Why Nuclear Went Broke, and What the Offtake Fixes

To understand why data-center offtake matters, you first have to understand how nuclear went broke in the deregulated power market, because the offtake contract is a precise remedy for a specific disease. Let me diagnose the disease first.

A nuclear plant has an unusual cost structure. Almost all of its lifetime cost is incurred before it generates a single kilowatt-hour: the capital cost of building it. Once it is running, its operating cost, mostly fuel and staff, is comparatively low and very stable. This means the economics live or die on two numbers: how much you spent to build it, and how many years of revenue you get to spread that cost across. Financiers capture this with the levelized cost of energy, which amortizes the upfront capital plus ongoing costs over all the energy the plant will ever produce. For nuclear, the capital term dominates that calculation, and the capital term is exquisitely sensitive to two things: the interest rate on the money you borrowed, and whether you finished on schedule.

Here is why schedule is so lethal. If you have borrowed billions and the project runs three years late, you are paying interest on that mountain of debt for three extra years while earning nothing, and that accumulated interest, capitalized into the project cost, can dwarf the original overrun. This is the mechanism, not incompetence alone, that turned the Vogtle expansion in Georgia and the abandoned V.C. Summer project in South Carolina into cautionary tales. The delays did not just add construction cost; they compounded financing cost. V.C. Summer was cancelled mid-construction with billions already spent and no plant to show for it. That is the nightmare every nuclear financier is trying to avoid.

Now layer on the market these plants were built to sell into. In a deregulated or merchant electricity market, a generator sells power at the fluctuating wholesale price. In the 2010s, cheap natural gas from the shale boom pushed those wholesale prices down and kept them there. A nuclear plant with high fixed costs, selling into a market flooded with cheap gas-fired power, could not clear enough revenue to justify itself. That is why plants like Three Mile Island Unit 1 shut down: not because they were unsafe or worn out, but because merchant prices no longer covered their costs. The plant was a victim of the market it sold into.

Now you can see exactly what the data-center offtake fixes. A power purchase agreement, or PPA, is a long-term contract in which a buyer agrees to purchase the plant's output at an agreed price, often for fifteen or twenty years. When that buyer is a hyperscaler with a pristine credit rating, the contract transforms the project's financial character. Instead of gambling on volatile merchant prices, the plant has a guaranteed revenue stream at a known price from a counterparty who will still be solvent in 2045. That certainty is precisely what lenders need to offer capital at a reasonable interest rate, and since the interest rate is one of the two numbers that dominate nuclear economics, lowering it improves the whole calculation.

Think of the PPA as de-risking the revenue side so that capital will tolerate the cost side. The reactor did not get cheaper to build. What changed is that there is now a buyer willing to underwrite that cost with a long, firm, creditworthy commitment, which is the exact thing the merchant market failed to provide in the last cycle. The hyperscaler is, in effect, renting out its balance sheet and its power appetite to make the project financeable.

There is a deeper point here about who values what. To a merchant power trader, firm nuclear output competing against cheap gas looked like a money-loser. To a data-center operator facing a hard physical constraint on power, that same firm, clean output is worth paying a premium for, because the alternative is not cheaper power, it is no power and a stalled buildout. The two parties value the same electrons completely differently, and that difference in valuation is the entire economic engine of the nuclear-data-center deal. The offtaker pays more than the merchant market would because avoiding the power constraint is worth more to it than the price premium costs. That is the trade, and it is why this cycle has a financial logic the last one lacked.