There is a mood I keep meeting this summer, in meetings and at dinner tables — a kind of arithmetic fatigue: people have done the sums, they have concluded the sums are bad, and they have decided that what follows from bad sums is nothing. The forests are burning on three continents. The heat arrives earlier each year and leaves later each time. The strait is still closed…
Against that, a solar array on the roof of a factory on the Swiss plateau has all the makings of a futile gesture.
This is about defending a policy of small steps, and doing so with complete clarity. Optimism is not what I have to offer.
The analogy, and where it breaks down
It is never too late to stop smoking. Not at forty, not at sixty, not on a hospital ward with the diagnosis already written down. Within twenty-four hours of the last cigarette, the carbon monoxide clears from the blood and the oxygen-carrying capacity begins to recover. Within a few weeks, the cilia in the airways start working again. None of this erases what came before: a person who stops at sixty does not get back the lung tissue they lost. But the curve bends — it bends immediately, and it bends whether or not the patient believes it will.
Let me be precise about where the analogy stops working: it gives way at an obvious point, one I would rather name myself.
When he stops, the smoker gets instant feedback. The relief comes fast enough to be felt, and that is precisely what makes the metaphor consoling. The climate system does not work that way. It carries decades of thermal inertia stored in the oceans; emissions stopped tomorrow would still leave warming already locked in for a generation. Our patient, by contrast, keeps deteriorating long after the cigarette is out. Anyone who tells you otherwise is selling something. So the argument is not that acting feels good, or that it would be quickly rewarded.
The argument is that the trajectory is the one thing we have ever managed to change, and that it remains changeable at every point along the curve — including this one. The lag is a reason to start earlier. It has never been a reason not to start.
What acting actually looks like
I spent the last ten days doing something that makes neither a great post nor a good photo.
Qanta is in the process of acquiring its first portfolio: ten projects in French-speaking Switzerland — rooftop photovoltaic, heat contracting, storage — held through a dedicated vehicle. This is our tenth “AssetCo,” but for the first time in Switzerland the funds are being raised in the form of a securitised product, in partnership with a leading private bank. Subject to the usual conditions, we will say more in the coming weeks.
Here is what those ten days consisted of. Land-register extracts. Easement deeds, one of which was rejected by the registry in 2024 because the right had been classified as a right of use rather than a building right, and which, as of this week, I still cannot prove was corrected and then registered. Bank agreements. Insurance notifications under a framework contract that — and this is the kind of detail that sinks a deal — covers nothing automatically: each site has to be individually declared. A tripartite contract in German whose article 10.4 states there is no metering tolerance threshold, and whose article 14.4 sets that threshold at 5%. Both articles appear in the same document. Both are in force.
None of this is climate work in the usual sense of the phrase. And yet all of it is the difference between a project that exists on paper and a project that produces electrons and heat. The gap between good intentions and installed capacity is not made of vision. It is made of paperwork, and paperwork is where projects die quietly.
The contradiction
I did this work with two artificial intelligence systems, and I can see exactly what that looks like.
The objection is a fair one and I don’t intend to dodge it: data centres consume considerable amounts of electricity, that consumption is growing fast, and a large share of it is currently met by carbon-based sources. Anyone building energy infrastructure without seeing this isn’t looking closely.
So let me put my own usage on the scale, stating up front what the number is worth.
Ten days of intensive use — long documents, extended context, a lot of generation — plausibly consumed somewhere between twenty and fifty kilowatt-hours, including data-centre losses. The estimate is shaky: neither provider publishes per-query consumption, and public estimates also vary by an order of magnitude.
It doesn’t matter. The smallest of the projects is a 225 kWp array on an industrial roof. Under Swiss conditions, it will produce around 214,000 kilowatt-hours a year. Fifty kilowatt-hours is about two hours of that single installation’s average daily output. Multiply my estimate by ten: you’re still under one day of a single roof, out of ten projects, over a twenty-five-year asset life.
The ratio is so lopsided that the imprecision becomes beside the point. That’s the honest version of the argument, and I believe it’s the only one worth making.
Still, it’s worth seeing what this arithmetic actually proves, because it proves less than it seems to. It answers the question of whether my ten days were worth their cost. It does not answer the aggregate question: whether the global build-out of computing capacity is being matched by the necessary generation, and on what timescale. That’s a real question, and it belongs far more to grid planning and electricity procurement than to individual usage — my own personal tally has nothing useful to say about it. I chose a flattering denominator. I’d rather admit that than pretend I hadn’t noticed.
What I believe
The instinct behind climate despair isn’t foolish. It’s an accurate reading of the physics, paired with a false conclusion about our capacity to act. The physics says the damage is real and partly irreversible. It doesn’t say the rest is already decided.
I spent ten days on easements, insurance annexes, and a contradiction between two articles of a contract in German. I’d be hard-pressed to say what those ten projects will change on the temperature record. Next to nothing, is the honest answer. But they will exist, they will produce, and they will still be there in 2050.
The lungs start recovering within a few days.
Not because it was early.
Because we stopped.