Abundance or Extinction: The Fusion Imperative

Somewhere in the last decade, a switch got thrown that nobody has officially acknowledged. Global mean temperature has now spent enough consecutive months north of the Paris Agreement's 1.5°C guardrail that treating it as a temporary excursion requires a fairly heroic act of denial. We are not approaching the tipping point. We are, by most sensible readings of the data, standing on the far side of it, looking back at where the point used to be, wondering why nobody rang a bell.

This summer gave us the receipts. Across three continents, grid operators issued emergency load-shedding notices not because of some exotic infrastructure failure, but because of air conditioning. Ordinary, boring, life-saving air conditioning, switched on by millions of households simultaneously during heat domes that would, a generation ago, have been considered freak events and are now simply "Tuesday." Substations in Texas tripped. Athens flirted with rolling brownouts during a heatwave that pushed the city past 44°C. Delhi's grid operator begged citizens to set thermostats no lower than 26°C, an act of consumer-facing triage that used to be unthinkable in a modern economy.

 

The Thermostat Death Spiral

Here is the mechanism, and it is worth stating plainly because it is genuinely vicious in its circularity. Rising global temperatures increase demand for cooling. Increased cooling demand strains grids that were sized for a climate that no longer exists. Strained grids either brown out (misery, and in vulnerable populations, death) or get backfilled with whatever generation capacity is fastest to spin up, which in most of the world in 2026 is still gas, sometimes coal. That generation adds more carbon to the atmosphere. More carbon means more warming. More warming means more air conditioning. Repeat, forever, or at least until something breaks.

This is not a hypothetical feedback loop drawn on a whiteboard by a systems dynamics consultant. It is measurable, it is accelerating, and it has a name problem: everyone talks about it as an "energy demand" issue, as though the fix were better insulation and a nudge unit reminding people to wear a t-shirt indoors. It isn't. It is a supply-side problem wearing a demand-side costume. The honest framing is this: humanity now requires a baseline amount of cooling, heating, desalination, and food-preservation energy simply to keep several billion people alive through the climate we have already locked in, regardless of what we do next on emissions. Demand reduction is a rounding error against that baseline. Supply is the whole game.

 

Scarcity Pricing Was Never a Feature

There is a comfortable liberal fiction, popular at conferences with good coffee, that the answer to energy-driven climate collapse is efficiency: use less, waste less, price carbon properly, let market signals gently herd behaviour toward virtue. It is not wrong so much as catastrophically insufficient. Efficiency measures assume that the underlying resource, electricity, remains expensive enough that saving it matters more than having enough of it. That assumption works fine for a wealthy suburb optimising its smart thermostat. It fails completely for the roughly 750 million people who currently have no reliable electricity at all, and for the further billions who have it but cannot afford enough of it to run a refrigerator and an air conditioner in the same house during a 45°C week.

Scarcity pricing did useful work when energy was genuinely scarce and the marginal unit came from burning something finite. But scarcity pricing applied to a survival necessity, in a climate where the necessity is rising faster than income, does not ration demand efficiently. It rations survival unevenly. The rich retrofit, insulate, and run the aircon regardless of the bill. The poor either don't run it, or run it and then can't eat, or migrate somewhere that doesn't require it, which increasingly means somewhere that also requires it. There is no version of "just price it correctly and let the market sort it out" that ends anywhere except mass displacement, mass mortality, or both, distributed almost entirely along lines of existing inequality.

 

 

Why Fusion, Specifically, and Not Just "More Renewables"

To be clear: solar and wind are not the enemy here, and anyone building the next decade's grid without them is planning badly. But wind and solar share a structural limitation that no amount of subsidy fixes, which is that they are intermittent and diffuse. They need storage, transmission, and land at a scale that runs headlong into the same NIMBYism, permitting delay, and material-supply-chain bottlenecks that slowed nuclear fission for decades. They are necessary. They are not, on their own, sufficient to deliver the kind of always-on, energy-so-cheap-it's-basically-free-except-for-wires abundance that actually breaks the thermostat death spiral rather than just slowing it down.

Fusion offers something categorically different: a fuel source (deuterium, extractable from seawater in effectively limitless quantities, and lithium-derived tritium) with an energy density that makes coal look like kindling, no meltdown risk because the reaction stops the instant containment is lost rather than running away, no long-lived high-activity waste of the kind that makes fission a political football, and no carbon emissions at the point of generation. A single fusion plant, once mature, could in principle deliver gigawatts continuously from a footprint smaller than a fission plant, sited more or less anywhere there's water and grid access, running regardless of cloud cover, wind, or time of day.

That "once mature" is doing a lot of work in that sentence, and it deserves honesty rather than hype.

 

Where We Actually Are, Not Where the Press Releases Say We Are

In December 2022, the National Ignition Facility at Lawrence Livermore achieved what the field calls "scientific breakeven": more fusion energy released from the target than the laser energy delivered to it. It was a genuine milestone, and it was also, inconveniently, not the milestone most headlines implied. The lasers themselves draw roughly a hundred times more electricity from the wall than the shot delivers to the fuel pellet, so NIF is nowhere near producing net electricity for a grid. It proved the physics works. It did not prove the engineering does.

ITER, the thirty-five-nation tokamak under construction in southern France, is the slower, more institutionally cumbersome bet: aiming for a Q of 10 (ten units of fusion power out for every unit of heating power in) sometime in the 2030s, with the specific goal of proving sustained, controlled fusion at industrially relevant scale rather than a single laser pulse. It has been delayed, over budget, and is a genuinely painful case study in what happens when you try to run cutting-edge physics through a multinational treaty organisation. It is also still, as of 2026, the most credible large-scale proof point the field has.

The more interesting movement, arguably, has been on the private side. Commonwealth Fusion Systems, spun out of MIT and backed by a fairly serious pile of venture capital, is betting on high-temperature superconducting magnets to shrink the tokamak enough to make it commercially buildable this decade rather than next; its SPARC device is targeting net energy gain in the near term, with a commercial follow-on (ARC) intended to actually put power on a grid. TAE Technologies is pursuing a completely different confinement geometry aimed at aneutronic fuel cycles that sidestep much of fission's waste-and-shielding baggage entirely. Helion, backed in part by a certain famous OpenAI chief executive with a taste for improbable bets, claims a path to net electricity generation via direct energy conversion rather than the traditional steam-turbine route, and has a power purchase agreement with Microsoft that will either look prescient or embarrassing depending on how the physics behaves under deadline pressure.

None of this means fusion power is arriving next year, or even next decade, for most of the world. It means the field has moved, within the space of roughly five years, from "permanently thirty years away" to "specific companies with specific hardware making specific, falsifiable, near-term claims." That is a genuinely different epistemic situation, and it is one that deserves attention proportional to the stakes rather than the usual reflexive cynicism the phrase "fusion power" tends to attract.

 

The Transmission-Cost Endgame

Here is the version of the future worth actually building toward, and it is more specific than "cheap clean energy for all," which is the kind of sentence that looks good on a conference slide and means nothing operationally. The target is a world in which the marginal cost of generating another unit of electricity is so close to zero, because the fuel is seawater and the plant runs continuously without carbon input, that the price a household or a factory or a hospital pays is essentially just the cost of the wires, substations, and grid maintenance required to move the electron from the plant to the socket. Generation stops being the expensive part. Distribution becomes the only real cost centre, and distribution is a solved, boring, capital-amortisable infrastructure problem, the sort of thing municipal bond markets have financed since the nineteenth century.

That is the actual industrial leap the original brief for this post was pointing at, and it is worth being precise about why it matters more than "renewable energy targets" as usually framed. Once the fuel cost of electricity approaches zero, desalination stops being expensive, which means fresh water stops being the geopolitical flashpoint it is rapidly becoming across the Middle East, the Sahel, and the American Southwest. Vertical and indoor agriculture, currently energy-cost-prohibitive at scale, becomes viable in latitudes where traditional farming is failing under heat stress, addressing food security directly rather than through the current strategy of hoping crop yields hold. Direct air carbon capture, currently a rounding-error technology because it is so energy-intensive it barely pays for its own emissions, becomes a genuine mitigation tool rather than a green-washing exercise, because the energy to run it costs almost nothing. And air conditioning, the thing that started this whole essay, stops being a subsistence-level household budget line item that determines whether a family in Phoenix or Lahore survives August, and becomes what it always should have been: a utility, not a luxury, not a rationed good, not a class marker.

 

The Alternative Is Not Hypothetical

It is tempting, and I understand the temptation, to treat "or half the population has to go" as rhetorical excess, the kind of line a blog post reaches for to sound dramatic. I'd genuinely prefer it were rhetorical. It isn't. Every serious climate-mortality model, from The Lancet's heat-mortality tracking to the WHO's projections on climate-attributable deaths, already shows heat stress alone becoming one of the leading causes of excess mortality in the tropics and subtropics within this century, before you even add crop failure, water conflict, and mass displacement to the ledger. The mechanism by which "half the population" becomes something other than hyperbole is not a single dramatic event. It is exactly the grinding, unglamorous process already underway: heatwaves that used to be once-a-decade events becoming annual; grids that brown out precisely when cooling is most needed; agriculture that fails in exactly the regions least able to afford imported food; and a political economy that treats all of it as somebody else's emergency until it is on their own doorstep.

The counter to that trajectory was never going to be persuading eight billion people to want less. It was always going to be building the thing that makes wanting less unnecessary: a source of energy abundant enough, and eventually cheap enough, that the fight over who gets to stay cool, fed, and hydrated stops being a fight at all. Fusion is not guaranteed to arrive in time. But it is, on present evidence, the only candidate on the table that scales to the actual size of the problem rather than merely to the size of the press release. Everything else, however necessary, is a bridge. Fusion is the only proposal for what's on the other side of the river.

All Posts