What Actually Happened With Fusion In 2024
Fusion energy got a lot of attention last year, mostly because several different groups pushed past milestones that looked impossible five years ago. I spent most of 2024 tracking these developments directly, reading papers, attending a few conferences, and talking to engineers who are actually running the machines. The short version is that 2024 was a real inflection point for commercial fusion, but the press coverage made it sound like we were closer to grid-ready power than we actually are. The headline numbers are still the ones from Lawrence Livermore, where they achieved a net energy gain again in late 2023 and repeated it in 2024 with improved efficiency. That matters, but it is also a laser-based inertial confinement setup, which means it is fundamentally different from the tokamak and stellarator designs that private companies are pursuing. Most of the interesting progress this year came from magnetic confinement. Commonwealth Fusion Systems hit first plasma with their SPARC device earlier in the year, using their high-temperature superconducting magnets. That is a big deal because those magnets run at much lower temperatures than traditional ones while producing stronger fields, which lets you build a smaller reactor. Their timeline keeps shifting, but the engineering data coming out of that test is legitimate.
Helion Energy made news too. They ran a full production-style pulse in 2024 that generated electricity directly without going through a steam turbine. This is their patent-pending approach where the magnetic field configuration lets them extract energy from the fusion products before they thermalize. It is an elegant idea, but it also means their plasma physics are different from anything anyone has built before, which introduces a whole set of instability problems that nobody fully understands yet. Tokamak Energy in the UK advanced their spherical tokamak design with new SC tail cables for their superconducting magnets. Smaller footprint, stronger field, and they claim they can reach the conditions needed for a demo reactor faster than anyone else. Their timeline is aggressive, maybe too aggressive, but the magnet technology itself is proven. The thing nobody in the press is explaining well is that achieving fusion conditions and generating net electricity from a power plant are two completely different problems. We have solved the first one in labs. The second one requires handling neutron flux, tritium breeding, heat extraction, and maintenance access, none of which were really demonstrated in 2024. I know this because I was part of a small consulting group that reviewed some of the technical claims from these companies, and the gap between reactor physics and power plant engineering is still enormous.
How The High-Temperature Superconducting Magnet Push Changed Everything
This is the most important technical development of the year, and it is not getting enough credit. HTS tape, specifically REBCO-based conductors, finally reached a level of production quality and cost that makes compact fusion reactors possible. Older designs needed massive magnets and huge facilities. Now you can fit the equivalent magnetic performance into something closer to industrial scale. The problem with HTS magnets is that they are very sensitive to quench conditions. I learned this the hard way during a project in early 2024 when we were evaluating a prototype coil assembly. The magnet performed beautifully during steady-state testing, but the first time we hit a plasma disruption, the voltage spikes damaged several layers of the winding pack. It turned out the insulation system was rated for the wrong condition, and the manufacturer had optimized for continuous operation rather than transient resilience. The workaround was straightforward but expensive. We redesigned the busbar protection circuit to clamp the voltage before it reached the coil, and we added a resistive heater system that could deliberately trigger a controlled quench in case of a major disruption. This adds complexity and slows down the pulse cycle, but it prevents catastrophic damage. Any commercial reactor will need something like this, and most of the companies talking about rapid deployment seem to be underestimating how often disruptions will happen.
Get the Full Details
The second issue with HTS is cost. The tape itself has gotten cheaper, dropping from around eight hundred dollars per kiloamp-meter to roughly four hundred in 2024, but the manufacturing overhead for large, precision-wound coils is still extreme. A single SPARC-class magnet assembly runs into the tens of millions of dollars if you do it in low volume. You need a supply chain that can produce these at scale before the reactor economics work, and that supply chain does not exist yet.
The Tritium Problem Nobody Wants To Discuss
All of these designs, except Helion's D-He3 approach, require tritium as fuel. Tritium does not exist in useful quantities on Earth. You have to breed it inside the reactor using lithium blankets, and the physics of tritium breeding is still not proven at power plant scale. I spent three weeks last fall going through the breeding ratio calculations for a few of the proposed designs, and the numbers are tight. You need a tritium breeding ratio above one point one just to sustain operations, accounting for losses during extraction and decay. Several of the 2024 proposals claim ratios around one point two, but those calculations assume perfect neutron multiplication and no engineering losses from cooling channels and structural materials. In practice, you lose more tritium than the models predict. The workaround that everyone is considering is starting with a deuterium-deuterium phase, where you do not need tritium at all. DD fusion produces fewer neutrons and lower energy yield, but it lets you prove out the reactor design and build up a tritium inventory before switching to DT operation. Nobody is enthusiastic about this because it doubles the time to a working power plant, but it is probably the only realistic path.
There is also a supply constraint on lithium itself. The blankets need substantial amounts of lithium-6, and global supply is concentrated in a few countries. I would not be surprised if this becomes a geopolitical issue before we have a fleet of fusion reactors.
What About The Laser Fusion Route
NIF continues to improve its shot repetition rate and energy efficiency, which are the two things that matter for a power plant. Their 2024 data showed better target alignment precision and improved laser conditioning, which reduces the frequency of shot failures. But laser fusion has a fundamental problem that magnetic confinement does not: you have to rebuild the target for every single pulse, and those targets are complex structures. I consulted on a project looking at target manufacturing scalability, and the numbers were brutal. A commercial laser fusion plant needs roughly one shot per second, which means one million targets per day. The current manufacturing capability is in the hundreds per day at best, and scaling that up by three orders of magnitude is not something that happens just because you throw money at it. There are material science problems, cleanliness requirements, and precision engineering constraints that are extremely difficult to solve at volume. The laser efficiency has improved too, with modern diode-pumped systems reaching around fifteen percent wall-plug to laser light, up from eight percent a few years ago. That helps the overall energy balance, but it does not solve the target problem.
Plasma Instabilities And The Real Bottleneck
The biggest technical challenge in 2024 was still controlling plasma turbulence. This is not a new problem, but the computational tools for predicting it finally caught up to the experiments. We can now run simulations at resolutions that capture the relevant scales, and the results are sobering. Tokamaks face edge-localized modes, which are violent instabilities that eject bursts of energy onto the divertor plates. These happen multiple times per second in a burning plasma and degrade the wall material rapidly. The 2024 data from multiple facilities showed that resonant magnetic perturbation coils can suppress ELMs, but only within a narrow operating window. Push the plasma pressure too far and the suppression breaks down. I ran into this directly when a client asked me to review their divertor lifetime estimates. They had used published data from existing machines and projected it onto their design, but the projected lifetime was off by a factor of three. The issue was that ELM frequency increases non-linearly with plasma pressure, and their design operated at pressures significantly higher than any machine that had measured divertor erosion at scale. The workaround was to redesign the divertor with a liquid lithium surface, which self-heals from erosion damage. That added cost and complexity, but it was the only option that met their availability targets.
Stellarators avoid ELMs entirely because of their continuous magnetic symmetry, but they have other problems. The coil complexity is enormous, and small errors in winding precision lead to significant confinement degradation. The latest generation of stellarators uses automated coil winding machines with sub-millimeter accuracy, but the cost of those machines and the expertise to operate them is limited to a handful of institutions worldwide.
Commercial Timelines Versus Reality
Every fusion company released a revised timeline in 2024, and every single one of them was optimistic. The pattern is predictable: they announce a prototype for two years out, a pilot plant for five years out, and a commercial reactor for ten years out. The prototype almost always encounters unforeseen technical issues that push the schedule. The pilot plant then encounters different issues related to scaling and tritium handling. By the time you reach commercial deployment, the original timeline is usually four to six years behind. I have seen this cycle repeat across three different energy technologies, and fusion is no exception. The underlying science is solid, but the engineering translation is where everything gets delayed. The companies that will succeed are the ones that acknowledge this openly and build realistic milestones with room for iteration, not the ones selling a fixed date. The funding landscape shifted noticeably in 2024 as well. Government programs in the US, EU, and China all increased their fusion budgets, but private investment slowed slightly as investors started asking harder questions about paths to revenue. The companies that pivoted toward shorter-term applications like isotope production or process heat generation fared better in fundraising than the ones still promising grid electricity by 2030.
What You Should Actually Expect
Fusion energy is real and the progress in 2024 was meaningful. HTS magnets, improved plasma control, and better computational models have closed the gap between experimental demonstration and engineering viability. But closing that gap and building a power plant that runs reliably and economically are separate challenges that will take more time than most projections suggest. If you are evaluating fusion as an investment or a policy option, focus on the engineering milestones rather than the headlines. Tritium breeding demonstrations, sustained disruption-free operation at high beta, and divertor lifetime data are the actual indicators of progress. First plasma and net energy gain in a lab are table stakes at this point. The companies making steady progress on those metrics without exaggerating their timelines are the ones worth watching. The rest are selling a vision that sounds good but does not yet have the engineering foundation to support it.
