Fusion Is Not a Physics Race. It’s a Governance Contest.

Want to know how a country really thinks about power? Don't listen to the speeches. Look for the hardest project on the table and see who owns it; that usually tells you everything.

Right now fusion is the project to watch. It's absurdly hard, staggeringly expensive and full of uncertainty. The New York Times lays out a clear split: China is mounting a big state-led push while the United States is steering private companies forward, with government in a supporting role. Read as climate reporting, sure; read as I do, it's a governance memo, one that reveals who is willing to shoulder the biggest risks.

Fusion goes beyond a simple technology race. Think of it as a live-fire test: two very different systems for managing high-risk innovation are being tried out in the real world.

China treats fusion much like it did high-speed rail and, increasingly, AI, approaching it as a coordinated national campaign. State media and outside reporting point to a tight network of government labs, universities and state-owned firms gathered around flagship machines such as the EAST tokamak in Hefei and the proposed CFETR reactor. Beijing lays out the goals, funds the work and folds it into national energy and security plans. The timetables are bold: experimental reactors this decade; a prototype in the 2030s; commercial systems by mid-century. Central planners are thinking in 25- to 50-year horizons.

On paper it looks tidy: a clear chain of command and a single story about why the project matters. If a provincial official or institute head drags their feet, someone above them has the mandate and the budget to force things along. It feels like a chess match where one player actually sees the whole board.

In the U.S., the work is split between public research and private startups. The Department of Energy and the national labs provide a research backbone, while a swarm of companies races to turn designs into viable reactors. Federal fusion funding is roughly a billion dollars a year, by most counts, and private investment in startups has already topped several billion. Firms such as Commonwealth Fusion Systems, Helion, and TAE are chasing different reactor concepts, timetables, and business models. The government runs milestone programs and issues grants; the markets supply most of the capital.

This isn't chaos so much as a conscious bet on competition. The idea is that rival teams will iterate faster and pursue more creative designs than a centralized program could. There's a political logic too: the risk is spread out; if one company falters, Congress doesn't have to shoulder the fallout the way Beijing would if a national flagship collapsed.

So you end up with two very different answers to the same basic question: who takes on the burden of execution when the physics aren't cooperating and any payoff is decades off?

In China, the state takes on the bulk of responsibility, which makes coordination easier. The party can bundle related projects (superconducting magnets, advanced materials, factory integration) into a single strategic program. Funding can be guaranteed for decades, outlasting any one leader, and officials can order supporting infrastructure and grid upgrades into being. That shows up as long-term roadmaps and large, integrated campuses instead of a loose scatter of industrial parks.

Putting responsibility in the state's hands makes the system brittle. When the state controls the narrative and the purse strings, it also becomes the owner of failure. That creates pressure to smooth data, avoid risks for the sake of careers, and dress political wishful thinking up as engineering fact. If the physics or materials turn out tougher than expected, the whole program can get locked into a sunk-cost trap, because admitting a delay is treated as a political wound rather than a routine technical setback.

In the U.S., the primary burden falls on private actors. Founders and investors reap the upside and take the embarrassment when things go wrong. The government buys options through grants, procurement commitments and clearer regulations. The advantage is a kind of market pressure: poor bets die quickly when the money dries up, and teams can pivot design choices without waiting for a decade-long plan to be rewritten in a ministry office.

That approach tends to drift. Venture capital has a short fuse; when milestones slip and the next shiny thing appears, funding can evaporate just when steady, unglamorous work is required. National priorities fragment into a patchwork of term sheets, each tailored to a fund’s return timeline instead of system-level resilience. The end result is that nobody is responsible for making sure experimental breakthroughs, grid upgrades, regulatory frameworks and workforce pipelines all line up in the same place at the same time.

Timelines make the divergence obvious. Beijing speaks like an engineer planning a dam; its horizon stretches for decades and completion is treated as all but guaranteed. The American side sounds like a startup at demo day, full of bold claims about net power “within a decade” and each team running on its own slightly different clock. China risks taking its planning documents too literally. The U.S. risks not believing anything long enough to finish the job.

What really matters is who ends up owning the failure. If China’s fusion push stalls, the costs are socialized across the political system, and that can either force a brutal internal reckoning or, worse, entrench denial. In the U.S., when a major fusion company implodes the losses are mostly private; politicians hold a few hearings, shrug, and move on. That seems safer until you remember that genuinely transformative infrastructure needs someone who can’t simply walk away.

At root this is a question of agency. If fusion actually works, whoever proves it first won’t just sell reactors; they’ll also export a playbook for how to make massive, slow, civilization-scale bets actually happen.

If China’s command-and-control model gets there first, people will see it as evidence that only a disciplined state, one that can ignore quarterly earnings and shrug off public dissent, can marshal the resources for projects at civilization scale. That logic would tempt other governments to centralize power and treat citizens as components of a five-year plan rather than as stakeholders in the system.

If the U.S. mix of public scaffolding and private initiative delivers, the signal is different. It would show that you can meet existential challenges with state support and distributed action, with markets, startups and independent labs shouldering extreme technical risk without a single central commander. That outcome would make other societies more likely to keep pluralism and competition in their designs, even when the stakes are civilization-sized.

The fusion race isn’t really about who flips on the first commercial reactor. Rather, it will decide which model becomes the default way we marshal people and institutions for problems that are too big for a single lab and that last longer than an election cycle.

At some point a government or a coalition of firms will announce that fusion is no longer just a physics experiment but a product. When that happens we won’t learn only about energy output. In the clang of steel and the hiss of plasma we’ll see which approach actually worked, whether a risk-bearing centralized state pushed it across the line or a loose, competitive network of firms and labs did. That choice will have consequences far beyond the grid.

Sources

• https://www.nytimes.com/2025/12/11/climate/china-fusion-energy.html
• https://asiatimes.com/2024/12/us-falling-behind-china-in-race-to-nuclear-fusion/
• https://www.wsj.com/world/china/china-us-fusion-race-4452d3be
• https://www.geekwire.com/2025/were-definitely-on-the-back-foot-u-s-risks-losing-fusion-energy-race-to-china-industry-leaders-warn/