The Commerce Department on Thursday, October 22, signed letters of intent for up to $389 million in CHIPS Act research awards, and two-thirds of it is going to fusion companies, according to NIST’s announcement.
Helion gets up to $175 million and Commonwealth Fusion Systems (CFS) up to $85 million. The other two recipients are Vertical Semiconductor, up to $73 million for vertical gallium nitride power chips for data centers, and Sibelco, up to $56 million for ultra-high-purity quartz.
That’s a strange-looking list for a law most people associate with smartphone and AI processors. It makes more sense once you look at what fusion machines actually need, and at what the government wants in return.
Who gets what
According to NIST, the four letters of intent break down like this:
- Helion, up to $175 million: high-voltage power semiconductors and power modules built for “high-peak-current pulsed-power applications in fusion power generation.” In plain terms, chips that can switch enormous currents in very short bursts. NIST says they will “power, control, and harvest energy from the magnets” in Helion’s generators.
- Commonwealth Fusion Systems, up to $85 million: semiconductors for radio-frequency plasma heating, power electronics, embedded electronics for fusion diagnostics, and radiation-hardened electronics, meaning chips designed to survive the neutron bath near a fusion reactor.
- Vertical Semiconductor, up to $73 million: vertical gallium nitride (GaN) power chips. GaN is a material that switches power more efficiently than silicon; “vertical” refers to current flowing through the chip rather than across its surface, which helps at high voltages. NIST says the money includes demonstrating the technology for “the world’s first 2.5 MW super intelligence data center rack.”
- Sibelco, up to $56 million: purification and finishing methods for ultra-high-purity quartz, used in the crucibles that grow silicon crystals and in other chipmaking hardware.
The common thread is power electronics: the unglamorous chips that convert and route electricity. NIST frames them as foundational to the grid and to energy-hungry data centers.
The government wants equity
The awards come with strings. NIST says Commerce will receive “a minority, non-controlling and non-voting equity stake in each company as a condition for making the awards, enhancing the return for the U.S. taxpayer.”
For Helion, TechCrunch reported in an October 23 update, that means a $175 million non-voting interest priced at its Series G valuation. Axios Pro has put Helion’s post-money valuation at $15.5 billion.
Our read: the original CHIPS Act was sold as grants and loans to rebuild domestic chipmaking. Under the current administration, Commerce has repeatedly turned that money into ownership, starting with its 2025 stake in Intel. Taking equity in two privately held fusion startups is a further step: the taxpayer is now a shareholder in companies that haven’t yet sold a single kilowatt-hour.
The catch: letters of intent aren’t final awards. NIST says the companies will do the work “under the final awards,” and terms still have to be negotiated. Helion CEO David Kirtley told TechCrunch that federal contracts can take months.
Why fusion needs chips
Most fusion designs, including CFS’s tokamak (a doughnut-shaped magnetic cage for plasma), would ultimately make heat, boil water and spin a turbine like any other power plant. Helion’s doesn’t.
Helion builds what’s called a field-reversed configuration. As TechCrunch explains, fuel is injected at both ends of an hourglass-shaped chamber, turned into plasma, and fired together with magnetic fields. The collision is compressed further to over 100 million degrees C. When atoms fuse, the plasma pushes outward against the magnetic field, and Helion captures that push directly as electrical current, a bit like regenerative braking in an EV.
That makes the power electronics central to whether the machine produces net power. Today those devices lose about 5% to 10% of the electricity flowing through them, Kirtley told TechCrunch. Because energy cycles in and out of the magnets repeatedly, small losses compound. “If we can cut that loss in half, that’s actually twice the power output of the fusion system,” he said.
The catch: that’s the CEO’s math for his own system, and it assumes the rest of the machine works as designed. Whether it does is still contested (more below).
Where the money goes
Per TechCrunch, more than $100 million of Helion’s award will go to semiconductor and advanced packaging R&D in Washington State, and about $80 million to tooling, design and process development in Vermont with GlobalFoundries. GlobalFoundries separately received more than $80 million in federal money for gallium nitride manufacturing, TechCrunch notes.
Kirtley said the Washington line would make a class of high-voltage, pulsed semiconductor that no US manufacturer currently produces. “There are no U.S. manufacturers right now that do it,” he said.
CFS’s award is spread across several kinds of electronics rather than a single production line. CFS is in a different phase. Its SPARC demonstration tokamak in Massachusetts is in commissioning, and the company has said it plans to begin operations in 2027 and then try to show net fusion energy, according to its SPARC page.
Timing, and a moving goalpost
Helion has a contract to supply Microsoft from its first plant, Orion, in Malaga, Washington. In June, the company said it had received the state radioactive-materials and air-emissions licenses needed for fusion operations there, with earthwork on the generator building begun in spring 2026.
Kirtley acknowledged the award-funded R&D won’t make it into Orion in time. It’s aimed at later plants.
Meanwhile, the Orion timeline has softened. Axios Pro reported in September that Helion’s website now says Orion will “begin initial operations in 2028,” without defining that phrase, and that a Helion executive said on stage, “We will not, in no universe, have 50 megawatts hit full operation in 2028.” Axios also reported that Helion changed its goal for the Polaris prototype from “net electricity” to “demonstrating electricity from fusion.”
Axios cited two academic papers this year that question key assumptions behind Helion’s approach. Helion has published responses arguing its design recovers energy more efficiently than older estimates suggest.
Helion also announced a $50 million private investment from an undisclosed investor alongside the award. TechCrunch says Helion has now raised more than $3.2 billion in total.
Dual use, stated plainly
Commerce isn’t hiding the defense angle. In NIST’s release, Bill Frauenhofer, who leads semiconductor investment at Commerce, said the work will strengthen the ability to “convert and deliver power for the grid, advanced data centers, defense systems, and emerging fusion energy.”
TechCrunch notes that high-power pulsed electronics are relevant to directed-energy weapons, which have drawn renewed interest as drone swarms overwhelm conventional air defenses. Kirtley said Helion remains focused on the grid: “Domestic energy supply, national security through energy dominance, and energy security — that’s our main focus.”
Why it matters: pulsed power electronics with no domestic supplier is exactly the kind of gap that national-security spending tends to fill. Fusion may be the stated use. It won’t be the only customer.
Who this affects
- Fusion companies: a signal that federal industrial policy money is now open to them, at the cost of giving the government a stake.
- Data-center builders: Vertical Semiconductor’s GaN work targets the rising power density of AI racks. A 2.5-megawatt rack would be far beyond today’s typical deployments.
- Chip suppliers: Sibelco’s quartz work addresses a narrow but critical input in growing the silicon crystals that become wafers.
- Taxpayers: who now own minority stakes in four more companies, with no vote.
The takeaway
The takeaway: fusion startups are now big enough to compete for chip-industry subsidies, and the government is buying a piece of them to do it. The chips themselves are useful whether or not fusion works on schedule, which may be the real logic of the deal.
What to watch:
- Whether these letters become signed awards, and on what equity terms.
- Whether Helion’s Washington production line gets built, and who else buys its chips.
- What “initial operations in 2028” at Orion turns out to mean.
- SPARC’s first plasma in 2027, and whether CFS’s net-energy goal holds.
Image: Bri / Wikimedia Commons, CC BY-SA 4.0
