Assuming that AI’s accelerated demand will still be around in the mid-2030s to “lift all boats” is probably not the best answer.
At The Fusion Report, we fully believe in a future where fusion plays a key role in our global energy picture, and particularly the energy picture for the US. But the question about what will be uniquely driving the demand for fusion energy in the mid-2030s when it expected to start coming online is a relevant question that needs a coherent answer. And it’s why a picture of oil tankers fits into an article about what will drive the demand for commercial fusion energy. But let’s start out the article with a discussion about what likely is and isn’t going to drive the demand for fusion energy in the mid-2030s.
1. For Fusion, All Energy Sources Aren’t Created Equal
The amount of electricity generated in the US in 2025 was 4,430 TWh; by 2030 it is expected to rise to 5,100 to 5,400 TWh annually, growing to 5,530 TWh in 2035. Of the electricity generated today, roughly 40% is from natural gas, 20% by coal, 19% by nuclear, and the remaining energy (~21%) from renewables. If fusion becomes commercially viable, it would most directly replace the electricity generated by fossil fuels, which is roughly 60% of all electricity today, but expected to drop to 50% by 2035. Because fusion is described as a low-carbon, on-demand source of power, it could do much of the same job these fuels do today without the carbon dioxide emissions that drive climate change. However, it is not likely to be able to replace renewable energy resources, which with battery backup can provide continuous baseload power like fusion at a much lower capital cost and risk.
2. Fusion Isn’t The Only Answer to Decarbonizing Electricity
Today, there are two obvious answers for decarbonizing electricity: solar energy/battery energy storage systems (BESS), and advanced nuclear fission. The combination of solar energy and BESS is here today and working at scale, in spite of the Trump administration’s attempts to stop it. The U.S. has installed a cumulative total of roughly 175 GWh of energy storage capacity to support its grid-scale solar deployments of 270 GW of solar generation, with storage deployed in over 90% of new grid-scale solar installs. Moreover, solidly Republican Texas leads the country with the largest share of upcoming and active utility-scale solar/BESS projects to be installed in the near future. And while total US solar generation (backed with BESS) only accounts for roughly 8.5% to 9% of total domestic electricity generation, even the US EIA’s conservative baseline projects that solar energy will account for at least 14% of total US electricity by 2035, with some independent estimates being as high as 30%.
A second choice that could decarbonize electricity is advanced nuclear fission energy. Currently, nuclear fission provides roughly 18%-19% of all electricity in the United States today, though over half of that capacity is over 30 years old. Moreover, while the Trump administration has spearheaded the idea of quicker regulatory approval for new nuclear fission plants, only two advanced reactors have been approved to date (Kairos Power’s Hermes Reactor in Oak Ridge Tennessee, and TerraPower’s Natrium Reactor in Wyoming), though several additional plants are in the process. And even advanced nuclear fission reactors do not solve the “long-lived nuclear waste” issue.
3. US Need For Fusion Isn’t As Strong As That In Europe or Asia
Today the situation in the Persian Gulf is far worse for countries in Asia than it I for the US or Europe because Asia imports roughly 84% of crude oil and 80% of liquefied natural gas through the Strait of Hormuz. Similarly, electric vehicle usage in Asia and in Europe is significantly higher than that in the US. Finally, growth in the demand for electricity in Asia and Europe are larger than that in the US, and the cost of electricity in Europe is 66% higher than that in the US. These factors make the need for fusion less urgent in the US than in either Asia or Europe.
4. The Electricity Demand for AI is Not Likely Going to Grow Fusion
Most optimistic forecasts show growth for the demand for electricity due to artificial intelligence (AI) to start to level off in the mid- to late-2030s, just about at the same time that fusion starts to become a meaningful electrical source. In fact, many more conservative forecasts show that demand will end up by 2035. A plateau by 2035 would most likely come from slower AI adoption plus hard infrastructure limits. The IEA says a “headwinds” case could push data-center energy demand to plateau beyond 2030, reaching about 700 TWh by 2035, with growth constrained by local bottlenecks and tight supply chains. Causes for an earlier slow-down could include the following:
- Slower-than-expected AI adoption, reducing growth for training and inference.
- Grid interconnection bottlenecks, permitting delays, and limited transmission buildout, which can delay new data-center capacity.
- Supply-chain constraints for power equipment, cooling systems, and other power transmission infrastructure.
- Efficiency gains in chips, servers, cooling, and workload scheduling that offset more usage.
- Local limits on water or power, which can cap data center expansion.
In any case, the growth in demand needed to kickstart commercial fusion energy in the US is not likely (or only partially likely) to be AI.
5. Government Support for Fusion in Europe and Asia is Stronger Than In The US
Europe and Asia tend to support commercial fusion more as a coordinated industrial strategy, while the U.S. has historically emphasized research, national labs, and long-horizon science. The EU has backed fusion through Euratom, EUROfusion, ITER, and a multi-year EU budget commitment, and countries like the UK, Germany, Japan, China, and South Korea have added national strategies, large public funding, and clearer commercialization goals; by contrast, U.S. federal support has been smaller relative to the scale of the challenge and more concentrated in the DOE’s Fusion Energy Sciences program, with commercialization funding still a limited share of the total. In practice, that means Europe and Asia are more likely to pair public money with industrial policy, regulatory planning, and shared infrastructure for private firms, while the U.S. has been stronger in startup innovation and private capital but less aggressive in direct government support for deployment. The gap is especially visible in China, where public spending and state-backed ecosystem building appear to be larger and more centralized than in the US.
Framing The Need For Fusion In The US More Effectively
Several things would be better drivers of the need for commercial fusion energy in the United States by 2035. One of the biggest drivers is the retirement of current electrical generation capacity, especially that from coal. Approximately 8%-10% (roughly 100 GW) of the current US electrical generation capacity is projected to retire by 2035, including 40% of the remaining US coal power plants. Additionally, the age of the current US nuclear fission electricity capacity is of significant concern, since over half of it will be 40 years or older by 2035, even though current plans do not have those plants being retired. And while solar plus BESS is a great answer for greenfield (new) power generation needs, using it to backfill existing power plant retirements is problematic, requiring expensive new transmission line capabilities.
The second biggest driver to spur the demand for fusion in the United States is likely to be electric vehicles (EVs); in this aspect, the Trump administration is (ironically) to be thanked. Because of the administration’s choices regarding ending EV tax credits and US automakers’ currently slow-rolling their EV plans, demand for EVs has probably pushed out a decade or more. U.S. EV adoption runs at 10% of market share for new EV sales, while in Europe it is roughly 27%, and in countries like China it exceeds 53%. If the US reached an EV adoption rates of 50%, it would increase total annual electricity consumption for charging by roughly 13%-15% (500 TWh-600 TWh) annually. This would require between 63-75 GW of new plants (in urban areas).
Conclusion: The Current AI Tide Will Not Lift Fusion Energy’s Boat
As the saying goes, a rising tide lifts all boats, but that doesn’t apply to boats that are still in dry dock being built. While misery (and business) ‘make strange bedfellows’, some seemingly perfect opportunities are more illusory than not. And so it is with AI for commercial fusion energy; no matter how big the demand is, fusion is not capable of supplying it in this decade. The likely outcome is that most AI data centers will be powered by natural gas, with some by nuclear fission, and maybe a few by fusion energy. And that is okay; things don’t always line up the way you’d hope for. But the worst thing you can have is the kind of an opportunity that you have no way of delivering against. And for commercial fusion energy, that is exactly what AI is, especially given its near-stranglehold that AI has on the venture capital market. But who’s to say that another opportunity won’t come up for commercial fusion energy, which is why fusion’s business strategy needs to start focusing on real opportunities instead of illusory ones.
