OpenAI Eyes Gigawatt-Scale Fusion Power From Helion as Sam Altman Steps Down Amid Deal Talks

OpenAI is in advanced discussions with Helion Energy on a landmark power agreement that could deliver gigawatt-scale electricity to fuel the explosive growth of artificial intelligence. The proposed deal would secure 5 gigawatts (GW) of capacity by 2030 and scale to 50 GW by 2035—an unprecedented leap that would make AI compute far more sustainable and cost-effective if realized.

The initial 5 GW allocation would represent roughly one-eighth of Helion’s planned production by the start of the next decade, a dramatic increase over the company’s current commercial commitments. Today, Helion’s public agreements include a 50-megawatt (MW) contract with Microsoft and a planned 500 MW facility with Nucor. By comparison, 5 GW equals 5,000 MW—100 times Helion’s existing 50 MW contract—and the full 50 GW target is 1,000 times larger.

What the Scale Looks Like

Each Helion reactor is designed to produce about 50 MW. Hitting the 2030 target would therefore require on the order of 100 reactors; meeting the 2035 ambition would demand around 1,000 reactors. Achieving this buildout will hinge on rapid advances in manufacturing, supply chains, site development, transmission capacity, and regulatory approvals, alongside sustained access to capital for large-scale deployment.

Why Fusion for AI

AI’s energy appetite is soaring as models and data centers expand. Fusion offers a compelling pathway: near-zero carbon emissions at the point of generation, no long-lived radioactive waste, and the potential for high capacity factors compared with intermittent renewables. For OpenAI, locking in a massive tranche of clean, dispatchable power could stabilize long-term energy costs and reduce environmental impact, while enabling continued growth in model training and inference.

Helion’s Technology

Founded in 2013 in Everett, Washington, Helion is developing magnet-based fusion systems that directly convert fusion energy into electricity. Unlike conventional nuclear plants that rely on steam turbines, Helion’s approach aims to turn plasma pulses into electrical power through direct energy capture, promising higher efficiency and fewer moving parts.

The company’s reactors use deuterium and helium-3 as fuel. In February 2026, Helion reported reaching plasma temperatures of 150 million degrees Celsius—an important scientific milestone on the path toward commercial operations. The long-term goal is to achieve and sustain net electricity generation at scale, a milestone that fusion technology has not yet demonstrated in commercial settings.

Governance and Conflict-of-Interest Safeguards

Sam Altman, OpenAI’s CEO and Helion’s largest individual investor, has stepped down from Helion’s board and recused himself from negotiations to mitigate conflicts of interest surrounding the proposed agreement. Altman’s stake in Helion is estimated at approximately $375 million. The move is intended to ensure clean governance as talks progress and to keep decision-making processes clearly delineated between the parties.

Key Hurdles Ahead

Despite the promise of the deal, significant challenges remain:

• Commercial validation: Achieving net electricity generation at a commercial scale is still ahead for fusion.
• Manufacturing scale-up: Building hundreds, and eventually thousands, of reactors will require new factories, talent pipelines, and streamlined supply chains.
• Site selection and infrastructure: Securing suitable locations, grid interconnections, and transmission capacity will be complex and capital-intensive.
• Regulatory and financing: Permitting, safety reviews, and multibillion-dollar financing rounds must align with aggressive deployment timelines.

Why It Could Be Historic

If executed, the agreement would be transformative for both AI and energy. OpenAI could anchor its infrastructure on vast quantities of clean, reliable power, stabilizing costs and reducing emissions as demand for compute accelerates. For Helion, it would mark a watershed commercial moment for fusion—shifting the technology from experimental milestones to large-scale market adoption.

Beyond the parties directly involved, the deal could catalyze new models for pairing energy development with digital infrastructure, from siting reactors near data centers to coordinating grid upgrades and demand response. It would also signal to investors and policymakers that fusion is moving from a distant prospect to a deployable solution within the next decade.

The Bottom Line

The proposed OpenAI–Helion partnership is audacious in scope: 5 GW by 2030, expanding to 50 GW by 2035, potentially requiring up to a thousand fusion reactors. The path is steep and contingent on scientific, industrial, and regulatory breakthroughs. Yet if the plan holds, it could redefine how the world powers AI—and mark the beginning of fusion’s commercial era.