The Second Reactor

On June 16, 2026, the Washington State Department of Health did something no regulator on Earth had ever done: it licensed a fusion power plant. Two permits — one for radioactive materials, one for air emissions — issued to a company called Helion, for a machine named Orion rising beside the Columbia River in Malaga. If Orion works, it will be the first machine in history to sell electricity made the way stars make it.

And every one of those first electrons is already spoken for. Microsoft bought them in 2023 — fifty megawatts, contracted before the machine existed, destined for data centers. Google has pre-purchased two hundred megawatts from ARC, the next great fusion plant, going up in Virginia. Seventy years humanity has chased this fire, and now that it is finally guttering to life, its first job is running the cathedral.

Hold that against the machine my last two essays traced. The compute-industrial complex needed no villain to raise its gigawatt campuses on your rate base; it will need no villain to claim the flame of the stars. It only needs a number: the smallest fusion plant anyone knows how to build makes tens of megawatts, and the economics improve from there in one direction — up. Fusion, as our species currently knows how to kindle it, is the most centralizable power source ever conceived. Read it as design and it resolves: of course the cathedral bought the fire first. The fire only comes in cathedral sizes.

So this essay owes you the question a sovereign household would ask next: could the fire ever come home? Could there be a second reactor — not the one hung in the sky, but one in the garage, beside the wall that remembers? The honest answer comes in three parts: a joke, a wall, and a door.

The why This is the third essay in a series. The first named the machine; the second, Fight Big Tech and Big Govt — Bring the Power Home, laid out the sovereign property that answers it — the roof, the wall, the shelf, the cabinet, the doorkeeper — and the six-clause petition to fund it. This one takes the single speculative question that blueprint leaves open and refuses to answer it dishonestly.

The joke: you can already own one

Here is the part almost nobody knows. You can buy a fusion reactor today.

Philo Farnsworth — the same restless mind that invented electronic television — spent his later years perfecting a tabletop device that genuinely fuses atoms. His design, the fusor, is simple enough that teenagers have built working ones in garages; there is an entire hobbyist "Neutron Club" whose membership requirement is achieving fusion at home. And industry sells Farnsworth's descendants off the shelf right now: sealed-tube neutron generators, five to six figures, small enough to hold, lowered down oil wells and pointed at cargo containers every day of the week. Real deuterium, really fusing. A fusion reactor that fits in a closet is not science fiction. It is a catalog item.

The joke is the direction of the power flow. A fusor drinks kilowatts from the wall and returns nanowatts of fusion — short of breakeven by a factor of roughly a billion. Sixty years of tinkering has moved that needle almost nowhere. The fusor is genuine fusion, and it is the world's most exotic space heater. You can own the fire today; you will simply pay, forever, for the privilege of feeding it.

The wall: why the fire won't shrink

Why can't it just be improved? Because plasma hates being small.

Fusion power is generated in the plasma's volume. The heat that keeps the reaction alive leaks away through the plasma's surface. Shrink the machine and that ratio turns executioner — more skin per unit of fire, until the fire cannot keep itself warm. This single geometric fact is why fusion machines grew to the size of buildings, and it does not care how clever you are.

Even the revolution that shrank fusion proves the rule. A new generation of high-temperature superconducting magnets collapsed the minimum viable machine to a fraction of the old designs — and the result of that collapse, a tokamak called SPARC, is being assembled right now in Devens, Massachusetts, from eighteen magnets weighing twenty-four tonnes apiece, chilled to −253 °C, each carrying more than thirty thousand amps. First plasma is expected between now and next year; net energy shortly after. That machine is the most compact net-power fusion device our species knows how to attempt. It is a building.

And beneath the plasma problem lies a harder floor. The workhorse fusion reaction throws off neutrons at ferocious energies, and stopping neutrons is set by nuclear physics: roughly a meter of shielding and blanket in every direction, a thickness that better magnets do not shrink by one centimeter. A meter-thick shell around anything already overflows a garage bay — before you have added the magnets, the cryogenics, or the turbine that turns heat into current. This wall is not regulation, not conspiracy, not a permitting queue. It is arithmetic, and arithmetic does not attend council meetings.

Arithmetic does not attend council meetings.

The door: the fire the law forgot to chain

Now set fusion beside its older sibling, because the comparison hides the most interesting fact in this whole story.

Fission can be built small — container-sized microreactors are real engineering — and yet you will never own one, and it is worth understanding exactly why, because the reason is three walls, and only one of them is made of opinion.

So fission's ceiling, even in the best regulatory future your petitions can win, is the sealed nuclear battery — a welded unit at your town, your campus, your co-op, delivered fueled, swapped by the vendor like a propane exchange, never opened on site. Sovereign siting. Never sovereign fuel. The fission loop is a custody chain by international design, and the enrichment plant at its head is the most tightly guarded industrial process humans run.

Now the inversion. Fusion's fuel is deuterium sieved from seawater, lithium, boron — unregulated commodities with no treaty, no safeguards regime, no custody chain, because none of it is bomb feedstock and none of it can melt down. And in 2023 the NRC made one of the quietest consequential decisions in its history: it chose to regulate fusion under the byproduct-materials framework — the tier of hospital accelerators and industrial X-ray equipment — not under reactor law. That is why Helion's historic licenses came from a state health department and not a federal reactor tribunal. Fission arrived first and was chained at the fuel. Fusion walked in after the chains were forged, carrying fuel no chain was built for.

Say it plainly, because it is the hinge of the whole essay: fission is blocked by law, not physics; fusion is blocked by physics, not law. The door to a household flame is not welded shut. It stands at the far end of a corridor that no one has yet proven passable.

One company is in the corridor. Avalanche Energy, outside Seattle, is building the Orbitron — a fusion device small enough to sit on a desk, holding 300,000 volts for hours at a stretch, designed as modular units that gang from kilowatts to megawatts, with ambitions running from lunar surface power to microgrids. That is the second reactor, conceptually: stack a few in the garage, power the property. And the honest ledger, in the same breath: the Orbitron descends from the fusor's electrostatic lineage, and the canonical physics objection to that entire family — laid down in a famous 1995 MIT thesis by Todd Rider — is that beam-driven fusion leaks energy through radiation and collisions faster than fusion can pay it back. Sixty years, and nothing in the lineage has escaped by less than orders of magnitude. Avalanche is betting its orbital geometry slips the trap. No net energy has been demonstrated. And if it works, the first customers will be satellites and soldiers, not households. The corridor is real. So is the darkness in it.

If the stars align

The stars are the only fusion reactors that have ever run at a profit, so alignment is precisely the right word for what the second reactor requires. Name the stars soberly — five of them, in order:

If all five rise, the picture — labeled honestly as speculation — looks like this: a welded cabinet in the garage beside the wall that remembers. The pack that burns seawater. In principle, the deuterium in a few gallons of ocean, fully fused, holds a household's year; refueling is a cartridge swap on the scale of a propane exchange, once in a long while. The blueprint from the last essay gains one box and loses its December problem — no more leaning on the grid when a Willamette Valley winter blots out the roof for a week.

But mark the odds like an adult. This is a lottery ticket, not a plan. Every serious fusion company on Earth is building for the grid, because the grid is the only place current physics permits fusion to live — and even there, the first watt remains unsold. The household flame requires a demonstration nobody has made. It may never come. That is not a reason to look away from the door. It is a reason to number the stars honestly, and to notice which one is within reach of a citizen with a stamp.

Write the law before the machine

Because one star is. Two clauses of standing work are available today, to anyone, and they cost nothing but a filing:

Where to file Both clauses route exactly where the six in the parent essay do — the county commission, the public utility commission, the statehouse, the company itself. The routing table, in ascending order of difficulty, is at the end of Bring the Power Home. One office, one clause, one deadline, and a demand for a response on the record.

And if you doubt that filings move walls, the proof is sitting in this year's Federal Register: an agency that once took decades to bless a reactor is proposing to permit small ones in months, because the public's understanding of the safety case finally shifted and the accident wall — the one made of opinion — began to dissolve. Petitions work on the walls made of opinion. The craft is telling which walls those are. The adversary wall and the custody wall around fission will not move for any pamphlet, and should not. The walls around a seawater flame are thinner than anyone has bothered to notice.

Meanwhile, the first reactor

The stars may align. Probably slowly; possibly never. But sovereignty was never waiting on the miracle.

The first reactor is already commissioned — licensed by no agency, insured by no one because it needs no insuring, delivering on schedule for four and a half billion years, its fuel custody handled ninety-three million miles beyond subpoena. The loop runs today on glass and chemistry: the roof, the wall, the shelf, the cabinet, the doorkeeper. Build that now, exactly as the last essay laid it out.

The build The engineering half of the shelf — hardware tiers, model serving, the records vault with retrieval, a bounded agent loop, and the power math to keep it running — lives in the companion field guide at DevForge Academy: The Half-Closet Server.

Then do one small, absurd, hopeful thing: wire a spare breaker in the garage panel, and label it in pencil — second reactor.

If the corridor ever lights, the socket is waiting. And if it never does, you have lost nothing, because you already own the fire. You simply own it at a distance, the way a household was perhaps always meant to: arriving fresh with every sunrise, eight minutes off the vine — once as kilowatts, once as liberty.