Fusion Story
The fusion story—how I see it.
Energy Scarcity: The Motivation
Progress at the civilizational level is ultimately constrained by a handful of foundational bottlenecks:
- Access to intelligence and knowledge
- Access to abundant, affordable energy
- Access to basic needs: food, shelter, and infrastructure
- Access to health and longevity
These are not independent problems. Advances in any one of them tend to unlock progress in the others. Energy, in particular, is a force multiplier: it powers the infrastructure that delivers food and clean water, it enables the compute that drives scientific discovery, and it determines how much human effort can be redirected from mere survival toward more ambitious ends.
A critical enabler for solving all three of the other challenges is access to vast amounts of scalable, cheap energy. Energy empowers societies and, crucially since 2025, it empowers the AI-driven intelligence explosion. It is highly plausible that by combining abundant energy with democratized intelligence, humanity's greatest problems will fall one by one.
The key question is: what technology can provide this scalable access to vast amounts of energy? In the long term, there are only two viable, scalable options: fission and fusion. Renewable energies like solar and wind, coupled with batteries, are essential, but ultimately serve as bridge technologies.
Fission
Nuclear fission is a technology that already allows developed countries like France and Sweden to operate almost fully decarbonized electricity grids. It is extremely versatile, and newer reactor designs—particularly Generation IV reactors—promise exceptional safety.
Fuel availability is sometimes raised as a concern, but it shouldn't be. The uranium available today would provide about 0.35 million TWh if burned in traditional light water reactors—enough for roughly 12 years of today's global electricity demand. However, if burned in fast breeder or molten salt reactors, its utilization increases dramatically, extending our supply to around 150 years. Beyond that, uranium is absurdly abundant in seawater and could be extracted once terrestrial reserves are depleted, though its economic viability remains to be seen.
Fission's challenges are widely debated, centering on two main drawbacks:
- Safety and Public Perception: Fission is not intrinsically safe (though modern designs come close). The reactor disasters at Fukushima and Chernobyl left a lasting psychological scar, leading to widespread public distrust. This distrust is reflected in the heavy regulations governing the technology.
- Nuclear Waste: Fission produces high-level radioactive waste that must be securely stored for timelines that may outlast human civilization. Storing or transporting this waste takes decades of planning, raising both societal and moral questions about burdening future life forms.
Today, the cost of fission-generated electricity is intrinsically linked to the immense regulatory requirements that must be met. It remains an open debate whether we can design less restrictive regulations—or safely deregulate the technology—to reap the benefits of lower electricity prices.
What about Solar + Batteries?
Solar and wind are fantastic solutions for our immediate clean energy needs; it's no surprise that solar is currently the fastest-growing clean energy technology. Standard lithium-ion batteries do a great job of balancing the day-to-night fluctuations of solar power.
However, to compensate for the longer fluctuations of wind power, grids require either long-term storage or dispatchable energy sources. To visualize the scale of the storage problem: storing just two weeks of electricity demand for Germany would require installing the battery capacity of 200 million electric cars. Using current technology (at roughly $20,000 per car battery), this would cost a staggering $4 trillion. It is highly questionable whether batteries, or any chemical energy storage solution, will ever be economical enough for large-scale, long-term grid supply.
A stable electricity grid without long-term storage or significant energy imports seems infeasible. Intermittent sources like solar and wind require a baseload, a dispatchable energy source, or massive imports to match a stable demand. Without long-term storage, this dispatchable source would need to be capable of supplying the full demand on its own. This begs the question: if we need to build that full dispatchable capacity anyway, why not just run those power plants continuously to get cheaper electricity?
Could we just produce or buy green hydrogen? Yes, but the sheer quantity required is daunting. Providing Germany with two weeks of energy via hydrogen would cost on the order of $20 billion. For that amount of money, we could develop and build several fusion power plants.
The most informative way to envision future energy grids is through scenario simulations using different price points for carbon-free baseload power. These simulations consistently show that if a baseload power source can be built for around $5/Watt, it will resiliently outcompete other setups in a decarbonized grid (similar conclusions are drawn across multiple research papers).
Ultimately, a heavily renewable grid will force adaptation on the demand side, rather than flexibly matching supply to demand. In grids dominated by renewables, strong price fluctuations will force consumers to adapt. While this won't necessarily decrease our quality of life, it will impose major restrictions, particularly in industrial settings where it remains unclear if technology can compensate for the downsides.
Is a combination of solar and hydrogen imports the long-term solution for an industrialized Europe? Likely not.
Fusion
Fusion is often praised as the ultimate form of energy—after all, it powers the stars and, by extension, all life on Earth. While I have dedicated my professional career to enabling fusion on Earth, the complete story of fusion is more nuanced.
Unlike fission, where the challenge is preventing the reaction from running away, fusion requires massive kinetic pressure confined within a specific volume to occur at all. Creating this pressure on Earth requires engineering extreme pressure gradients, which is both difficult and expensive.
Currently, fusion is a high-technology endeavor relying on superconductors, cryogenics, extreme vacuum technology, and high heat-flux materials. It requires engineering within extreme magnetic fields, under severe neutron and gamma irradiation, while managing corrosion and erosion. Like fission, fusion requires radioactive activation control, though to a much lesser degree. Fusion does not generate long-lived radioactive waste, nor does it run the risk of a runaway meltdown. While fusion environments cause materials to degrade—requiring a replacement of internal components every few years—it remains a fundamentally safe process.
Simply put: making fusion work commercially is not easy. So why make the effort?
The answer lies in its unparalleled energy density. The power that the first wall of a fusion reactor sees is about 1,000 times stronger than what a solar panel receives in bright daylight. Moreover, once a fusion power plant is built, it requires a negligible amount of fuel.
To illustrate how little fuel is needed: the deuterium flowing naturally in a single standard-sized German river, burned in a fleet of stellarators, could power the equivalent of 1,000 Earths.
Because fusion fuel is effectively free, the economics of a fusion power plant are entirely defined by its capital costs (CAPEX)—the superconductors that last 40 years, the steel that degrades under neutron bombardment, and the tungsten facing the plasma.
Ultimately, commercial fusion is about building the most powerful machines as cheaply as possible.
I am often asked how fusion could ever be economical when a natural gas plant only costs $1/W to build. Even if first- or second-generation fusion plants cost $4–$8/W, the levelized cost of electricity (LCOE), assuming good availability, would sit around $20–$40/MWh. Fission currently sits at over $100/MWh. How is fusion cheaper?
Because fusion has the prospect of being regulated in a vastly different, more favorable environment.
This highlights the critical importance of regulatory frameworks. Does this mean fusion shouldn't be regulated? Of course not. A fusion plant is not harmless; it is a machine containing immense magnetic forces comparable to the gravitational weight of the Eiffel Tower. The reactor vessel must contain neutron-activated materials, parts of the blanket are chemically toxic, and the machine's short-lived radioactive components must be handled appropriately. It must be regulated correctly, but without the crippling over-regulation of fission. Regulators hold the future price of fusion electricity in their hands.
Regulations aside, achieving commercial fusion comes down to manufacturing and assembling components efficiently. This is where AI and advanced intelligence will be tremendously helpful. Fusion machines—stellarators in particular—are heavily intelligence-enabled. The upcoming explosion in AI won't just make fusion a reality; it will directly result in more powerful, cheaper fusion capability.
ITER and the Steady March of Progress
I sometimes hear people ask, "How do you plan to achieve what scientists have failed to do for decades, like at the ITER project in France?" or "Forget it, fusion is just an unsolved academic exercise."
This always makes me pause. We aren't just tinkering in the dark; we know exactly where the finish line is. In technical terms, it is the Lawson criterion ($Q > 1$, or $nT\tau > 10^{21} \ \text{keV/m}^3\text{/s}$).
Recent experimental machines have been approaching this target at an exponential rate. In fact, laser fusion has already breached $Q > 1$ (though it still needs to solve repetition rates and overall efficiency). We are steadily building machines that are increasingly powerful, though correspondingly more expensive. ITER itself is a generational mega-project born from the promising results of JET in the 1980s, designed to definitively prove the science at scale.
The Promise of Stellarators
Looking ahead, I am convinced that we will eventually unlock pure deuterium fusion. The most likely path forward involves tuning stellarator reactors for extreme MHD (magnetohydrodynamic) stability and ultra-strong turbulence suppression. While pure deuterium fusion is out of reach for our current generation of machines, the long-term prospective is incredibly bright.
Prospect: How will this new world play out?
To be expanded...
Explosion in Intelligence
To be expanded...
Questions for Journalists
To be expanded...