The 18% problem: can helium-cooled reactors decarbonize heavy industry?

Industrial heating accounts for roughly 18 percent of annual global greenhouse gas emissions, and about half of that comes from processes too hot for renewable electricity to replace. Cement kilns, steel furnaces, chemical crackers, and petrochemical heaters need temperatures above 700°C, often above 1000°C. Solar panels and wind turbines make electrons, not raw heat. Electric arc furnaces can cover some of that ground for steel, and induction heating handles a slice of low-to-mid temperature industrial loads, but the high-temperature wedge remains stubbornly fossil.

That is the engineering problem X-energy wants to solve. The Rockville, Maryland company, founded in 2009, is building a helium-cooled, pebble-bed reactor called the Xe-100. According to X-energy's product spec page, the unit is rated at 80 MWe and 200 MWt, with a helium outlet temperature of 750°C and steam delivery at 565°C. The vendor targets 95 percent reliability, a 60-year plant life, four to twelve reactors co-located on a single site, online refueling, and load-following operation. Compared with typical commercial reactors in the multi-hundred-megawatt range, 80 MW is roughly one-tenth the size. That is the point: the unit is sized to sit on a chemical or industrial site and deliver process heat directly, not just electricity to a distant grid.

The pitch is not just smaller. Helium stays a gas at high temperature without the high pressure a conventional pressurized-water reactor needs (about 155 atmospheres), which is what makes 750°C operation feasible in a package that can be factory-built and shipped. The fuel is uranium, formed into tennis-ball-sized "pebbles" coated in three ceramic layers. The "TRISO" name refers to that three-layer shell (tristructural-isotropic), and each particle is its own containment system, which is the basis for the design's inherent-safety claim.

The Dow chemical facility in Seadrift, Texas, has cleared its first NRC hurdle: an Environmental Assessment with a Finding of No Significant Impact, as X-energy announced. The project targets four Xe-100 units in the early 2030s. Amazon has announced a 320 MW, four-unit cluster in Richland, Washington, also aimed at the 2030s, per the company's announcement. The United Kingdom has signed on for up to 6 GW of Xe-100 capacity in the northeast of England, with first power also targeted in the 2030s. And X-energy's TRISO-X subsidiary has its fuel-fabrication license from the NRC, with the first pebble factory due in 2028.

The demand pull behind all of that is not just industrial heat. Data centers, driven by the build-out of AI training and inference clusters, are the single largest new source of US electricity demand, with FERC and the IEA framing data centers as the swing factor in their growth projections. The IEA's Energy and AI report projects data-center electricity demand roughly doubling again by 2030. Some of that load can ride on the same helium-cooled reactor the cement plant sits next to, which is part of why this design is being pitched to two decarbonization problems at once.

The only commercial-scale helium-cooled reactors running today are the twin HTR-PM units at Shidaowan in China's Shandong province, and MIT Technology Review reports that they "have struggled to operate on a continuous basis." The US Energy Information Administration's cost comparison puts small modular reactor electricity at more than six times the cost of utility-scale solar, before any factory-construction learning curve is credited. And the waste profile is a real tradeoff: the Xe-100 may produce roughly ten times the volume of spent fuel per unit of energy as a conventional reactor, though the individual pebbles are lower-activity and arguably easier to store.

None of those problems are fatal. They are the engineering frontier. A reactor that runs at 750°C, ships in factory sections, and is sized to live next to a cement kiln is a different kind of answer to a different kind of problem than the gigawatt-scale light-water units that defined the last seventy years of civilian nuclear. The TRISO-X pebble factory is due in 2028, Dow's Texas project targets the early 2030s, and Amazon's Richland cluster aims for the same window. Whether the Xe-100 can deliver process heat at industrial scale by then is now a question with a clock attached.