The question
Why did the United States start a war with Iran on Saturday 28 February 2026?
The answer the evidence supports
Not the official one. The official story, that Iran was racing to rebuild a nuclear bomb and diplomacy had failed, is the one explanation the evidence kills outright (odds effectively zero). The administration’s own intelligence chief put in writing, 19 days after the strike, that Iran had made no attempt to rebuild its enrichment capability, and tried to leave that sentence out of her spoken testimony. The execute order went out 36 minutes before the mediator announced a deal was within reach. The IAEA’s own report, circulating that same day, said the negotiations were ongoing. Technical talks were booked for the following Monday. Iran had spent February offering concessions, not making threats.
What actually happened, on the numbers
A dead heat between two stories, with two more close behind.
Spur of the moment. Trump was handed a once-only chance to kill Khamenei (Netanyahu’s tip that the leadership would be in one place on Saturday morning), his envoys told him Geneva was going nowhere, and he gave the order on instinct. The strongest support: the US struck two weeks before its own forces were fully in place, with no economic plan, no legal letter, no funding request, no allied notifications. Everything about the follow-through looks like a decision made days, not months, in advance.
Kicking them while they’re down. Iran’s currency had collapsed, millions were in the streets, the regime had just shot thousands of its own people. Washington and Jerusalem saw a wounded regime and swung at it, with the nuclear story as cover. Support: Trump’s day-one call for Iranians to “take over your government”, reporting that Israel’s classified objective was regime change, and the fact this administration had run the identical play in Venezuela eight weeks earlier.
America jumped the queue. Israel was going to strike regardless, its plan existed, originally scheduled for mid-2026, so the US went first to control the result. This is, remarkably, the Secretary of State’s own on-record version.
The fix was in. The decision was really taken back in December at Mar-a-Lago, and February was theatre. Held up mainly by one unverified leak (the US asked Britain for bomber bases 17 days before the strike). If that leak is ever confirmed, this number jumps.
What died
The official story (0%). The machine — bureaucratic momentum and the arms-money engine together — 2.7%. And wag-the-dog (2.1%: Netanyahu’s motive is real, but on the US side the political returns went negative fast and Trump started blaming his own advisers).
On the arms industry, because it is the objection everyone raises
The industry profited enormously and positioned early. That is not in dispute, and it is also not evidence — it is true under every explanation on this list, which is the exact trap this method exists to avoid.
The first version of this report argued against the arms-money story on the grounds that the $87.6bn funding request arrived 116 days late. That argument was wrong, and it has been corrected. Money always comes after the shooting: seven and a half months after the June 2025 strikes on the same target, roughly two years after the Ukraine drawdowns began, months after the Red Sea firings. Expend first, fund the replacement second, is simply how the system runs.
What killed the arms-money story was looking for the pressure directly and not finding it. There is no documented industry or congressional advocacy for this strike in the three months before it; the lobbying surge is post-strike and reactive. When the Pentagon asked for $200bn on 19 March the resistance came from defence-side members of Congress, and the ask was cut to $87.6bn — of which about $7.6bn is traceable to replacing what Iran cost. By August the government was pressing contractors to build faster, not the other way round. And the campaign burned the newest and scarcest munitions in the inventory while demonstrably older stock sat untouched in the magazines, which is the opposite of using up what is sitting in storage.
What survives is narrower. Defence money funds the think tanks whose people spent the fortnight before the strike arguing publicly for it. That is a real channel — but it belongs to the story about the option set being narrowed before it ever reached the President, not to a story about the budget.
Roughly a 55% combined chance the decision formed in February itself, an opportunistic strike at a wobbling regime through a suddenly-open window, versus roughly 40% that the timing was substantially set by Israel or by a prior US–Israeli understanding. It was not intelligence-driven, and the evidence for that is unusually strong because it comes from the administration’s own officials speaking against their own interest.
What to watch
Confirmation of the 11 Feb UK basing request (raises “the fix was in”). Any document showing the decision predates 23 Feb (same). Evidence the Khamenei tip was manufactured to force Trump’s hand (raises “jumped the queue”). The drafting history of the $87.6bn funding request, the one thing that could resurrect the arms-money theory.