Michael Heumann

The Fusion Report continues its series on the challenges commercial fusion energy must overcome, this time examining fusion fuel and the systems that inject it into fusion machines. Despite claims that fusion fuel is essentially free, a hypothetical 1GW deuterium-tritium plant would burn roughly 100 grams of deuterium and 150 grams of tritium per day, with tritium costing roughly $30,000 per gram today, or about $1.6B per year, until breeding is proven. Even with breeding, the lithium-6 consumed annually would run roughly $20M per year, and the initial fill of a breeder blanket with 40 to 100 metric tons of enriched lithium could cost $1.6B to $4B at today’s prices. The article also compares fueling approaches across machine types: inertial fusion machines fire cryogenic D-T pellets at 1 Hz or faster, magnetic fusion machines inject D-T gas that is heated into a continuously fueled plasma, and hybrid systems such as Helion’s Polaris accelerate two plasma clouds into each other.

The Fusion Report examines how US nuclear deterrence work has shaped commercial fusion energy. After performing 1,054 nuclear weapons tests between 1945 and 1992, the United States shifted to validating its stockpile through modeling, simulation, and high energy density experiments under the Stockpile Stewardship Program. That mission funds the facilities inertial fusion knows best: the National Ignition Facility at Lawrence Livermore, the Z machine at Sandia, and the Omega Laser Facility at the University of Rochester. All three are decades old, and a recent Hudson Institute paper outlines plans and milestones for a joint program between the NNSA and commercial inertial confinement fusion companies. The article argues that a well designed partnership could accelerate both commercial fusion and nuclear weapons stewardship as the US competes with China on both fronts.

The first wall is the innermost boundary facing a fusion plasma, and no material yet does everything asked of it. Tungsten leads on heat handling but can embrittle under neutron damage, while steels, vanadium alloys, silicon-carbide composites, and ultra-high-temperature ceramics each trade one strength for another. A newer approach abandons the solid surface entirely in favor of flowing liquid metal. ITER has switched from beryllium to tungsten in its revised baseline, European suppliers have opened series-production lines, and Germany’s DINERWA project is developing next-generation alloys for high-heat-flux testing. Yet no fusion first wall has established a commercial operating record, and that gap, between industrialization and a bankable power plant, is where the real question sits.

The Fusion Report examines what will actually drive US demand for commercial fusion energy when it is expected to start coming online in the mid-2030s. The author argues that counting on AI is a mistake: most optimistic forecasts show electricity demand growth from AI starting to level off in the mid- to late-2030s, and the IEA cites a headwinds case in which data-center energy demand plateaus beyond 2030, reaching about 700 TWh by 2035. Nearer-term answers for decarbonizing electricity already exist, including solar plus battery storage, which the US EIA’s conservative baseline projects will account for at least 14% of total US electricity by 2035, and advanced nuclear fission. The author suggests stronger framing for fusion’s US case: roughly 100 GW of generation capacity is projected to retire by 2035, including 40% of the remaining US coal power plants, and if US EV adoption reached 50%, charging would increase total annual electricity consumption by roughly 13%-15% (500 TWh-600 TWh). The conclusion: fusion’s business strategy needs to start focusing on real opportunities instead of illusory ones.