Commonwealth Fusion Systems has never generated a single watt of commercial electricity. But Google just agreed to buy 200 megawatts of it anyway.
The two companies announced a strategic partnership that includes a power purchase agreement tying Google to CFS’s planned ARC fusion reactor in Chesterfield County, Virginia. The deal locks in a buyer for power from a plant that has not been built, using a technology that has never operated commercially, in a grid region where even conventional projects wait years for permission to connect.
That grid region is managed by PJM Interconnection, the operator that coordinates electricity across 13 states and the District of Columbia. PJM runs the largest wholesale electricity market in the United States, and its interconnection queue has become one of the most severe bottlenecks in American energy development. According to PJM’s own queue data, hundreds of generation projects are stacked up waiting for grid access studies, with many applicants waiting three years or longer before receiving approval to connect. CFS’s ARC plant now joins that line.
What the deal actually says
The public details are thin. Google will purchase 200 MW from the ARC plant under a long-term offtake agreement, the standard contract structure used when a buyer commits to electricity from a generator before it starts operating. The 200 MW figure is the only hard number either company has disclosed. No construction start date, projected cost, or target year for first power delivery appeared in the announcement.
That silence is telling. Power purchase agreements for operational wind and solar farms typically include specific commercial operation dates and price terms. When those details are absent, it usually signals a project still in early development, where the timeline depends on technical milestones that have not yet been reached.
The financial structure of the contract is also unknown. Whether Google’s commitment includes milestone-based payments, cancellation provisions, or price escalators has not been disclosed. Without access to the underlying agreement or related regulatory filings, it is impossible to determine whether this functions as firm project finance or something closer to a research sponsorship with an option to buy power later.
Where CFS stands technically
CFS spun out of MIT’s Plasma Science and Fusion Center in 2018 and has raised more than $2 billion in private funding, according to the company. Its approach centers on a compact tokamak reactor design that uses high-temperature superconducting (HTS) magnets to confine plasma at the extreme temperatures needed for fusion reactions.
The company hit a genuine technical milestone in September 2021 when it successfully tested a 20-tesla HTS magnet, demonstrating the magnetic field strength its design requires. That test was a meaningful step, but it addressed only one component of a reactor. CFS has since begun construction of SPARC, a demonstration reactor at its facility in Devens, Massachusetts, designed to prove that its tokamak can achieve a burning plasma, the condition where fusion reactions sustain themselves.
SPARC is not a power plant. It is an experiment meant to validate the physics before CFS attempts to build ARC, the full-scale commercial reactor that Google’s contract covers. The gap between a successful demonstration and a grid-connected power station involves billions of dollars in additional capital, years of engineering, manufacturing scale-up, and regulatory approvals that no fusion company has yet navigated.
It is worth noting that the National Ignition Facility at Lawrence Livermore National Laboratory achieved scientific net energy gain (ignition) in December 2022, proving that fusion reactions can produce more energy than the laser energy used to trigger them. But NIF uses a fundamentally different approach (inertial confinement) and was never designed for commercial power generation. No fusion device of any kind has produced electricity for the grid.
The grid bottleneck
Even if CFS solves every technical challenge, the ARC plant still has to get through PJM’s interconnection process. That process was designed around familiar generation types: natural gas turbines, solar arrays, wind farms, and battery storage. A fusion reactor presents a generation profile that grid operators have never modeled. PJM has not published guidance on how it will evaluate the output, reliability, or safety characteristics of a fusion plant during interconnection studies.
The queue itself has become a graveyard for projects using proven technology. Solar and wind developers routinely abandon their applications after waiting years without completing the study process. A novel facility with no operating analogs could face additional scrutiny or modeling challenges that extend timelines further. Whether having a creditworthy offtaker like Google accelerates a project’s progress through the queue is not established by any public PJM documentation.
Virginia, specifically, has become one of the most contested territories for new generation capacity. Northern Virginia hosts the densest concentration of data centers in the world, and the resulting electricity demand has strained the regional grid. Google, Microsoft, Amazon, and Meta have all been competing aggressively for clean energy supply in PJM’s footprint, signing deals for nuclear, solar, wind, and geothermal power as their consumption has surged.
How this compares to other fusion bets
Google is not the first tech giant to place a wager on fusion. In May 2023, Microsoft signed a power purchase agreement with Helion Energy, committing to buy electricity from a fusion plant that Helion said it would have operational by 2028. That timeline was widely viewed as ambitious, and as of mid-2026, Helion has not announced a completed reactor. The Microsoft-Helion deal set a precedent for corporate fusion PPAs, but it also illustrated how far ahead of the technology these contracts can run.
The Google-CFS agreement follows the same pattern at a larger scale. CFS’s 200 MW commitment exceeds the 50 MW or more that Helion initially discussed with Microsoft, and CFS has disclosed more technical progress (the magnet test, SPARC construction) than Helion had made public at the time of its Microsoft deal. But the core dynamic is identical: a tech company with enormous and growing electricity needs is betting that a fusion startup can deliver something no one has delivered before.
These deals reflect a broader shift in how large energy buyers think about procurement. Traditional clean energy purchasing focused on wind and solar farms that could be built within two to four years on well-understood terms. As grid congestion, permitting delays, and rising demand have made those options harder to secure, some buyers are reaching further out on the risk curve, backing technologies that may not deliver power until the 2030s or later.
What to watch next
The verifiable fact at the center of this story is narrow but real: Google and CFS have signed a binding commercial agreement for 200 MW of fusion power from a specific plant in Virginia. That contract exists. Everything required to turn it into flowing electrons does not yet exist.
The milestones that will determine whether this deal produces electricity or remains a high-profile research partnership are concrete and trackable. First, CFS needs to complete SPARC and demonstrate a burning plasma. Second, the company needs to secure financing for ARC’s construction, likely requiring several billion dollars beyond what it has raised. Third, CFS must file for and advance through PJM’s interconnection process. Fourth, it must build and commission a reactor that operates reliably enough to fulfill a long-term supply contract.
Each of those steps could take years, and failure at any one of them would leave Google’s 200 MW commitment as a paper agreement rather than a power source. For anyone tracking the intersection of Big Tech’s energy appetite and the long-shot promise of fusion, the CFS-Google deal is the clearest marker yet of how far corporate climate strategy has moved beyond proven technology and into territory where ambition and engineering risk are genuinely hard to separate.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
