Helion raises $465M at $15.5B valuation as AI data-center power demand fuels fusion race
The AMW Read
Tripling valuation to $15.5B in 17 months with explicit AI compute-demand narrative meaningfully updates the AI infrastructure player map and signals capital-compression arc; significance is segment-level as fusion-timeline risk ties to downward capex flexibility for hyperscalers.
Helion raises $465M at $15.5B valuation as AI data-center power demand fuels fusion race
Nuclear fusion startup Helion has raised $465 million in a Series G round led by Thrive Capital, nearly tripling its valuation to $15.5 billion from $5.4 billion in January 2025. The round includes participation from Alta Park Capital, Lux Capital, Peak XV Partners, SoftBank Vision Fund 2, and Ford Motor executive chair Bill Ford. Total funding now stands at $1.5 billion. Proceeds will accelerate commercial deployment and expand manufacturing capacity for Helion's first power plant, Orion, under construction in Malaga, Washington. The company previously signed a 2028 power purchase agreement with Microsoft and a 500MW development deal with Nucor.
This funding signals that the AI industry's insatiable demand for clean, baseload electricity is being treated as a tangible addressable market rather than a distant aspiration. Helion — backed by OpenAI co-founders Sam Altman and Greg Brockman — positions fusion as the long-term solution to the coming compute-energy bottleneck that hyperscalers face as data-center fleets expand. The $10.1 billion valuation step-up in 17 months reflects capital markets pricing in both technical milestones — Polaris reaching 150 million degrees Celsius with fusion fuel — and the structural force of AI-driven energy procurement.
The round exemplifies the capital-compression arc specific to AI infrastructure: frontier energy startups are attracting mega-rounds at valuations once reserved for foundation-model labs. Helion's ability to raise $890 million across two rounds in 18 months, with explicit AI compute demand as the narrative engine, updates the player map for AI infrastructure. The key unresolved question remains whether fusion timelines can match hyperscaler buildout schedules — Microsoft's 2028 target is audacious by any fusion standard, and any delay would ripple into the compute-expansion calculus for the largest AI training clusters.
#Helion #NuclearFusion #AIDataCenters #EnergyInfrastructure #HyperscalerCompute #ThriveCapital #CleanEnergy