Nuclear Fusion Took Big Leaps in 2025. Here’s What Mattered Most

The promise of nuclear fusion feels simple. Just as stars fuse hydrogen into heavier elements to produce energy, a fusion reactor generates massive amounts of energy by combining lightweight particles with minimal risk to the environment.

That sounds like the dream scenario for clean energy. That said, the many challenges of nuclear fusion make it seem more like fantasy than reality; fusion is always ten years away, as the joke goes.

But that doesn’t mean scientists have given up. In fact, quite the opposite. 2025 witnessed a surge in fusion research from both established and newcomer stakeholders. Scientists reported increased fusion energy outputs compared to previous years, improved reactor hardware, and a wide range of experimental and theoretical developments supporting different parts of fusion energy. Here are the most notable advancements in fusion from the past 12 months.

A big fusion lab more than doubled its energy output

Let’s start with the Lawrence Livermore National Laboratory’s National Ignition Facility (NIF)—arguably the largest, most established fusion research facility in the world. This year, NIF continued firing its reactor, tweaking even the most minute details of its setup in hopes of the slightest improvement in energy yield efficiency.

They must have been doing something right. This year, NIF reported that it had broken its own 2022 record for fusion yields, more than doubling energy production.

China announced ambitious nuclear goals—and delivered

In 2025, fusion research became more international. Perhaps most importantly, China entered the race for fusion with ambitious targets for nuclear power, which it intends to hit by 2030. The plan includes targets not only for fusion but also for nuclear fission, the process in which reactors split heavy atoms to generate energy.

Whether China will deliver on fusion goals remains to be seen, but the country has certainly been aggressively rolling out new technologies and reactors for its fission programs.

An international project continued to make progress

On the other hand, ITER—the international collaboration aiming to run the world’s largest fusion reactor by 2034—announced a steady stream of small milestones this year. A newly revised project plan took effect in January, and member states shared updates on their contributions to the massive reactor, including the main superconducting magnet system and the wires that keep it running.

AI entered the fusion scene. No nightmares…yet

For better or worse, AI permeated virtually every corner of society this year. Fusion was no exception, although, fortunately, these AI programs proved to be rather helpful to researchers. This became especially pertinent for predicting experimental results, given the tremendous amount of time, money, and other resources required to perform a single experiment with a reactor.

For example, NIF researchers taught an AI model to predict whether a reactor run would achieve ignition. It performed so well that it ended up saving scientists considerable time and cost in setting up reactor runs. Another team at MIT combined physics with machine learning to create a program that predicted plasma behavior inside reactors with stunning accuracy.

Smaller stakeholders made themselves known

2025 also brought some memorable headlines from startups and smaller-scale labs, each testing their own ideas for bringing fusion to market. For instance, in the spring, California-based firm TAE Technologies introduced Norm, a reportedly cheaper yet more powerful alternative to older, bigger designs. A couple months later, energy startup Marathon Fusion proposed a reactor setup that—essentially—turns mercury into gold for possible revenue opportunities outside the production of energy.

In more academic pursuits, the Berlinguette lab at the University of British Columbia in Canada created a proof-of-concept for Thunderbird, a bench-top reactor operating on a combination of plasma science and electrochemistry.

Key incremental progress

There were several research projects this year not necessarily linked to building and deploying reactors but instead on the smaller, sometimes theoretical, details of fusion.

To take a flashy example, in July physicists at the SLAC National Accelerator Laboratory blew up gold with giant lasers and disproved a renowned model about the state of matter under extreme conditions. This offered the researchers a closer look at how matter responds to fusion-inducing environments.

Similarly, a separate study by University of California San Diego researchers investigated how diamond—a common ingredient in fusion fuel capsules—responded to the stress of fusion experiments.

Of course, I’d be remiss not to mention a winning experiment from this year’s Gizmodo Science Fair, in which a team led by Texas A&M University researchers discovered a way to safely extract lithium-6, a key fuel for nuclear fusion. The project took off from several unexpected scientific connections the team drew between electrochemistry and fusion engineering, highlighting the inherently multidisciplinary nature of fusion energy.

Honorary mention

I’ll end this list with this video of the wild scene inside a nuclear reactor. Ambitious tech projects founded on lofty, obscure science can sound like the crazy ramblings of scientists and engineers, and honestly that’s a fair assessment for some of the hype. Still, as the past year shows, these experiments are very real, with real people working on real science.

Plasma is better in colour! Watch one of our latest #plasma pulses in our ST40 tokamak, filmed using our new high-speed colour camera at an incredible 16,000 frames per second.

Each pulse lasts around a fifth of a second. What you’re seeing is mostly visible light from the… pic.twitter.com/jWKmcl0tEx

— Tokamak Energy (@TokamakEnergy) October 15, 2025

Here’s to another (reasonably) momentous year of fusion science—and, hopefully, the last decade it’s 10 years away.