Valar Atomics’ Ward 250 microreactor completed a zero-power fueled criticality demonstration, becoming the second advanced reactor to reach that technical threshold under the Department of Energy’s Reactor Pilot Program. The company has stated its intention to connect the reactor to an AI data center designed for near-zero water consumption, a pairing that would test whether small nuclear units can solve both the power and cooling bottlenecks facing hyperscale computing facilities. With DOE now confirming three separate advanced reactor designs have achieved criticality in rapid succession, the federal timeline for moving from laboratory proof to commercial heat delivery is under intense scrutiny.
Why Ward 250’s criticality changes the clock for nuclear-powered AI
Criticality, in reactor engineering, means a sustained nuclear chain reaction has been established at zero power. It is the first physical proof that a reactor design works as modeled, but it is not the same as generating electricity or delivering usable heat. That distinction matters because the headline promise of feeding an AI data center still depends on steps that no microreactor startup has completed at commercial scale in the United States. The Ward 250 milestone confirms the physics; the engineering and regulatory path to grid connection remains ahead.
What makes the Reactor Pilot Program’s pace notable is the comparison with earlier DOE-hosted microreactor efforts. Safety documentation approval for the MARVEL microreactor at Idaho National Laboratory followed a multi-year review cycle before the reactor could attempt initial criticality. Valar Atomics reached its own criticality as the program’s second entrant, and DOE has since confirmed a third design cleared the same bar. If the interval between zero-power criticality and first commercial heat delivery compresses by 18 months or more relative to earlier INL-hosted projects, it would signal that the Reactor Pilot Program’s streamlined testing framework is producing measurable acceleration. That hypothesis, however, cannot yet be confirmed because no public DOE document specifies a delivery date for Ward 250’s first commercial load.
The practical stakes are straightforward. AI data centers consume enormous amounts of electricity and, in conventional designs, enormous amounts of water for cooling. Operators in water-stressed regions face permit constraints and community opposition. A reactor that can supply both electricity and process heat while the facility itself uses minimal water would remove two barriers at once. Valar Atomics has described exactly that configuration, though the specific cooling technology and its regulatory approval status have not appeared in any primary DOE release reviewed for this article.
DOE records confirm three rapid-fire criticality milestones
The strongest evidence for Ward 250’s achievement comes directly from the Department of Energy. DOE’s Office of Nuclear Energy published a formal announcement confirming that Valar Atomics’ Ward 250 completed a zero-power fueled criticality demonstration as the second advanced reactor to do so under the Reactor Pilot Program. The first was Antares Nuclear’s Mark-0, which DOE recognized in a separate release using identical milestone language to define what criticality means and what it does not.
A subsequent DOE announcement covering the program’s third advanced reactor criticality referenced both the Antares Nuclear and Valar Atomics milestones, providing secondary official corroboration and placing Ward 250 in a broader federal narrative. The Reactor Pilot Program’s stated goal is to reach criticality for multiple advanced reactor concepts by a specified deadline, according to the program’s own overview page. Three designs clearing that bar in close sequence suggests DOE is meeting or exceeding the pace it set for itself.
The safety authorization process that preceded these tests followed a pattern visible in the MARVEL microreactor timeline at Idaho National Laboratory. DOE approved safety documentation before MARVEL could attempt initial criticality, a step that involves reviewing the reactor’s safety analysis report, environmental assessments, and operational procedures. Valar Atomics would have gone through an analogous review, though the specific documents and their approval dates have not been published separately from the criticality announcement itself. The Office of Nuclear Energy’s broader stream of press releases underscores how tightly DOE curates public information about these early-stage reactor tests.
Open gaps between criticality and a working data center
Several questions sit between the confirmed milestone and the headline’s full promise. No primary DOE or INL document directly links Ward 250’s criticality to a specific data-center offtake agreement. The claim that the reactor will feed an AI facility built for near-zero water consumption originates from company statements and secondary reporting, not from the official criticality releases. Those releases contain no operational power output figures, no heat-delivery data, and no timeline for when Ward 250 might transition from zero-power testing to sustained operation.
The “almost waterless” cooling design is similarly unanchored in primary federal records. Conventional nuclear plants use large volumes of water for cooling, and eliminating or drastically reducing that requirement would involve air-cooled or gas-cooled heat rejection systems that carry their own engineering tradeoffs, including lower thermal efficiency and larger physical footprints. Whether Valar Atomics has secured regulatory approval for such a system, or whether the data center’s cooling architecture has been reviewed independently, is not addressed in any DOE publication examined here.
There is also a scale question. Microreactors like Ward 250 are designed for modular deployment and relatively modest thermal output compared with full-scale commercial nuclear plants. AI data centers, by contrast, are trending toward gigawatt-scale power demand as model sizes and utilization rates increase. Even if Ward 250 operates as intended, it is unlikely to supply the entire load of a large facility. Instead, it would function as a dedicated baseload source that reduces dependence on the grid and potentially anchors a hybrid system combining nuclear, renewables, and grid power.
Translating that concept into a working project requires more than physics. Any reactor delivering power or heat to a commercial data center must navigate licensing, local siting approvals, and integration with the facility’s electrical and thermal systems. Cybersecurity and physical security requirements increase when a nuclear asset is colocated with a high-value computing target. None of those topics appear in DOE’s criticality-focused announcements, which deliberately confine themselves to the technical milestone.
Regulatory and policy implications
The rapid sequence of three advanced reactor criticality events raises policy questions beyond Ward 250 itself. If the Reactor Pilot Program is successfully shortening the time from design to initial criticality, regulators will face pressure to show comparable efficiency in authorizing higher-power tests and commercial demonstrations. At the same time, community groups and nonproliferation advocates may argue that compressed timelines risk eroding public oversight and environmental review.
For AI developers and data center operators, the key regulatory unknown is how federal and state authorities will treat nuclear-powered computing campuses. Existing frameworks were written around utility-scale reactors feeding regional grids, not dedicated units serving a single industrial customer. Whether a microreactor paired with a data center is classified as an industrial facility, a power plant, or some hybrid will shape everything from insurance requirements to emergency planning zones.
Policy debates will also track equity concerns. Siting nuclear-powered AI infrastructure in water-stressed or economically disadvantaged regions could bring jobs and tax revenue, but it could also concentrate risk and heighten mistrust if communities feel decisions are being made without transparent consultation. The absence of detailed public documentation about Ward 250’s eventual deployment location and cooling system leaves those questions unresolved for now.
What to watch next
The Ward 250 criticality event is best understood as a starting gun, not a finish line. In the near term, observers will be looking for evidence that the reactor can safely ramp above zero power, deliver stable heat output, and interface with non-nuclear systems such as power conversion equipment and experimental cooling loops. Any DOE or company disclosures about higher-power testing will be crucial in assessing whether the technology is on track for commercial relevance.
On the data center side, concrete information about site selection, cooling architecture, and offtake contracts will determine whether the promised nuclear-powered AI campus moves beyond concept art. Clearer statements from regulators and local authorities about licensing pathways will indicate how replicable such a configuration might be in other regions.
Until those pieces come into focus, Ward 250 stands as a significant but bounded achievement: proof that another advanced reactor design behaves as intended at zero power, and a signal that federal programs can move multiple concepts through that gate in quick succession. Whether that momentum ultimately reshapes how AI infrastructure is powered and cooled will depend on engineering, regulation, and community consent that have yet to be fully tested.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
