Vol. IV, No. 85
Covering 9 March - 22 March 2026
Monday, March 23, 2026
Late edition · A Relentless publication
All the fortnight that mattered, in technology and in the world, read next to what we were building at the time.
THE WAR

The promised quick blow becomes a grinding war, and the "days" stretch into weeks

The Defense Secretary had promised uncontested air supremacy within days. Weeks in, Iranian missiles still cross the Gulf, Israeli troops are fighting in southern Lebanon, the dead climb into the thousands, and the war that was sold as swift and surgical has become the thing such wars always become: long, wide, and beyond anyone's control.

The war entered the phase its architects had promised it would skip. The American Defense Secretary had said, in the first days, that the United States would achieve uncontested air supremacy within days; two weeks on, Iranian ballistic missiles still reached Gulf targets, Israeli ground forces were fighting through southern Lebanon against Hezbollah, and the campaign that was sold as swift and surgical had settled into the grinding, widening shape that such campaigns reliably take once the first plan meets the enemy's refusal to collapse on schedule. The Iranian dead passed into the thousands, the leadership decapitated at the top had been inherited by Mojtaba Khamenei, and the state that was to be broken in a night was instead absorbing the blow and answering it, day after day, across a region with no clean front and no visible end.

The war's expansion followed the logic every wider war follows, each side striking what it could reach. Iranian missiles and drones continued to hit bases and facilities across Qatar, Kuwait, Bahrain, and the UAE; Israeli strikes and ground operations pushed into Lebanon; the oil infrastructure of the Gulf remained under threat and under fire. What had been promised as a matter of days was now measured in weeks with no terminus, and the gap between the promise of the swift war and the reality of the long one was, as it always is, filling with the dead.

THE ECONOMY

The oil shock hardens into a recession warning

Crude held above a hundred and ten dollars as the Strait of Hormuz stayed effectively closed to the shipping that feared it, and the war's economic shock began to set into something structural. Recession models that had sat quiet for a year jumped toward the threshold that has historically preceded a downturn; the supply shock of doubled oil, arriving atop existing tariffs, presented every central bank with the one problem it cannot solve, an inflation it can only worsen by fighting or tolerate by ignoring. And the great bet of the age, the artificial-intelligence build-out on which so much of the economy's growth had come to lean, faced its first genuine external test, as the cost of the energy that is its true foundation doubled overnight and the question this paper has pressed for three years, whether the digital future could survive a collision with the physical one, stopped being rhetorical.

IN BRIEF

Europe arms, the streets persist, Eid under fire

The war's nuclear logic propagated through a frightened Europe: France moved to increase its nuclear stockpile for the first time in decades and reached a deterrence arrangement with Denmark, as a continent that had spent eighty years sheltering under a borrowed umbrella began, in fear, to build its own. American antiwar demonstrations persisted in the cities even as the political system declined to constrain the war. And across the Muslim world, Eid al-Fitr arrived at the end of Ramadan under a sky of drones and missiles, a festival of mercy observed in the middle of a war, its ordinary joys pressed against an extraordinary grief.

The Column

The war power that Congress gave away

The Senate declined to limit a war it never voted for. What concerns me is the constitutional machinery that was supposed to make exactly this hard, and why its quiet failure is a more dangerous story than the war it failed to check.

Amid a war launched without a declaration and now waged on the president's authority alone, the Senate this fortnight declined even to constrain it, rejecting the resolution that would have limited the executive's power to continue. The war is the loud story. The quiet one, and the one that will outlast this particular conflict, is the constitutional machinery that failed here in plain sight: the deliberate, centuries-old design that placed the power to make war not with the president but with the Congress, and that has now been hollowed to a formality that the legislature will not even invoke. That failure is the more dangerous story, because a war ends and this does not.

The design was not an accident or an oversight, and it was not naive about speed. The framers of the American constitution, having studied every republic that preceded them, placed the war power in the legislature for a specific and hard-headed reason, which James Madison stated with unusual bluntness: that the executive is, of all the branches, the one most structurally prone to war, because war concentrates power in the executive, flatters it, and frees it from the ordinary constraints of peacetime, so that the branch which conducts a war is exactly the branch that can least be trusted to decide whether to start one. The remedy was to divide the decision from the execution, to require that the many, in the legislature, authorise what the one, in the executive, would carry out. It was a machine built to make war hard, deliberately, because its designers understood that a republic's chief danger was not that it would find war too difficult but that it would find it too easy.

Let me put the counterargument at full strength, because it is the argument that has won, in practice, for eighty years, and it is not frivolous. The world moves faster than an eighteenth-century legislature; modern threats arrive in minutes, not months; a president who had to assemble a congressional majority before responding to a missile would be a president presiding over a crater, and the framers' machine, whatever its wisdom, was built for a world of sailing ships and cannot govern a world of hypersonics. There is real force here. Genuine emergencies do demand that one person be able to act. But notice what this argument, valid for the true emergency, has been stretched to cover: not a response to an attack in the minutes it takes to launch one, but a deliberate, months-in-the-planning, region-reshaping war of choice against a nation of ninety million, precisely the kind of war the machine was built to make the legislature own. The emergency exception has swallowed the rule, and a design meant to reserve to the many the gravest choice a state can make has been reduced to a power the one exercises alone, with the many declining even to object.

And here is why the abdication matters beyond this war, why it is worse than any single conflict it fails to stop. The framers' insight was that the check was not a brake on necessary wars but a filter against unnecessary ones, that requiring the many to authorise war would not prevent the wars that had to be fought but would prevent some fraction of the wars that merely could be, the wars of choice and opportunity and executive temptation that a lone decider, flattered by the power that war confers, will always find reasons to wage. Remove the filter, and you do not get faster necessary wars; you get more unnecessary ones, because the one institution designed to make the executive pause before the gravest act has stopped performing its only essential function. The Senate's refusal this fortnight was not a vote about Iran. It was the legislature confirming, in the middle of exactly the kind of war the machine was built to check, that the machine no longer runs, and a republic that has let the war power migrate wholesale to the executive has not made itself safer against its enemies. It has made itself defenceless against its own worst impulses, which is the danger the whole design was built to guard, and which it now no longer guards at all.

Field Notes
A Relentless build, told plainly

The war just moved my whole problem

I wrote in February about the tool I am building to decide where, on a strained grid, the great new data centres can go. This fortnight the war changed the ground under that work, because the one variable I model above all others, the price and availability of power, just moved more in three weeks than it had in three years. No client is named.

I have spent this year building software to solve a problem I described in these pages in February: given that you want to build a data centre that draws as much power as a small city, where, on an already-strained electrical grid, can you actually put it? The whole tool turns on modelling the constraints that make a site viable, and the constraint that dominates all the others is power, its cost, its availability, the capacity of the grid to deliver it. And this fortnight the war reached into the middle of that model and moved the largest variable in it further than three years of ordinary change had done, because a doubling of the oil price ripples through the entire energy system, into the cost of the gas that sets the marginal price of electricity, into the economics of every site my software evaluates.

What this has taught me, watching the numbers move, is something about the difference between a model and the world it models, and it is a lesson worth more than the immediate disruption. When I built the tool, energy cost was a variable, but it was a slow variable, the kind you treat as roughly stable while you optimise around it, and a great deal of the tool's cleverness lived in trading that stable cost off against the other constraints. A war teaches you, brutally, which of your stable assumptions were only stable because nothing had yet disturbed them. The good models, the ones worth building, are the ones that hold up when a variable you had quietly treated as fixed suddenly triples, and the test of my tool this fortnight was not whether it produced a clever answer under normal conditions but whether it produced a sane one under a shock, whether it degraded gracefully or broke.

The lesson generalises past data centres, and it is the one I keep relearning in this work: build for the shock, not for the demo. It is easy, and it is fatal, to build a system that is elegant under the conditions you expected and brittle under the ones you did not, because the conditions you did not expect are precisely the ones that arrive, unannounced, in a fortnight like this one, and find out whether you built for the world or only for the weather you happened to have while you were building. The war did not break my tool, because I had, from earlier and smaller shocks, learned to treat the stable variables as merely quiet rather than fixed. But it moved the ground hard enough to remind me that everything I build sits on assumptions that hold only until they do not, and that the whole craft, in the end, is knowing which of your certainties are load-bearing and building so that the failure of any one of them bends the system instead of shattering it.

The Ledger
AI
The oil shock's first fortnight of real economic damage sharpens every question about the AI build-out's foundations; a build-out whose true input is cheap, abundant power now watches that input double in a war it cannot influence. The bet's physical basis, long abstracted, is suddenly the whole story.
Data centers & power
As the Field Notes above describe, the war moved the single largest variable in the siting problem further in three weeks than three years of peace had. Energy is no longer the quiet substrate; it is the front line, and the build-out's economics move with the fighting.
Rates
A supply-shock inflation the central banks cannot fix, atop tariffs, into a slowing economy: the recession models cross toward their historical threshold, and the Fed's tools are, for a war-driven oil spike, worse than useless.
Real estate
US 30-year mortgage near 6.6 percent and rising on the inflation shock; the domestic market's long strange calm gives way to the war's economic weather.
India tech
India absorbs the war's energy shock directly, its Gulf-dependent oil and its millions of regional workers both exposed; the frugal, constraint-first posture this paper has long noted in Indian engineering looks, in a fortnight like this, less like a limitation than a preparation.
What we called wrong
Nothing to retract. The "quick, surgical" war has become the long, wide one this paper warned it would, on the timetable of every such war before it, and the constitutional check that might have forced the question was, as this column notes, not even invoked.
The Back Page

Eid in the shelter

The festival of Eid al-Fitr came this fortnight at the end of Ramadan, the celebration of mercy and renewal that closes the month of fasting, and it came to much of the Muslim world under a sky of drones. There is an old human stubbornness in the observance of a feast during a war, in the insistence on the sweets and the prayers and the children's new clothes while the missiles cross overhead, and it is easy to read it as either heartbreaking or futile, the small rituals of ordinary life dwarfed by the enormity around them. But there is another way to see it, and this paper, which has spent a hard fortnight cataloguing the machinery of destruction, chooses it. The feast kept in the middle of the war is not a denial of the war; it is a refusal to let the war be the only thing, an assertion, against the missiles, that the people beneath them are still people, still keeping their days and their faith and their love for one another, still human in exactly the way that the abstractions of war are designed to make us forget. The bombs fell on Eid. And the families, where they could, still gathered, still prayed, still pressed the ordinary sweetness of the festival against the extraordinary bitterness of the hour, and in that pressing there is a kind of resistance that no army can defeat, because it is the simple, unkillable insistence of ordinary life on continuing to be lived, even here, even now, even under this sky.