A bipartisan proposal in the US House would direct $10 billion toward commercializing fusion energy, funding new test facilities, demonstrations and research. Representative Don Beyer estimates that the United States could reach fusion at scale within two to 10 years—his forecast, tied to the ability to model plasma with artificial intelligence.

Psst!Add AlexTech.ai asPreferred Source on Google

The bill, the American Leadership in Fusion Act, was introduced on September 24 by Representatives Zoe Lofgren and Jay Obernolte, with Don Beyer, Lori Trahan and James Gallagher as co-sponsors. Lofgren is the top Democrat on the House Science, Space and Technology Committee; Obernolte chairs its Research and Technology Subcommittee. The legislation identifies fusion as a strategic national priority and aims to strengthen the US position relative to China.

How the $10 billion would be allocated

The plan assigns $3.8 billion to major new test facilities addressing challenges involving fusion materials and the fuel cycle. Another $3 billion would support a new demonstration program, with funding tied to technical milestones; $2 billion would expand the existing milestone-based development program. The remaining funds would go to smaller testing facilities and research and development.

That approach ties a substantial share of the funding to measurable project progress rather than distributing it solely according to a calendar. Beyer connected the structure to a study by the bipartisan Special Competitive Studies Project. The lawmaker said $10 billion could resolve the remaining engineering problems through funding awarded in stages.

The proposal would also establish the Energy Department's Office of Fusion in law and authorize the National Science Foundation to create a workforce training program with the department. The measure draws on recommendations from federal bodies and industry, including the Fusion Energy Sciences Advisory Committee and the Fusion Industry Association.

Plasma is the engineering challenge; AI could help model it

Fusion joins atomic nuclei and can produce energy without the carbon emissions associated with combustion. To trigger the reaction on Earth, operators must generate and sustain plasma, a gas of charged particles in which fusion takes place. Controlling its behavior remains a central engineering challenge. In earlier work, a Chinese startup used AI to simulate reactor behavior, illustrating how computational models can contribute to reactor design.

Beyer told Nextgov/FCW that modeling plasma with AI—which, in his view, was not previously possible—is one reason he puts fusion at industrial scale within two to 10 years. That is the lawmaker's forecast, not a deadline set by the bill.

The timelines for experimental projects show how gradual the path remains: in Japan, a startup plans to begin power-on tests at a pilot reactor in 2027. If the first stage succeeds, tests with the completed facility are expected around 2030. The US proposal must now continue through Congress.