21 September 2026

 

In February 2026, three US military transport aircraft carried the components of an unfuelled nuclear reactor from California to Utah. The airlift was intended to demonstrate something central to Valar Atomics’ pitch: a nuclear power plant need not remain an enormous structure assembled permanently in one place.

Four months later, the reactor achieved a controlled nuclear chain reaction. By August, Valar had raised $1 billion in a Series B funding round led by Sequoia Capital, as well as securing a $200 million credit facility.

That is unusually rapid progress in an industry where projects are commonly measured in decades.

Valar’s proposition contains an intriguing paradox. It wants to make reactors small enough to manufacture repeatedly, then place hundreds of them together at enormous industrial sites. The company believes this combination can deliver the repeatability of a factory product alongside some of the economies of a conventional nuclear power station.

Whether it works will depend on much more than reactor physics.

 

What is Valar Atomics?

 

Valar Atomics is a California-based nuclear energy startup founded in 2023 by Isaiah Taylor, an unconventional figure in a deeply conservative industry. Taylor left high school at 16, taught himself to code and built software businesses, including a platform serving vehicle repair shops. He has no formal training in nuclear engineering.

His interest in the field is not entirely new. Taylor has said that his great-grandfather worked as a physicist on the Manhattan Project and that he spent years studying why civilian nuclear power had failed to expand as once expected.

His conclusion was that nuclear was not merely building reactors badly. It was selling the wrong product.

Electricity must generally be consumed when and relatively close to where it is generated. Connecting a new power station to a congested grid can take years. Industrial fuels, by contrast, can be stored, transported and traded internationally.

Taylor began imagining nuclear sites that could produce electricity but also use high-temperature heat to make hydrogen. Combined with captured carbon dioxide, that hydrogen could eventually be converted into synthetic aviation fuel, diesel and other hydrocarbons.

Taylor’s contribution is not a new branch of reactor physics. It is an industrial thesis: that nuclear power can be standardised, vertically integrated and financed like a hard-tech manufacturing company.

 

Who runs Valar Atomics?

 

The technical responsibility rests with a more experienced team. Chief nuclear officer Mark Mitchell has worked on high-temperature gas reactors for decades, including South Africa’s Pebble Bed Modular Reactor programme. He was previously president of USNC-Power, which was developing another nuclear microreactor.

Valar’s leadership also includes president Muhammad “Mo” Shahzad, head of operations Kip Mock and head of projects Max Ukropina.

Taylor supplies the industrial vision and fundraising ability. Mitchell and the wider engineering team must turn it into a reliable commercial reactor.

 

How does the Ward 250 reactor work?

 

Valar’s Ward 250 is a high-temperature gas-cooled research reactor. Instead of the water used to cool most commercial nuclear plants, it circulates helium through a graphite core.

Helium remains chemically stable at high temperatures and does not boil inside the reactor. The heat can generate electricity, but could also serve industrial processes that are difficult to electrify directly.

Ward 250 uses tristructural isotropic, or TRISO, fuel. Each tiny uranium fuel particle is enclosed within layers of carbon and ceramic material designed to retain radioactive fission products. Thousands of these particles are incorporated into larger fuel forms. The resulting fuel can tolerate temperatures far beyond normal reactor operating conditions.

These are useful properties, but they are not uniquely Valar’s. High-temperature gas reactors have been researched for decades, while several other nuclear startups are also using TRISO fuel. Earlier gas-cooled projects have encountered expensive engineering and maintenance problems.

Valar’s value will therefore not come from discovering a new reactor type. It must come from simplifying, manufacturing and operating one economically.

 

What has Ward 250 proved?

 

In June 2026, Ward 250 achieved zero-power criticality at the Utah San Rafael Energy Lab. It became the first US Department of Energy-authorised reactor built and operated outside the national laboratory system.

Criticality means that a reactor can sustain a controlled nuclear chain reaction. It validates essential aspects of the design, including the fuel arrangement, neutron behaviour and control systems.

Valar subsequently increased the reactor’s output and generated around 10 kilowatts of electricity, which it used to power an Nvidia DGX Spark computer built around the company’s Blackwell architecture. It was an effective demonstration that the experimental reactor could produce usable electricity, although several competitors are also pursuing nuclear power for AI data centres.

This was a genuine technical milestone, but Ward 250 is a 250-kilowatt-thermal research reactor. It is not yet the five-megawatt commercial system Valar ultimately intends to develop. Nor has it demonstrated years of reliable operation, competitive electricity prices or factory production.

Valar is effectively making three separate bets. The first is technological: that its reactor will operate safely and reliably. The second is industrial: that standardised units can be manufactured repeatedly at falling cost. The third is commercial: that data centres, industrial plants and fuel producers will buy enough energy to support entire nuclear gigasites.

Ward 250 provides evidence for the first. The other two remain largely untested.

 

How the Trump administration accelerated Valar

 

Valar’s speed cannot be separated from American politics.

In May 2025, President Donald Trump signed four executive orders intended to accelerate reactor testing, reform the Nuclear Regulatory Commission and expand domestic nuclear fuel production. Nuclear power was presented less as climate policy than as an instrument of AI leadership, industrial strength and national security.

The Department of Energy subsequently selected Valar and nine other developers for its Reactor Pilot Program. The programme allowed experimental reactors to proceed under DOE authorisation rather than first completing the conventional NRC process. Trump set a target of bringing at least three advanced reactors to criticality by 4 July 2026. Valar was among the companies that delivered.

Taylor has enthusiastically embraced the administration’s language of American energy dominance and appeared alongside Trump at the White House. Valar’s investors include Anduril founder Palmer Luckey and Palantir chief technology officer Shyam Sankar, prominent figures in America’s defence-oriented hard-tech movement.

Government support did not design or construct Ward 250, but it gave Valar a much faster experimental pathway and signalled to investors that advanced nuclear reactors might have a clearer route towards commercialisation.

Supporters see overdue reform of a system that made nuclear experimentation painfully slow and expensive. Critics worry that political deadlines and pressure on an independent regulator could weaken scrutiny. DOE authorisation of an experimental reactor is not the same as approval for widespread commercial deployment, for which Valar still faces substantial regulatory work.

 

What is a nuclear gigasite?

 

Large nuclear reactors benefit from scale. One control room, security operation, turbine and containment structure can support an enormous electrical output. Dividing that capacity among many small machines may require more equipment per megawatt.

Factory production is supposed to compensate. Building the same reactor design repeatedly under controlled conditions could reduce labour costs, improve quality and allow engineers to incorporate lessons from each unit into the next.

Valar’s gigasite model attempts to recover both advantages. Reactors would remain small enough for repeated manufacturing, but hundreds would share security, staff and industrial infrastructure at one location. Capacity could be added gradually rather than through a single multibillion-dollar project. If one reactor stopped operating, it would represent only a small part of the site’s output.

Valar also plans to control more of the supply chain, from TRISO fuel fabrication and reactor deployment to long-term operation. This could remove bottlenecks and improve quality control. It could equally leave a young company attempting to master several complex industries at once.

 

From AI data centres to synthetic fuels

 

Synthetic fuel may be the most original part of Valar’s business model—and its most speculative.

High-temperature nuclear heat could help produce hydrogen from water. That hydrogen could then be combined with captured carbon dioxide to manufacture fuels compatible with existing aircraft, ships and distribution networks.

The attraction is clear. Unlike electricity, liquid fuels can be stored and sold around the world, giving a gigasite access to a much larger market than its local grid.

But each conversion adds equipment, cost and energy losses. Valar has not yet demonstrated an integrated synthetic-fuel plant or published sufficient commercial data to show that its fuel could compete with petroleum or other low-carbon alternatives.

Its most concrete publicly announced external project is a collaboration with Nvidia to explore a 30-megawatt, nearly waterless AI facility in Utah. The companies have described this as an exploration rather than a completed commercial deployment.

No Valar gigasite is operating, and the company has not publicly disclosed binding orders for fleets of reactors or detailed costs for delivered electricity, industrial heat or synthetic fuel.

A billion-dollar investment represents confidence in the opportunity. It is not customer demand.

 

Who are Valar Atomics’ competitors?

 

Valar is not alone in trying to industrialise nuclear power.

Its closest competitor may be Aalo Atomics, which is developing factory-built, sodium-cooled reactors that can be clustered to supply data centres. As MTN previously explored, Aalo is responding to the growing mismatch between the electricity required by AI infrastructure and the capacity available from local grids.

X-energy provides the closest major technological comparison. Its Xe-100 also uses helium cooling and TRISO fuel, but the company is further along in developing utility and industrial projects with partners including Amazon and Dow.

Radiant and Antares are concentrating more heavily on portable or deployable power for military bases and remote locations. Other developers, including Oklo, Kairos Power and Deployable Energy, are pursuing different reactor architectures and markets.

Valar’s distinction is not any single component. It is the combination of small standardised reactors, very large shared sites, vertical integration and products extending beyond electricity.

Ward 250 has shown that Valar can build a small reactor quickly. Its billion-dollar test is whether building hundreds together can combine the repeatability of a manufactured product with the economics of a large power station.