The Illusion of Fusion, the Gluttony of AI, and the Near-Term Necessity of Fission

By Eric Anders

As the AI boom accelerates, the American energy system is heading toward a breaking point. Data centers are multiplying. Large language models like GPT-4o and Claude 3 are only the beginning—AI workloads are projected to triple U.S. data center electricity demand by the end of the decade. Meanwhile, domestic energy policy is failing to catch up. Climate commitments are sliding backward. Grid bottlenecks remain unresolved. And energy prices are expected to rise, not fall, in the coming two years.

Derek Thompson recently pointed out that if MAGA Republicans succeed in rolling back clean energy incentives—as they’re currently attempting through Trump’s tax proposals—electricity prices will rise just as demand explodes. The GOP’s backward-facing agenda strips subsidies from wind and solar while offering minimal structural reform for new clean energy technologies like nuclear. It’s the worst of both worlds: an energy-constrained economy and no credible decarbonization plan.

Tech companies like Google and Microsoft seem to understand the danger. They’ve begun signing power purchase agreements with nuclear developers—Kairos Power, Constellation, X-Energy, and even fusion startups like Commonwealth Fusion Systems.

But let’s be clear: fusion is not the answer in the next two decades. These are speculative offtake agreements, not production timelines. No fusion reactor has delivered net energy to the grid. Betting on fusion today is like planning your 2030 food supply based on algae farms on Mars.

If the U.S. is serious about climate goals, energy abundance, and geopolitical resilience, it has only one viable option in the near term: fission.

Fission as the Only Scalable Hope

Fission is the only technology capable of delivering clean, firm, scalable baseload power in the 2030–2040 timeframe. That means:

• SMRs (Small Modular Reactors) that can be factory-assembled and deployed rapidly

• Life extensions and uprates of existing nuclear plants

• Next-gen designs already operating or under construction in other countries

But here’s the paradox: while SMR technology is advancing globally, the U.S. remains paralyzed by its own regulatory inertia. The Nuclear Regulatory Commission (NRC) continues to treat advanced fission with the same 1970s rulebook it applies to traditional gigawatt-scale light water reactors. As a result:

• No U.S.-designed SMR has broken ground.

• NuScale, long touted as a leader, recently had its Utah project canceled after years of delay and rising costs—due largely to regulatory uncertainty and permitting delays.

• Developers like Kairos Power and X-Energy are stuck in an expensive, protracted design certification process that could take another five to ten years.

Meanwhile, SMRs Are Working—Abroad

Other countries have moved ahead. Russia’s Akademik Lomonosov, a floating SMR barge, has been producing electricity in the Arctic since 2020. China’s Linglong One—a 125 MW pressurized water SMR—achieved major construction milestones in 2023 and could be grid-connected soon. South Korea’s SMART reactor, and Canada’s GE-Hitachi BWRX-300, are also leading global deployment efforts.

If these technologies are available and being deployed abroad, why aren’t we importing them? Why not streamline approval for allied-nation SMRs that already meet international safety standards?

The answer lies in the same outdated regulatory structure that stifles our own domestic development. The U.S. makes it all but impossible to certify foreign nuclear designs for deployment—either through procurement restrictions, licensing barriers, or national security red tape. If we can buy French submarines and Korean EV batteries, we can buy SMRs from our allies. But only if the political will exists to deregulate those purchases.

The Renewables Subtext: Subsidies Without Reform

It’s not enough to blame the NRC. Part of the reason nuclear regulation remains untouched is that legacy environmentalism has fixated almost exclusively on renewables—and done so for decades. Billions in subsidies flow annually to wind and solar, yet no comparable effort has been made to update nuclear permitting, even as nuclear plants remain the largest source of zero-carbon electricity in the U.S.

This subsidy imbalance—combined with ideological hostility toward nuclear power—creates a political ecosystem in which renewables expansion is pursued without grid realism, while nuclear is left to wither. The result: a brittle, intermittent grid and rising carbon emissions as gas fills the gaps. And with AI and electrification on the rise, this model is not just unsustainable—it’s a national vulnerability.

Abundance Is Not Just About AI

Let’s also be clear: AI is not the only reason we need abundant energy. True abundance is about lifting industrial constraints, powering electrified transportation, desalinating water, producing green hydrogen, and rebuilding domestic manufacturing. The AI sector may be the immediate pressure point, but the energy hunger of the 21st century goes far beyond ChatGPT.

To meet that hunger, we need:

• Massive new generation capacity

• Clean firm power sources

• Reformed permitting

• Regulatory parity with renewables

• Open procurement for proven SMR technology, even if it’s not American

Conclusion: Fission or Failure

If the U.S. continues to treat fission as an afterthought—caught between anti-nuclear ideology on the left and fossil nostalgia on the right—we will miss the narrow window for energy abundance.

Fusion is decades away. Renewables are essential but incomplete. Gas is dirty and geopolitically risky.

Only fission can anchor the coming surge in demand with clean, scalable, politically independent power.

And if we want to compete in the energy future—whether it’s AI, hydrogen, industry, or climate—we need to unshackle nuclear innovation now.

Not in 2045.

Not after the next election.

Now.