In December of 2022, at the Lawrence Livermore National Laboratory in California, the National Ignition Facility produced 3.15 megajoules of fusion energy from 2.05 megajoules of laser input. Humanity recreated the physics of a star in a building. By aiming 192 lasers at a fuel capsule the size of a peppercorn, a fusion reaction on Earth produced more energy than was put into it. This is the first time in history that humans have created a net energy gain in fusion, or what is scientifically termed “Q > 1”, where Q is:

This matters because fusion is what literally powers the sun. The sun's core (roughly 15 million degrees Celsius) is hot enough that hydrogen atoms slam into each other and fuse into helium, releasing enormous amounts of energy (Britannica). This is done with heat and pressure. The sun's gravity compresses the core to 265 billion times Earth's atmospheric pressure and holds it there for billions of years. So even though fusion at that temperature is exceedingly slow, the sheer scale and duration add up to a star. On Earth, with no gravity to help, we make up for the lack of pressure with more temperature: 100 to 150 million degrees Celsius, ten times hotter than the sun's core.

All of that complexity, done on Earth, was supposed to be the hard part. Physicists had been theorizing and working on this problem for 75 years (Fusion from then to now), asking if a fusion reaction on Earth could ever produce more energy than we put into it. We proved it, but oddly enough, this is where the actual hard part comes in: engineering. 

In order to create and deliver those 2.05 megajoules of energy, the lasers had to draw 300 megajoules from the grid. While the National Ignition Facility achieved a Q > 1 net gain, the “engineering gain” (Q_eng > 1) had not been achieved.

So while we solved the physics, we still have a major engineering problem. Fusion commercialization is currently blocked by materials and efficiency. Fusion releases most of its energy as neutrons, which are electrically neutral, so they fly straight through the magnets holding the plasma and slam into the reactor's lining, wrecking it. And there's no facility on Earth that can test materials against them.

The distinction: 

Our intuition says that physics will be hard to understand, while engineering should be practical and solvable. Fusion completely inverts this. We can hold a 150 million-degree plasma in midair with magnets. What we can’t do is build the wall behind it: the neutrons ignore the magnets, and nothing we know of survives these neutrons for 30 years. The reason is that engineering must be cost-effective and last for long periods, providing continuous, low-cost, and reliable solutions. Physics problems can be solved once, in a lab, for a fraction of a second. The science proves it works, while engineering proves we can harness it effectively. 

What is even more interesting is that we keep solving more physics problems. There was a lot of concern that the 2022 ignition shot was a fluke, but since then, the NIF has produced 5.2 and 8.6 megajoules in subsequent attempts (nearly 3x the initial amount). We also used to believe that fusion's own exhaust (Helium) would cause turbulence that made it hard to maintain heat until simulations of the Smallest Private-Funded Affordable Robust Compact Reactor (SPARC) and International Thermonuclear Experimental Reactor (ITER) actually found out the opposite: the exhaust suppressed the turbulence and boosted heating! 

Meanwhile, the Department of Energy’s (DOE) roadmap still lists six unsolved engineering areas (structural materials are still partly physics), neutron test facilities do not exist yet, and no one has produced tritium (the fuel) at scale. Each of those is worth understanding on its own.

The three engineering challenges that stand out (none of which have been addressed at the power-plant scale) are Materials, Fuel, and Heat. 

Materials: No metal has been proven to survive fusion neutrons (seven times more energetic than fission neutrons) for a plant’s expected 30-year life. We do not even have a facility capable of proving it. Europe is building a test facility (IFMIF-DONES) in Spain, but it is not running yet. 

Fuel: Fusion runs on tritium, which occurs in nature only in trace amounts, with today's supply being a byproduct of Canadian fission reactors. The world's entire stockpile is about 20 kilograms, and a 1-gigawatt plant (roughly what powers a mid-size city) burns 55 kilograms a year. So the reactor has to breed its own: line the wall with lithium, and the escaping neutrons convert the lithium into tritium. It works on paper, but nobody has done it at scale.

Heat: The plasma vents its waste heat onto one component at the bottom of the fusion reactor, which absorbs about what a rocket engine nozzle does. A rocket fires for minutes while fusion has to hold for decades.

Companies like Commonwealth Fusion Systems originally aimed to have SPARC operating in 2025. As of 2026, the machine is roughly 75% built, with first plasma now targeted for 2027 and net energy gain shortly after. Even then, SPARC won't produce electricity. That's ARC, the follow-on plant, sometime in the early 2030s.

Nearly every fusion milestone to date has slipped and the reason seems to always be that the engineering complexity gets underestimated. 

Private investment has surpassed $15bn, and Microsoft actually signed a power purchase agreement with Helion for electricity before 2030. The International Energy Agency (IEA) now tracks fusion alongside other emerging energy tech and has listed a first demonstration plant as a milestone to watch by 2030. 

In June 2026, the DOE published its Fusion Science and Technology Roadmap, the most detailed public blueprint yet for closing the gap. Its stated goal is to deliver the public infrastructure that supports a private fusion industry: test facilities, materials databases, neutron sources. Not experiments to find out whether fusion works.

We have answered the question of does nature even permit this on Earth? (physics). Now we need to answer: can we make it efficient, durable, and repeatable? (engineering).

Takeaway: For 75 years, fusion's problem was physics: does nature permit this to exist on Earth at all? December 2022 answered yes. What is left is the engineering of materials that survive neutrons, fuel that does not exist yet, and heat nothing can hold. Instead of requiring discovery, it requires engineering. 

Have a great weekend,

Josh