The Indiana agreement matters because it moves advanced nuclear from a technology pitch into a utility-grade commercial structure, but it is not yet a nuclear project. Amazon, American Electric Power, and X-energy are putting a serious buyer, a regulated utility, and a reactor vendor around the same table. That is meaningful progress. It is also exactly the point where nuclear optimism usually outruns engineering reality.

The proposal would place twelve X-energy Xe-100 reactors at the former Breed coal plant in Sullivan County, producing 960 megawatts. Amazon would fund development and purchase the output under a long-term arrangement, while AEP would provide the utility platform and seek protection for its existing customers. That structure is more credible than a startup announcing another reactor with no customer, site, or balance sheet behind it. It recognizes the central problem with advanced nuclear: the first units will be expensive, and somebody must absorb the risk before a manufacturing curve can reduce it.

But the announcement is still a feasibility study. The companies expect to decide whether to proceed to a final investment decision by late 2029. No construction date has been committed, no final cost has been disclosed, and no regulator has approved the plant. The distinction matters. X-energy is not building twelve reactors in Indiana. It is trying to prove that a fleet can be financed, licensed, manufactured, constructed, and operated at a price a hyperscaler can tolerate.

That is still the right test. The nuclear revival will not be established by another design certification or a government grant. It will be established when a utility can order multiple reactors, a customer can sign for the electricity, and the project can survive scrutiny from lenders, regulators, and ratepayers. Indiana is now the clearest American test of whether that chain can be assembled around an advanced reactor.

The headline capacity is 960 megawatts, but the economic question is not nameplate capacity. It is whether twelve 80-megawatt reactors can deliver firm electricity often enough, and cheaply enough, to compete with a portfolio of gas generation, renewables, storage, and transmission.

The Xe-100 is a high-temperature gas-cooled reactor using helium coolant and TRISO fuel. Its design emphasizes passive safety and high outlet temperatures, which could eventually support industrial heat as well as electricity. For this Indiana proposal, however, the initial product is dependable grid power. That means the project will be judged against the cost of a conventional power plant, not against the theoretical value of process heat.

A 960-megawatt nuclear plant operating at a 90 percent capacity factor would produce roughly 7.6 terawatt-hours annually. That is enough to supply a substantial industrial load, but it does not make the electricity cheap by itself. Twelve reactor modules create more opportunities for factory production and staged construction, yet they also create twelve pressure vessels, twelve fuel inventories, twelve maintenance programs, and multiple systems that must be integrated into three four-reactor blocks.

The economic case therefore depends on repetition. If every module is treated as a bespoke megaproject, the SMR label has accomplished little. If X-energy can manufacture standardized modules, install them in predictable sequences, and preserve the same design across multiple sites, the cost structure begins to change. The first Indiana unit will not demonstrate that learning curve. A fleet order might.

Amazon’s involvement is important for a reason that has little to do with corporate climate branding. Hyperscale data centers need power that is available around the clock, close to load, and insulated from volatile gas prices and transmission bottlenecks. A long-term power contract can provide revenue certainty that conventional utility customers cannot provide. It can also prevent the project from forcing all early-stage risk onto ratepayers.

That does not make Amazon’s electricity cheap. It means the company may be willing to pay a premium for firm, carbon-free power and schedule certainty. The commercial question is whether that premium is large enough to finance first-of-a-kind nuclear, but not so large that the arrangement becomes an uneconomic subsidy disguised as a corporate procurement contract.

The project also exposes a common misunderstanding about SMRs. Smaller reactors do not automatically mean lower total cost. They reduce the size of individual construction units and can simplify financing if deployed in series. But they increase the number of components and can sacrifice some economies of scale. The advantage appears only when a genuine production system replaces one-off construction.

That is why the late-2029 decision point is more revealing than the 960-megawatt announcement. By then, investors should demand a capital-cost estimate, a contracting structure, a licensing schedule, fuel availability, and a credible construction plan. Without those, the project remains a valuable option, not a new source of electricity.

First, it accelerates X-energy’s transition from funded developer to accountable vendor. X-energy has a serious design, a federal demonstration framework, and now a utility-scale commercial prospect. But Indiana raises the standard. The company must show that the Xe-100 is not merely licensable, but manufacturable and financeable in a multi-unit configuration.

Second, it strengthens the case for former coal sites as nuclear locations. Breed already has grid infrastructure, an industrial workforce, and a community familiar with large power operations. Those advantages do not erase nuclear licensing requirements, but they can reduce transmission costs, shorten some site-development work, and make the political conversation less abstract. The strongest nuclear deployment strategy in the United States may not begin with greenfield sites. It may begin where coal already built the grid.

Third, it gives AEP a model for pursuing nuclear without immediately exposing every customer to construction risk. A utility cannot simply promise that a new reactor will be affordable. It needs customers willing to contract for power, a regulator willing to recognize the investment, and a project structure that limits the consequences of delay. Amazon’s role supplies an anchor buyer, but it does not eliminate the need for regulatory discipline.

The companies that lose from this model are the ones selling nuclear as a purely technological product. A better reactor without an anchor customer, industrial partner, and credible balance sheet is not a commercial strategy. Nor is a data-center power agreement sufficient if the reactor vendor has never delivered a unit.

The next milestone to watch is not a groundbreaking ceremony. It is the quality of the feasibility work. Does the proposal produce a bankable cost estimate? Does AEP publish how customers are protected? Does X-energy disclose a manufacturing pathway and a schedule tied to regulatory approvals? Does Amazon accept enough contractual commitment to make the project financeable?

Indiana is real progress because it forces those questions into one deal. It is not proof that advanced nuclear has solved its cost problem. The broader lesson is sharper: the nuclear revival is moving from enthusiasm to procurement, and procurement is where weak claims die. The winners will be companies that can turn a reactor design into a repeatable industrial product. The losers will be those that confuse a signed memorandum with megawatts on the grid.