear industry has been grappling with aging reactors for years. The offtake model addresses both problems simultaneously: it unlocks capital for assets that already exist and would otherwise face decommissioning, and it does so without requiring a new reactor license or construction timeline. The result is capacity that can reach the grid in years rather than decades.
Hyperscale data center operators are now acting as anchor offtakers for first-of-a-kind advanced nuclear projects as well, providing the bankable revenue contracts necessary to unlock project finance and shift developers from research and development to commercial execution. By securing a bankable offtake agreement, a company can demonstrate a clear return on investment, making it possible to attract the large-scale debt and project equity needed for construction. The two recent deals sit at the more conservative end of that spectrum (existing plants, not new builds), but the mechanism is the same: contracted revenue from an investment-grade offtaker substitutes for regulatory certainty.
What This Means for Capital Allocation
Every major AI company has now signed at least one nuclear deal in 2026, committing to more than a dozen agreements worth nearly 10 gigawatts of capacity, per Forbes. That figure represents a structural shift in where power infrastructure capital flows, and it has direct consequences for investors with exposure to energy infrastructure, private credit, or climate technology.
The practical implication is a bifurcation in the nuclear investment landscape. Assets backed by a long-term offtake from a creditworthy technology company now carry a fundamentally different risk profile than merchant nuclear plants or early-stage advanced reactor developers without contracted revenue. Investment is shifting away from exclusively funding intermittent renewable projects and toward direct capital allocation for firm power technologies capable of supporting 24/7 AI workloads, evidenced by multi-billion-dollar acquisitions in the nuclear supply chain and direct offtake agreements for next-generation power sources.
For investors evaluating private exposure to energy infrastructure or climate tech, the offtake structure is now the variable that most determines underwriting quality. Capacity without a contracted buyer remains speculative. Capacity with a 20-year commitment from an investment-grade technology company is a different asset class entirely, closer in its cash flow profile to contracted infrastructure than to commodity generation. The distinction matters for how a position is sized, where it sits in a portfolio's liquidity spectrum, and how the underlying risk correlates with the rest of a family office's holdings.
What remains unresolved is the long-term rate environment that will govern these contracts. Fixed-price 20-year agreements that look attractive at current power prices become more complex to model in scenarios where grid buildout materially changes the supply-demand balance in PJM and adjacent markets. That uncertainty does not undermine the offtake structure; it is simply the variable that any rigorous underwriting process should stress-test first.
