Breakthrough Energy's Sixth Fellows Cohort Bets on Fusion, Efficient Chips and Hydrogen

Breakthrough Energy has selected the incoming cohort for its Fellows program. The 2026 group is the sixth cohort in the program's history, and TechCrunch received an exclusive look at the list ahead of publication on September 22, 2026. TechCrunch

The cohort spans agriculture, nuclear fusion, robotics and data center power efficiency. TechCrunch Food systems and fusion appear in the same early-stage batch.

Fusion includes two technical bets. Borealis Fusion is pursuing fusion power using proton-boron fuel, a mix of hydrogen and boron studied as an alternative to the more common deuterium-tritium fuel. StarWarden is working on heat management technology for future reactors. One effort focuses on fuel-cycle physics. The other focuses on how materials and components survive extreme heat.

Three companies address semiconductors and data centers. Bedrock Semiconductor, K1 Semiconductor and Vari Flux are working on improving efficiency of semiconductors and data centers. K1 Semiconductor is developing a method to process gallium nitride and silicon carbide for power electronics, the two materials widely used for high-voltage switching, RF power and fast-charging designs. Improvements in processing affect conversion losses, thermal budgets and operating costs.

Hydrogen logistics get separate treatment. AmyHyTech and Ammovolt Energy are developing ways to use and transport hydrogen with ammonia. Ammonia is easier to liquefy and move than elemental hydrogen. The tradeoff involves cracking efficiency, purity and NOx control at point of use. Cracking here means converting ammonia back to release hydrogen.

Demand framing came from Eric Toone, managing partner at Breakthrough Energy Ventures. Toone said electrifying everything implies a 3x increase in demand for electricity. He described data centers as a 10% bump against basically flat demand for electricity over the last 30 years. That places AI infrastructure inside a larger electrification load curve, not outside it.

Breakthrough Energy describes itself as a global platform to accelerate clean energy innovation and build the industries of the future. Breakthrough Energy Bill Gates is the founder. The organization states that the world's demand for electricity will double by 2050. That projection is undated background material, and it sits below the September 2026 cohort details in priority.

There is also a leadership change inside the program. Ashley Grosh previously led the Breakthrough Energy Fellows program and departed to become a partner at WovenEarth Ventures. Benjamin Gaddy, senior director, is serving as acting director of the program.

The Fellows effort sits alongside a larger investment vehicle. Breakthrough Energy Ventures raised over $839 million for its third flagship fund, reported in August 2024. Utility Dive Earlier, undated background coverage put the portfolio at 19 startups, including three energy-storage startups and two startups using microbes to cut emissions in agriculture products. Those older figures provide context only and do not update the 2026 Fellows roster.

The broader context here is load growth meeting physical constraints. Grid planners spent three decades optimizing for flat demand. Electrified transport, heating and industrial processes change that baseline. Data centers add a concentrated, always-on layer with tight power quality and backup requirements.

In my view, the cohort selection works as a hedge across that stack. Proton-boron fusion, if it works at net energy, avoids neutron activation and tritium breeding issues that complicate deuterium-tritium designs. Advanced heat management is needed whether or not that fuel cycle succeeds. More efficient power electronics help under any demand scenario. Ammonia as a hydrogen carrier addresses storage and shipping without waiting for new pipelines.

Worth flagging for technology teams is the operational implication. Efficiency gains in power conversion and thermal management compound. A percentage point saved in a converter, multiplied across racks and substations, reduces cooling load and upstream generation need. That is less visible than a new reactor concept, but it deploys faster and fits existing procurement cycles. I have watched my own children adopt new devices with no thought for the grid behind them. Enterprise buyers no longer have that luxury. The next phase of infrastructure growth will be measured in megawatts as closely as in FLOPS.