The $22 Billion ITER Reactor Won't Run Until 2039. Private Startups Are Racing to Beat It.

The $22 Billion ITER Reactor Won't Run Until 2039. Private Startups Are Racing to Beat It.

The biggest fusion project in human history just got delayed again.

ITER — the €22 billion international fusion reactor in southern France — announced its latest setback: full operations won't begin until 2039. That's a four-year slip from the previous schedule, which was itself a slip from the schedule before that.

The joke in the physics community has always been that fusion is "thirty years away, and always will be." ITER's delays have made that joke feel less funny and more like prophecy.

But here's what's changed: the race no longer belongs to governments alone.

Over the past five years, private fusion startups have attracted $14.24 billion in investment. In just the twelve months leading to July 2026, 56 fusion companies raised $4.48 billion — a record. Twenty-four funding rounds closed in the first half of 2026 alone.

The private sector isn't waiting for ITER. It's trying to eat ITER's lunch.

—

The Startup That Bet It Could Beat Everyone

Helion Energy might be the most aggressive bet in the history of energy.

The Washington-based startup doesn't use a tokamak at all. Instead, it fires pulses of magnetic fields to compress plasma — a fundamentally different approach that's smaller, faster to iterate, and potentially cheaper to build at scale.

In 2026, Helion raised $465 million in new funding at a $15.5 billion valuation. Its timeline? Produce electricity from fusion by 2028. That's not a typo. Two years from now.

Microsoft already signed a power purchase agreement with Helion — the first commercial fusion PPA in history. If Helion delivers, Microsoft gets grid-scale fusion power before ITER even finishes construction.

If Helion misses, the timeline slides. But the bet itself has already reshaped the industry's psychology. It proved that someone could stand up and say "we'll do this in years, not decades" — and still raise billions.

—

Commonwealth Fusion Systems: The MIT Offense

If Helion is the maverick, Commonwealth Fusion Systems (CFS) is the institutional favorite.

Spun out of MIT's Plasma Science and Fusion Center, CFS has raised over $2.4 billion — more than any other private fusion company. Their approach uses a tokamak, like ITER, but dramatically smaller. The key innovation is high-temperature superconducting magnets that are far stronger than anything ITER had access to when it was designed in the 1990s.

In June 2026, CFS announced that peer-reviewed papers had validated their magnet and plasma confinement approach. Their experimental reactor, SPARC, is designed to achieve net energy gain — the same milestone ITER is chasing, but in a machine roughly 1/40th the size.

CFS doesn't claim 2028. Their target is net energy gain by the early 2030s, with a commercial power plant shortly after. That's still years ahead of ITER's 2039 timeline.

—

The Field Behind Them

Helion and CFS get the headlines, but they're not alone.

Zap Energy

uses a "sheared-flow Z-pinch" — a concept that's been around since the 1950s but was considered impractical until modern diagnostics and materials made it viable. No magnetic coils required. If it works, it could be the simplest and cheapest fusion reactor ever built.

TAE Technologies

has raised over $1.2 billion and uses a field-reversed configuration that could theoretically run on hydrogen-boron fuel — the holy grail of fusion because it produces almost no neutron radiation. Their latest machine, Norman, has sustained plasma at temperatures exceeding 50 million degrees.

Tokamak Energy

builds compact spherical tokamaks in the UK. Their private approach to what was once a purely government technology has been validated by repeated plasma shots at their Oxfordshire facility.

Realta Fusion

and

Type One Energy

represent the newest wave — both founded within the last five years, both pursuing alternative magnetic confinement geometries that could reduce reactor size and cost by orders of magnitude.

—

Why ITER's Delay Actually Matters Less Than You Think

Here's the counterintuitive truth about ITER's 2039 slip: it might not matter.

ITER was never designed to be a power plant. It's a science experiment — a proof-of-concept reactor designed to demonstrate that sustained fusion at power-plant scale is physically possible. Its "Q-factor of 10" goal means producing ten times more energy than it consumes to run. That's a physics milestone, not a commercial one.

The private companies aren't trying to prove physics. They're trying to build machines you can plug into the grid and sell electricity from. That's a different problem — and arguably a more solvable one, because the physics of fusion has been understood for decades. What was missing was the engineering: better magnets, faster diagnostics, advanced materials, and the ability to iterate quickly.

Governments build ITER-scale projects on decade-long timelines. Startups build, test, break, and rebuild in months. That speed difference is why $14 billion of private capital has flooded into this space. Investors aren't betting on fusion physics. They're betting that nimble engineering teams can close the gap faster than a 35-nation consortium managed by committee.

—

The Real Race: Cost Per Megawatt

The question was never "can fusion work?" The physics is settled. Stars do it every day.

The question is: can we do it cheaply enough to matter?

ITER costs €22 billion and climbing. A single reactor. No commercial output. When (if) it achieves Q=10 in 2039, it will have taken over 40 years from initial design.

Helion's valuation is $15.5 billion. CFS has raised $2.4 billion. If either company delivers a working power plant — even a small one — the cost per megawatt will make ITER look like building a cathedral to prove that stone can stand up.

The fusion industry's own association says the majority of private companies expect commercially viable plants by the mid-2030s. Some say sooner. The DOE's roadmap targets the same window.

That means the 2030s could see fusion power plants coming online — not from the €22 billion government project, but from the startups that decided to move faster.

—

What This Means for AI Infrastructure

In our last piece, we covered why the AI boom is driving a datacenter energy crisis. The connection here is direct.

Microsoft — one of the largest AI infrastructure operators in the world — has already placed its fusion bet. Their PPA with Helion isn't philanthropy. It's a strategic hedge: if AI compute demand keeps doubling, and grid capacity keeps stalling, the company that secures its own clean energy source wins the AI race.

The same logic applies to every hyperscaler. Google, Amazon, Meta — all of them need power at a scale that current grids can't deliver. Fusion, if it arrives in the 2030s, could be the unlocking technology for the next decade of AI infrastructure.

And the companies that get there first won't be governments. They'll be startups.

—

Adam Silva is the founder of Adam Silva Consulting, building agentic commerce infrastructure for the AI economy.