DOE releases fusion roadmap to speed commercial power by mid-2030s

The hard part in fusion is no longer just keeping the plasma hot enough, long enough. DOE’s finalized roadmap shifts the fight to the ugly middle of commercialization: infrastructure, hardware gaps, supply chains and the schedule for pilot plants that could lead to grid power in the mid-2030s.

The June 9 roadmap turns a broad ambition into a more specific federal playbook. DOE says the plan is a national strategy to accelerate fusion development and commercialization, but it also makes plain that it is not a binding funding commitment. Future money still depends on congressional appropriations, which means the community now has a roadmap, not a blank check.

What changed from the earlier draft is not the destination but the amount of structure around it. The final version uses a Build-Innovate-Grow framework. In the Build phase, DOE is putting emphasis on the physical and digital infrastructure needed to speed research, testing and scale-up. In the Innovate phase, it is calling for transformative R&D, cost-competitive power plants and an AI-fusion digital convergence platform, a signal that DOE expects high-end computation and data integration to sit alongside plasma physics and materials science. The Grow phase moves the conversation into public-private partnership programs, supply chains, workforce pathways and links to advanced nuclear deployment.

DOE has also narrowed the technical target list to six core areas that readers can actually track: structural materials, plasma-facing components, confinement systems, fuel cycle, blankets, and plant engineering and integration. That matters because it gives labs, universities and private firms a way to measure whether the roadmap is pushing on the right bottlenecks, or just reciting the usual fusion wish list.

The department has been explicit that it wants to align public investment and private innovation, and it has pointed to the scale of the U.S. fusion base it is trying to organize. A March 2026 DOE presentation said the ecosystem spans 61 universities, 15 national laboratories and 23 private companies, with more than 1,500 full-time equivalents, over 300 graduate students and more than 120 postdocs. The finalized roadmap builds on an October 2025 version, and the broader planning effort had already been taking shape through 2023 and 2024.

That is the real accountability test now. DOE has named the milestones, the core technologies and the timeline to the mid-2030s. What it has not locked down is the funding stream, so the fusion community can judge this document by whether the next budget cycle starts turning those milestones into hardware instead of another promising strategy on paper.