The Sun in a Bottle: Nuclear Fusion and Quantum Computing
Book: Quantum Supremacy: How the Quantum Computer Revolution Will Change Everything Author: Michio Kaku ISBN: 978-0385548366
Every twenty years, physicists claim that fusion power is just another twenty years in the future. This joke has been going around for decades and it’s still accurate. Chapter 15 takes on nuclear fusion, why it’s so hard, and whether quantum computers could finally break the cycle of delays.
As someone who works with complex systems every day, I appreciate the engineering honesty in this chapter. The physics of fusion is solved. The engineering is what keeps failing us.
Why Does the Sun Shine?
Kaku starts with the basics. The sun is powered by fusion. Hydrogen nuclei smash together to form helium, and a tiny bit of mass disappears in the process. That missing mass becomes energy, exactly as Einstein’s E = mc2 predicts.
One gram of heavy hydrogen can produce 90,000 kilowatts of electrical energy. Equivalent to eleven tons of coal. The fuel is hydrogen, which you can extract from seawater. Practically unlimited.
On paper, fusion is the perfect energy source. No carbon emissions. No meltdown risk. No long-lived radioactive waste. The by-product is helium, which is actually commercially useful. If a fusion reactor has an accident, the reaction just stops. Compare that to Chernobyl, where the fission process kept generating heat even after the reactor was shut down, eventually blowing the roof off and spreading radioactive material across Europe.
The Tokamak: A Doughnut Full of Problems
The most popular fusion reactor design is called the tokamak. Russian invention. The basic idea is straightforward: take a doughnut-shaped chamber, wrap it with wire coils, inject hydrogen gas, and heat it to millions of degrees until it becomes plasma. The magnetic field from the coils keeps the plasma from touching the walls.
Simple concept. Brutal engineering.
In a star, gravity compresses hydrogen evenly from all sides. Gravity is monopolar, it only pulls in one direction, so stars form naturally. That’s why we see billions of them. Magnetic fields are bipolar though, they always have a north and south pole. Trying to evenly squeeze plasma inside a doughnut with magnets is like trying to evenly squeeze a balloon animal. You push one spot and air bulges out somewhere else.
This is why fusion has been “twenty years away” for seventy years. The plasma keeps becoming unstable. Tiny irregularities in the magnetic field get amplified because the plasma itself has its own magnetic field that interacts with the reactor’s field. A feedback loop. Small wobbles become big wobbles, and sometimes the plasma touches the reactor wall and burns a hole through it.
ITER and the Competition
The world’s biggest bet on fusion is ITER, the International Thermonuclear Experimental Reactor, funded by thirty-five nations. A monster machine weighing over 5,000 tons, with a torus sixty-four feet in diameter. Its magnets generate a field 280,000 times stronger than Earth’s magnetic field.
Physicists measure fusion efficiency with a number called Q. Q equals energy out divided by energy in. When Q equals 1, you hit breakeven. The world record before ITER hovered around Q = 0.7. ITER is designed to eventually reach Q = 10.
ITER is slow though. Testing started around 2025, full power maybe by 2035. Even then, it won’t connect to the electrical grid. The next step after ITER is DEMO, planned for 2050, designed for Q = 25.
Commercial fusion power before mid-century? Unlikely. As BBC’s Jon Amos put it: “Fusion is not a solution to get us to 2050 net zero. This is a solution to power society in the second half of this century.”
Competitors are moving faster. The SPARC reactor from MIT uses high-temperature superconductors, a class of ceramic materials discovered in 1986 that superconduct at 77 degrees Kelvin instead of near absolute zero. The coolant changes from liquid helium at $100 per pound to liquid nitrogen at $4 per pound. That shifts the entire economics.
SPARC raised over $250 million from investors like Bill Gates and Richard Branson. Pocket change compared to ITER’s $21 billion, but the approach is different. Smaller, cheaper, faster iteration.
On the laser side, the National Ignition Facility in California took a completely different path. Instead of magnets, NIF uses 192 laser beams focused on a pea-sized pellet of hydrogen fuel, compressing it to 350 billion times atmospheric pressure. In December 2022, NIF made history by achieving Q greater than 1 for the first time. Scaling that up to power a city is another problem entirely though.
Where Quantum Computers Come In
The equations governing plasma behavior and magnetic fields are known. The problem is that these equations are tightly coupled. The plasma affects the magnetic field, the magnetic field affects the plasma, and small perturbations can cascade into instabilities.
Classical computers struggle with this kind of coupled simulation. Quantum computers, being quantum mechanical themselves, could potentially simulate the plasma-magnetic field interactions directly. Instead of spending $10 to $20 billion building a reactor and finding out the design doesn’t work, you could test virtual reactor designs on a quantum computer. Change parameters, run the simulation, check stability. Repeat until you find an optimal configuration.
AI is already getting involved. DeepMind has been used to modify fusion reactor operations at the Swiss Federal Institute of Technology. Combine AI with quantum computers and you get a system that can vary magnet configurations and analyze the results to increase the Q factor.
There’s another angle too. Nobody fully understands why high-temperature ceramic superconductors work. They’ve existed for over forty years and there’s still no consensus on the theory. A quantum computer could simulate the electron distribution inside these materials and figure it out. Once you understand the mechanism, you can systematically search for better superconductors instead of discovering them by accident.
My Take
One of the more grounded chapters in the book. Kaku doesn’t oversell fusion. He acknowledges the decades of broken promises and explains the physics behind the delays clearly. The balloon analogy for why it’s hard to squeeze plasma in a torus is genuinely helpful.
The quantum computing angle is speculative but logical. If you can simulate plasma behavior accurately, you save billions on physical prototyping. The coupling between AI and quantum simulation for reactor optimization feels like it could actually happen within a decade or two.
I keep coming back to the timeline problem though. Fusion won’t save us from climate change. It’s too slow. The real value of fusion is as a long-term energy foundation for the second half of this century. Quantum computers might be the tool that finally gets us there. Not by solving the physics, but by solving the engineering.