What Really Happened to Fordow

Of all the questions the Twelve-Day War left behind, one carries more weight than the rest, and it is still not cleanly settled: what actually happened to Fordow. On the American and Israeli account, the war was fought to end Iran's path to a bomb, and Fordow was the target that mattered most: the deep, hardened enrichment plant that Israel's air force could not reach and that the United States built a one-of-a-kind weapon to destroy. Whether that weapon worked is not a footnote. It is the difference between a war that achieved its stated aim and one that merely delayed the problem while scattering the most dangerous part of it.

It is also the one question that cannot be answered by looking. Fordow's enrichment hall sits under a mountain near Qom, behind tens of meters of rock chosen, deliberately, to defeat exactly the kind of assessment the world tried to make in the days after 22 June. No commercial satellite images through it. No seismograph reports the condition of a centrifuge. The people best placed to know, Iran's own engineers, are not talking, and the people who spoke loudest, the governments that ordered the strike, had the most at stake in the answer being yes.

This is not a hard case because the facts are locked in a classified file. It is hard because the target was engineered to make the facts unknowable from outside, and because the moment those facts mattered most, the one organization built to establish them was shown the door. The contradictory intelligence assessments that followed are not competing attempts to deceive. They are competing ways of guessing at the same thing no one could see, and the gaps between them are a map of exactly where the evidence gives out.

This briefing does not tell you that Fordow was destroyed, and it does not tell you that it survived. It tells you what the primary data can actually establish, the seismic record, the satellite imagery, the physics of the weapon, and the IAEA's own count of Iran's uranium, and, just as carefully, where that evidence runs out. In a story this contested, the boundary between what is known and what is merely asserted is the most valuable thing a record can hold. Everything below is an attempt to hold it, and to be honest at the one place it matters most: the edge.

The mountain that made the question unanswerable

The most consequential claim of the entire Twelve-Day War was made in a single word. Hours after the B-2s pulled off target, President Trump told the nation that Iran's enrichment sites had been "completely and totally obliterated," with Fordow named as the primary target. Within days, a leaked Defense Intelligence Agency assessment put the setback at only a few months; the CIA countered that the program was "severely damaged" and would take years to rebuild; and the Pentagon estimate landed at up to two years. Four agencies, four answers, spanning from a few months to many years. That is not a story about political spin, though there was plenty. It is a story about a specific mountain near Qom, and about a hard epistemic fact that mountain was engineered to create: the central question of this war cannot be answered by direct observation. Everything that follows in this briefing is an exercise in inference from indirect signals, because that is the only kind of evidence the target permits.

What the evidence actually shows begins with why the target exists at all. Fordow, formally the Fordow Fuel Enrichment Plant, is a uranium-enrichment facility built roughly 80 to 90 meters beneath a mountain outside Qom, and its depth is not incidental. It was dug into rock precisely to survive an air attack, revealed to the world in 2009 as a covert site whose only military logic was survivability. Inside sit roughly 2,700 IR-1 and IR-6 centrifuges, the machines that made it the program's assessed engine for bulk 60 percent HEU production, the near-weapons-grade material that was the whole proliferation concern. When Israel opened the war on 13 June, its aircraft could reach Natanz and destroy its above-ground plant, but Fordow was reported not impacted in the first wave. It sat too deep for Israel's munitions. Only the United States held a weapon built for it: the GBU-57 Massive Ordnance Penetrator, a 30,000-pound bomb designed to bore through as much as 60 meters of concrete, and its first-ever combat use on 22 June was against this single mountain. The engineering symmetry is the point: a facility built to be unreachable, struck by the one weapon built to reach it, and even then the reach ended at the rock, not at the centrifuge hall.

Here is the epistemic wall. You cannot see inside a mountain from orbit, and you cannot see inside it from a seismograph. No commercial satellite, however sharp, images through 80 to 90 meters of rock; the centrifuge cascades sit in a chamber no camera outside Iran has ever photographed. The seismic record makes the limit even starker. A live query of the USGS earthquake catalog for the Fordow box returns zero cataloged events across the entire war, and the independent European-Mediterranean network confirms the same null. Twelve of the heaviest conventional bombs ever built, dropped in sequence, did not register on a single global seismograph, because even 30,000-pound bombs release orders of magnitude less energy than a small nuclear test. Iran's own local stations recorded only about magnitude 2 blast tremors near the site. And seismology, even where it registers a blast, can time and size an explosion but cannot tell you whether a centrifuge broke. A tremor is not a status report. This is why the widely shared "magnitude 4.3 bunker-buster" figure is unsupported: it appears in no primary catalog, and the real M2.5 many outlets pinned on the US strike was an Israeli strike days earlier, on 16 June. The absence of a large signal is itself the finding.

So how does anyone assess a hardened target they cannot look inside? By reading the surface and inferring the interior. Analysts do not observe the centrifuge hall; they observe the mountain's skin and the human behavior around it, and reason backward. They count the craters: Maxar imagery showed six penetration holes in two clusters of three along the ridge, matching the two ventilation shafts. General Dan Caine's official account described twelve bombs walked into those two shafts, the first to strip the concrete caps and the rest to enter and detonate below, and the visible surface pattern corroborated the aim point. They read the pre-strike activity: Maxar caught about 16 cargo trucks and bulldozers sealing tunnel entrances on 19 to 20 June, read as deliberate backfilling. They watch for subsidence, the ground sinking as a void collapses beneath it: months later, Orbital Sidekick's hyperspectral sensors detected subsidence over an area roughly 315 by 275 meters with concrete pushed to the surface. They even pick up heat: NASA's VIIRS sensor flagged a thermal anomaly near Fordow the night before the bombing. And they track the aftermath: months on, CSIS found virtually no rehabilitation at the site, an absence of activity that is itself evidence. Craters, shafts, subsidence, the presence or absence of repair crews: these are proxies, every one of them, for a hall no instrument can reach.

Why this matters is that it defines the honest limit of the whole record, and it is the frame this platform is built on. The IAEA, the one body with a mandate to know, was blunt about the boundary. Director General Grossi told the Security Council on 22 June that craters were visible but that no one, including the IAEA, could yet assess the underground damage at Fordow. Four days later the agency's position sharpened to a careful inference rather than an observation: Fordow's centrifuges were "likely no longer operational," while the degree of internal damage still could not be determined with certainty. That is the ceiling of what can be known from outside a mountain: a probabilistic read of surface evidence, not a count of shattered machines. It is why "obliterated" and "set back a few months" could coexist for weeks without either being provably wrong, and why the loss of IAEA continuity of knowledge, with over 400 kilograms of 60 percent uranium falling outside monitoring, compounds the blindness rather than resolving it. The rest of this briefing does not pretend to see through the rock. It does the only rigorous thing available: it lays out each indirect signal, seismic, satellite, thermal, and diplomatic, states honestly what each one can and cannot tell you, and lets the reader watch inference do the work that observation cannot.

How the strike was aimed

The single hardest thing to prove about the June 2025 strike on Fordow is what happened eight or nine stories underground, inside a hall no camera can reach. But there is a prior question the evidence answers far more cleanly, and it is worth separating from the damage debate: how was the weapon actually delivered? On this, the public record is unusually good. The United States briefed a specific mechanical account of the attack, and commercial satellites photographed the ridge above the mountain two days later. The two independently corroborate each other. That corroboration is the reason the AIM of Operation Midnight Hammer can be stated with confidence even while its RESULT stays disputed.

Start with the weapon. Fordow was built to defeat exactly the kind of air campaign it faced: the enrichment hall sits roughly 80 to 90 metres under a mountain near Qom, packed with about 2,700 IR-1 and IR-6 centrifuges, deep enough that no conventional Israeli munition could reach it. The only tool in any arsenal designed for that problem is the GBU-57A/B Massive Ordnance Penetrator, a roughly 30,000-pound (about 14-tonne) bomb rated to punch through up to about 60 metres of concrete before its delayed fuze fires. Only one aircraft can carry it, the B-2 Spirit, two per plane. On the night of 21 to 22 June, seven B-2s of the 509th Bomb Wing flew a 37-hour round-trip mission from Whiteman Air Force Base in Missouri, refueling in the air, while a decoy flight of B-2s tracked west toward Guam and drew the open-source watchers. That much is confirmed by the Pentagon's own briefing and by the Air & Space Forces Magazine reconstruction: the first two MOPs fell on Fordow at about 2:10 a.m. Iran time, and the operation as a whole put 14 MOPs on target, with the widely reported split being 12 on Fordow and 2 on Natanz (a split the Pentagon never officially confirmed, and one worth flagging as an estimate rather than a fact).

The mechanical account came a few days later, from Gen. Dan Caine, Chairman of the Joint Chiefs. His briefing, detailed by DefenseScoop, is the most granular public description of how the bombs were used, and it is not the caricature of a mountain being flattened. Caine described the 12 bombs being "walked" into TWO ventilation shafts that ran down into the underground complex. The sequence he laid out was deliberate: the first weapon removed the concrete cap sealing a shaft, then weapons two through five entered the open shaft at over 1,000 feet per second and detonated deep inside, with a sixth bomb held in reserve as a backup in case any of the earlier ones failed. Run that pattern across both targeted shafts and the arithmetic lands on twelve. This is a precision-threading account, not a saturation-bombing one: the plan was to use the facility's own ventilation architecture as a delivery channel to place explosive energy inside the buried hall rather than trying to crush the whole mountain from above.

What makes that account credible is that it can be checked against imagery collected by parties with no stake in the Pentagon's framing. Maxar Technologies captured Fordow on 22 June, and analysis reported by Bloomberg and the Institute for Science and International Security (ISIS) found six penetration holes along the ridge, arranged in two clusters of three, directly over the underground hall. Six holes, two clusters, two shafts: the surface pattern matches the briefed method almost exactly. Caine's account has each shaft receiving multiple weapons in sequence, and the imagery shows multiple closely spaced impacts at each of two points rather than a random scatter across the site. When the IAEA's Rafael Grossi told the UN Security Council on the same day that craters visible at Fordow indicated the use of ground-penetrating munitions, he was describing, from a neutral vantage, the same holes. Three independent readings of the ridge (the US briefer, a commercial imagery firm and its analysts, and the UN watchdog) converge on the same physical fact: bunker-busters went into two specific points, in the pattern the aiming plan called for.

The imagery also captured the days BEFORE the bombs fell, and this is where the picture gets more textured. Maxar frames from 19 to 20 June, analyzed by The War Zone, showed about 16 cargo trucks lined up at Fordow's tunnel entrance along with bulldozers working to seal the entrances. Analysts read this two ways at once, and both may be true: as pre-strike backfilling, Iran packing the tunnel mouths with earth to blunt or misdirect the coming blast, and as possible removal of material before the attack. The trucks are a real, dated observation from orbit. What was ON those trucks is not something a satellite can see, and it feeds directly into the separate, unresolved question of where Iran's roughly 440 kg of 60% enriched uranium ended up, a stockpile the IAEA verified at 440.9 kg on 13 June and then lost continuity on. Note the honest limit here: the trucks corroborate that Iran was ACTIVE at the site beforehand; they do not, by themselves, prove either a successful evacuation or a failed one.

Why this matters is a point about what satellites can and cannot referee. The imagery settles the aim decisively. We know the weapon (the GBU-57), the delivery method (bombs walked into two ventilation shafts), and that the surface result (six holes in two clusters of three) matches the briefed plan, cross-checked by ISIS's own post-attack assessment. That is a genuine, sourced finding, and it is more than most reporting conveyed in the noise of "obliterated" versus "only set back a few months." But the aim being well-corroborated is precisely what exposes the boundary of orbital evidence. As Grossi told the Security Council, no one, the IAEA included, could yet assess the underground damage at Fordow from the surface. Six clean entry holes above a hall tell you where the energy went in. They cannot tell you whether the centrifuges 80 metres below survived, were wrecked, or had already been moved. The strike was aimed with real precision at a target picked for being unreachable, and the picture from orbit proves the first half of that sentence far better than it will ever prove the second.

What the ground recorded: an earthquake that never came

The most durable myth of the Fordow strikes is a number: magnitude 4.3, the "bunker-buster earthquake" that supposedly shook central Iran when American B-2s put twelve GBU-57 bombs into the mountain on 22 June 2025. It circulated within hours, got attached to breathless "was this a nuclear test?" speculation, and hardened into received fact. It is worth stating plainly, because almost nothing else about Fordow can be: the earthquake never happened. The single most differentiating seismic finding of the entire war is a null result. Neither the US strikes on Fordow nor the earlier Israeli ones were large enough to enter the global seismic catalogs at all. The ground kept a record, and the record says the teleseismic networks saw nothing.

What the evidence actually shows is a query anyone can rerun. The USGS earthquake catalog, interrogated for the Fordow box (34.4 to 35.4 degrees north, 50.4 to 51.5 east) across 13 to 24 June, returns zero cataloged events. The 22 June strikes did not register teleseismically. That is not the absence of a good source; it is a positive finding from the primary global catalog. To guard against a single-network gap, the same box run against the independent European-Mediterranean Seismological Centre returns the identical null: two independent global catalogs, built from different station networks and different analysts, agree that nothing at Fordow crossed their detection threshold. EMSC's only nearby events in the window were natural M3.2 and M3.3 quakes, 130 to 150 km away, unconnected to any strike.

So what did the blasts actually register on? Local Iranian instruments, and only local ones. Mehdi Zare of Iran's Institute of Earthquake Engineering and Seismology (IIEES) reported that Iranian stations recorded over 1,500 explosion-induced tremors across the twelve days, and that the blasts near Fordow, Natanz, and Isfahan reached only about magnitude 2. That is the true shape of the signal: not one clean seismic slam but a scatter of roughly 1,500 small, close-range thumps, none of them large enough to escape Iran's own borders on a seismograph. A magnitude-2 event is felt, if at all, only in the immediate vicinity. It does not propagate as the teleseismic body wave the worldwide network is built to catch, which is precisely why the global catalogs are empty.

Against that primary record, the viral "M4.3" figure has no home. It appears in no primary seismic catalog, is attributed to no named monitoring agency, and is contradicted directly by both USGS and EMSC. It should be treated as an unverified aggregator number, and its persistence is instructive: it survives because it sounds like the kind of thing bombing a mountain should produce. The magnitudes that do circulate turn out, on inspection, to be three different events braided together and misdated. The M2.5 tremor near Qom that many outlets pinned on the American strike was in fact reported by Iran's Seismological Center on 16 June, coinciding with an Israeli airstrike near Fordow, six days before the US operation. And the genuine magnitude-5.1 earthquake widely reported "near Fordow" was a natural tectonic quake on 20 June near Semnan, roughly 200 km east of the site, unrelated to any strike. Three separate signals, two of them nothing to do with the US bombs, collapsed by rumor into one imaginary "obliteration quake."

Why the physics guarantees this, and why it is not a monitoring failure, is the part the myth ignores. A conventional bomb couples very little of its energy into the earth as a seismic wave. Even twelve 30,000-pound GBU-57s, walked into the same two ventilation shafts in sequence, release orders of magnitude less energy than the smallest underground nuclear tests the global network is tuned to detect. A nuclear detonation vaporizes rock and radiates a sharp, near-instantaneous, and (crucially) isotropic pulse in all directions; a shaped, delayed-fuze penetrator detonates inside a void it has already bored, so much of its energy vents into the tunnel and blows debris upward rather than driving a spherical shock into surrounding rock. Chemical explosives couple poorly into hard rock to begin with, and a sequenced strike is not one large event but a string of small, separated ones. The sum stays below the teleseismic threshold. This is the same discriminant the CTBTO uses for test-ban monitoring, and it runs the other way too: the complete absence of any large isotropic signal anywhere in Iran during June 2025 rules out the "accidental nuclear test" speculation entirely. If a nuclear device had gone off underground, the network would have caught it and flagged its distinctive signature. It caught nothing of the kind, because nothing of the kind occurred.

Why it matters, and this is the honest limit that defines the whole Fordow question, is what seismology can and cannot tell you. The seismic record is genuinely powerful for what it is built to do: it can time a blast to the second, confirm its conventional (not nuclear) character, and bound its energy. On all three counts it delivers a clean verdict. What it categorically cannot do is see inside the mountain. A seismograph measures ground motion; it does not image a hall 80 to 90 metres down, and it cannot tell you whether a single IR-6 centrifuge was knocked off its rotor. The seismic null does not mean the strikes failed; a strike that pulverized every centrifuge in Fordow and a strike that missed the hall entirely would produce the same M2 tremor on the same local stations and the same empty global catalog. This is the discipline the rest of the damage debate keeps losing. The ground can tell you an earthquake never came. It cannot tell you what died in the dark, and anyone who claims the seismographs "confirmed" Fordow was destroyed, or that they prove it survived, is reading a story into an instrument that was never designed to tell it.

What orbit could and could not see

The single most contested claim of the entire war was President Trump's assertion that Fordow had been "completely and totally obliterated," delivered in a national address hours after the strikes. Fordow sits ~80 to 90 metres under a mountain near Qom, so the one question that mattered, whether the ~2,000 to 2,900 centrifuges in its underground hall survived, was also the one question no camera in orbit could answer directly. What satellites could do was document the surface with forensic precision, and that surface record is unusually rich. The honest reading of it is neither "obliterated" nor "barely scratched": it is severe surface damage consistent with, but not proof of, destruction below. Everything in this section is what commercial and scientific satellites actually recorded in June 2025, and, just as importantly, what they structurally could not.

What the evidence actually shows. Start with the optical record, because it is the most legible. High-resolution imagery from Maxar Technologies on 22 June, analysed by Bloomberg's graphics team and the Institute for Science and International Security, showed six penetration holes punched into the ridge over Fordow's underground hall, arranged in two clusters of three. That geometry is not incidental. It matches the account Joint Chiefs Chairman Gen. Dan Caine gave in his 26 June briefing: twelve GBU-57 bunker-busters were "walked" into two ventilation shafts, the first weapon stripping the concrete cap, weapons two through five entering the exposed shaft at over 1,000 feet per second to detonate deep inside, and a sixth held as a backup. The surface craters, in other words, corroborate the aiming method the US described, weapon by weapon. This is a genuine case where independent commercial imagery and an official military account converge rather than collide.

The imagery also caught the preparation of the site. Maxar frames from 19 to 20 June, read by The War Zone, showed roughly 16 cargo trucks queued at Fordow's tunnel entrance and bulldozers sealing the entrances, which analysts interpreted as deliberate pre-strike backfilling, possibly to protect the tunnels, possibly to move material, possibly both. Read alongside the IAEA's own record that Iran's roughly 400 kg stockpile of 60% enriched uranium was verified in mid-June and then lost to monitoring, with imagery suggesting most was trucked into the Isfahan tunnel complex around 9 June, the satellite picture of trucks and sealed doors becomes part of the war's central unresolved thread: what was still inside the mountain when the bombs arrived.

Beyond ordinary optical cameras, two more specialised sensors saw things no wire photographer could. The first is thermal. NASA's VIIRS instrument, running the Nightfire algorithm, flagged an infrared anomaly about 3 km north of Fordow on the night of 21 June, detected by two passing satellites at 22:57 and 23:23 UTC, hours before the bombers arrived. The Payne Institute at the Colorado School of Mines, whose researcher Mikhail Zhizhin led the analysis, estimated the heat source at roughly 1,200 K and 2 MW of radiative output. This was a pre-bombing event visible only from orbit, its cause unresolved, and it matters precisely because it is the kind of signal that never appears in the news cycle and only a scientific sensor pointed continuously at the Earth's night side could have caught.

The second is hyperspectral. The commercial constellation operated by Orbital Sidekick detected ground subsidence over an area of roughly 315 by 275 metres at Fordow, with concrete apparently pushed up to the surface, reported by Asia Times. Subsidence on that scale is the surface fingerprint of a void collapsing underground, exactly what you would expect if a large buried hall had failed. It is the closest any orbital sensor came to inferring the state of the hall itself. And it is still an inference, not an image, of what happened below.

The honest limit, which is the whole point. Every one of these signals reads the surface, or reads the ground's response to what happened underneath. None of them sees the centrifuge hall. Optical imagery cannot penetrate overburden: eighty to ninety metres of rock is opaque to a camera at any resolution. Six impact holes prove six weapons reached the ridge and detonated; they do not prove the hall below is rubble, nor that every cascade stopped spinning. Subsidence over a footprint of roughly 300 metres is consistent with a hall collapse; it does not confirm one, and hyperspectral analysts said as much. Even the IAEA's Rafael Grossi told the Security Council on 22 June that craters were visible indicating ground-penetrating munitions, and then said plainly that no one, the Agency included, could yet assess the underground damage at Fordow. That is the ceiling on satellite forensics stated by the person with the most authority to state it. This is also why the battle-damage assessment splintered across agencies, from a leaked DIA estimate of "a few months" to Pentagon and CIA figures running to years: the imagery was compatible with several of those conclusions at once, because it stops at the rock.

Why it matters. Satellites became the independent referee in a dispute that governments, each with an interest in the answer, could not settle among themselves. That role is real and valuable: without Maxar and Planet, "obliterated" and "barely damaged" would be competing assertions with no public evidence between them, and instead we have a documented, dated, forensically consistent record of six holes over two shafts, sealed tunnels, a pre-strike thermal flare, and measurable ground subsidence. But the referee can only rule on what crossed the surface. The centrifuge hall's true state was, by late 2025, assessed as severely damaged and inferred to be collapsed, and not confirmed by any orbital sensor, because confirming it would require seeing through a mountain, and nothing in orbit can do that. The lasting lesson of Fordow is that the most advanced imaging fleet ever pointed at a target can tell you, with great precision, everything except the one thing you most want to know.

Obliterated, or set back months? The assessment war

Within hours of the last GBU-57 detonating inside Fordow's mountain on 22 June 2025, the war over what had actually happened underground began, and it was fought not with bombs but with assessments. President Trump told the nation the three nuclear sites had been "completely and totally obliterated", and the White House later published a page titled "Iran's Nuclear Facilities Have Been Obliterated, and Suggestions Otherwise Are Fake News." That single word, obliterated, is the claim this section exists to test. The honest answer, as the record stood by late 2025, is that no external party could see inside Fordow's enrichment hall, that the professional assessments genuinely disagreed, and that the disagreement was not noise but a real function of different methods, different evidence, and different incentives. Assessing damage to a hall buried under eighty to ninety metres of rock is inherently probabilistic, and anyone offering a clean number was either guessing or selling.

What the evidence actually showed was a spread, not a point. At one end sat the administration's framing. At the other, two days after the strike, came the most explosive counter-claim: a leaked preliminary Defense Intelligence Agency assessment, reported by CNN on 24 June, that concluded the strikes had not destroyed the core of Iran's program and may have set it back only by months. The DIA judgment found the most heavily fortified sites largely structurally intact and assessed that Iran's stockpile of enriched uranium had likely been moved before the strikes. Three things about that assessment matter and are usually stripped away when it is quoted: it was preliminary, it carried low confidence, and it had not been coordinated across the intelligence community. The White House called it flatly "wrong". So the two poles of the public fight, "obliterated" and "months," were an official political claim against an uncoordinated, low-confidence intelligence draft, and neither was a settled finding.

The middle of the range is where the more durable 2025 assessments actually landed. On 25 June, CIA Director John Ratcliffe issued a statement saying the agency had "a body of credible evidence" that the program was severely damaged, that several key facilities were destroyed and would have to be rebuilt over years, and DNI Tulsi Gabbard echoed that rebuilding Natanz, Fordow and Isfahan would likely take years. Read carefully, that is a deliberate middle position: it countered the DIA "months" leak while stopping short of Trump's "obliterated." "Severely damaged" and "years to rebuild" is a real claim, but it is not "destroyed," and the gap between those two is precisely the space the assessment war was fought in.

The most rigorous independent technical work came from the Institute for Science and International Security (ISIS), whose post-attack assessment of the first twelve days analyzed IAEA reporting and commercial imagery rather than classified intercepts. ISIS estimated the strikes had knocked out on the order of 22,000 centrifuges across Fordow, Natanz and Isfahan, a figure later reported by Iran International, and read the sites as significantly damaged and largely disabled. Crucially, ISIS did not say "destroyed." The distinction is the whole ballgame: disablement of installed centrifuges is not the same as destroying the program, because centrifuges can be rebuilt, and the material that matters most, Iran's near-weapons-grade uranium, is portable. That is why ISIS analyst David Albright, reading Maxar imagery of the Fordow craters and blocked shafts, stressed that heavy surface damage and sealed tunnels were visible from orbit while internal enrichment-hall damage remained unverifiable from imagery alone, an excavator working the ventilation-shaft impact points by 30 June looked to him like damage assessment and possible backfilling, not proof of what survived below.

The most cautious voice, and arguably the most authoritative on the physics, was IAEA Director General Rafael Grossi. On 22 June he told the UN Security Council that craters at Fordow indicated ground-penetrating munitions but that no one, including the IAEA, could yet assess the underground damage. By 26 June the agency assessed Fordow was directly impacted and that its centrifuges were likely no longer operational, while cautioning it could not determine the internal damage with certainty and could not verify the fate of the enriched-uranium stockpile. Then, on 29 June, Grossi added the finding that undercut "obliterated" most directly: Iran could resume enriching uranium "in a matter of months," noting it could have "a few cascades of centrifuges spinning" relatively quickly, and he flagged Iran's roughly 400-plus kilograms of 60% enriched uranium as the unresolved concern. That stockpile is the crux. The IAEA had verified 440.9 kg of 60% material on the eve of the strikes, and after the attacks the agency lost continuity of knowledge, unable to confirm the stockpile's location or condition, with imagery suggesting much of it had been moved into the Isfahan tunnel complex around 9 June. You cannot "obliterate" a program while the material that defines it is unaccounted for and probably intact underground.

Why it matters is that these were not five people describing one fact and lying at different rates. They were five instruments pointed at a mountain from different angles, each seeing part of it. The administration measured the operation's execution, which was near-flawless. The DIA draft measured structural survival and stockpile displacement, on thin, uncoordinated early evidence. ISIS measured centrifuge disablement from imagery and safeguards data. Grossi measured what could be verified, which after Iran suspended IAEA cooperation on 2 July was almost nothing. The word "obliterated" fails not because the strikes were ineffective, the damage was real and heavy, but because it converts a probabilistic, partially observable, still-classified question into a marketing certainty. The defensible 2025 statement is narrow and unsatisfying: the strikes inflicted severe, likely years-costly damage to enrichment infrastructure, they did not verifiably destroy Iran's stockpile of near-weapons-grade uranium, and no one outside the mountain could confirm the state of Fordow's hall. On the war's central question, the most honest finding is that the record did not resolve it, and every party claiming otherwise was reaching past its own evidence.

The material that outweighs the buildings

The world watched twelve GBU-57 bunker-busters walk into two ventilation shafts above Fordow and asked the wrong question. The question was never whether the concrete broke. It was whether the enriched uranium, the only part of Iran's program that a bomb can either destroy or scatter but not un-make, survived, and where it went. Centrifuges are machines: intricate, sanctioned, and slow to replace, but replaceable. Near-weapons-grade uranium is a stock, an accumulated inventory of physics that took years of enrichment cascades to build and that no airstrike can enrich backward. Destroying the halls sets back Iran's ability to make more. It does nothing to the material Iran had already made, and by mid-June 2025 that material was very close to the amount and the enrichment level that matter.

What the evidence actually shows is a number, a date, and then a hole in the record. The last time the IAEA physically verified the stockpile, on 17 May 2025, Iran held 408.6 kg of uranium enriched to 60% U-235. Enrichment did not pause for the diplomacy. By 13 June, the day before Israel's opening strikes, the Agency later calculated Iran's holding at 440.9 kg of that same 60% material, a figure the IAEA published only in its September verification report because after mid-June it could no longer count. Sixty percent is the number that should stop a reader: it is not weapons-grade (that is roughly 90%), but the hard, energy-intensive work of enrichment is front-loaded, so jumping from 60% to 90% is a short technical step, not another marathon. The Bulletin of the Atomic Scientists put the proliferation stakes plainly: the 400-plus kilograms of 60% uranium that fell outside IAEA monitoring after the strikes is, if further enriched, enough for roughly ten nuclear weapons. That is the metric that outweighs every crater on the ridge.

Here is where honesty about the limits has to govern the story, because the single most important fact is one nobody outside Iran can confirm: where the material is. The strikes may not have eliminated the stockpile so much as scattered it, or missed it entirely. A leaked preliminary US Defense Intelligence Agency assessment, reported on 24 June, found that Iran's enriched-uranium stockpile had likely been moved before the strikes, and that the most fortified sites remained largely intact. Satellite analysts read the same signature: Maxar imagery from 19-20 June showed roughly 16 cargo trucks queued at Fordow's tunnel entrance and bulldozers sealing the openings in the days before the bombs fell, movement that analysts read as deliberate pre-strike relocation and backfilling, not panic. The most cited working hypothesis is that much of the material was trucked into the deep tunnel complex at Isfahan around 9 June, where it may now sit buried under a site the Tomahawks could crater but not reach. None of this is verified, and it must not be dressed up as if it were: the honest formulation is that the stockpile is unaccounted for, its location assessed rather than known, and that it is possible some fraction was buried inside Fordow's collapsed halls, some at Isfahan, and some elsewhere. A buried stockpile is a delayed problem, not a solved one. Uranium does not degrade; a mountain of debris is an obstacle to recovery, not a substitute for destruction.

The reason this uncertainty became permanent rather than temporary is the diplomatic aftermath, and here the dates are firm and 2025-sourced. The strikes did not just damage buildings; they severed the IAEA's continuity of knowledge, the unbroken chain of inspection and monitoring that lets the Agency say with confidence how much fissile material a country holds and where. On 25 June, one day after the ceasefire, Iran's parliament voted 221-1 to suspend cooperation with the IAEA. On 2 July 2025, President Masoud Pezeshkian signed that suspension into law, barring inspectors from Iran's nuclear sites without approval from the Supreme National Security Council. From that moment the Agency was, in its own September accounting, unable to visit any safeguarded facility except the Bushehr power plant, and could not verify the current location or condition of the 440.9 kg. Director General Rafael Grossi had already warned, in an interview aired 29 June, that Iran retained the roughly 400-plus kilograms of 60% material and that his central concern was verifying none of it had been diverted, a concern he could no longer act on. By late 2025 the record had not closed the gap: the IAEA Board's 20 November resolution still had to demand that Iran account for the stockpile, which is the clearest possible admission that, as of the period this briefing covers, the material remained unaccounted for.

Why it matters is a distinction the "obliterated" framing erased. There are two clocks running on any nuclear program, and the strikes ran them in opposite directions. The first is the production clock: how fast Iran can spin new material. On that clock the campaign scored a real, if disputed, win. Grossi assessed Fordow's centrifuges were likely no longer operational, and ISIS estimated the strikes knocked out on the order of 22,000 centrifuges across Fordow, Natanz and Isfahan, which sets production back by a span that agencies pegged anywhere from a few months to many years. The second is the breakout clock: how fast Iran could turn what it already has into a weapon. On that clock the strikes may have moved the hands very little, because the 440.9 kg of 60% material is the input that clock runs on, and it appears to have been moved out of the buildings before the buildings were hit. Grossi's own warning, that Iran could have "a few cascades of centrifuges spinning" again within months, only matters because the stockpile that those cascades would finish enriching was not destroyed with the halls. Bombs are good at concrete and bad at inventory. The Twelve-Day War demonstrated that a nation can hit a mountain with the heaviest conventional weapon ever built, film the craters, and still not be able to say, months later, where the ten-weapons' worth of uranium actually is. That gap between the visible damage and the invisible material is the honest center of the Fordow question, and it is why the concrete was never the point. Reading a stockpile through enrichment levels and breakout arithmetic is specialist judgment that AI labs now pay for as well: the nuclear and CBRN roles listed on AITraining.jobs, a specialist jobs board we operate, are for experts who audit how AI models handle exactly this kind of dangerous knowledge.

The verdict the evidence supports, and the one it does not

Lay the signals side by side and a careful, defensible reading emerges. It is not the reading anyone with a stake in the war wanted, because it refuses to be a headline. But it is the one the primary data will bear.

On the aim, the record is strong. The satellite imagery corroborates the official account of how the weapon was used: six penetration holes in two tight clusters along the ridge over Fordow, matching the two ventilation shafts that General Caine said the bombs were walked into. The pre-strike truck movement and the sealing of tunnel entrances are visible in the 19-20 June imagery. The thermal anomaly the night before is in NASA's sensor record. On the question of whether the United States hit what it aimed at, and hit it hard, the evidence is consistent and points one way: it did.

On the internal result, the record is honest about its own blindness. The seismographs are empty at Fordow not because nothing happened but because conventional bombs cannot register through that much rock, so the null tells you nothing about the centrifuge hall. The optical imagery stops at the surface. The subsidence over an area consistent with a hall collapse is a strong indirect sign of severe internal damage, and it is the best single piece of evidence that the strike did real harm underground. But subsidence is an inference, not a photograph of a broken cascade. The most that the imagery, the seismology and the physics can jointly support is: severe damage is likely, total destruction is unproven, and continued operation is unlikely in the near term. That is a genuine finding. It is also a range, and the range is the truth.

Which is why the four-agency split is not merely politics. When a leaked Defense Intelligence Agency assessment said "a few months" and the administration said "obliterated," the two were not only disagreeing about Iran; they were filling the same evidentiary gap with different assumptions. Assessing a buried, hardened hall from outside is inescapably probabilistic. Reasonable analysts, looking at the same craters and the same absence of a seismic signature, can land on "set back years" or "set back months" depending on what they assume about redundancy, about spare centrifuges, about how much of the cascade was spun down and moved before 22 June. The honest position is not to pick the estimate that flatters a prior. It is to report the spread, name the sources, and mark the assumptions that drive the difference. The independent technical assessment from the Institute for Science and International Security did exactly that, and its measured conclusion, significant damage and disablement across the sites without a claim of permanent elimination, has held up better than either government's slogan.

Fordow also cannot be judged alone, because the program it served did not live under that one mountain. Israel's opening strikes had already destroyed the above-ground enrichment plant at Natanz and battered the wider complex, and the American raid on 22 June also struck Isfahan, the home of Iran's uranium-conversion work and, by several analyses, a plausible resting place for the very stockpile that later slipped out of the accounts. A verdict on whether the strikes ended Iran's path to a weapon is therefore not a verdict on a single centrifuge hall. It is a verdict on three sites, on the material that could move between and beyond them, and on the monitoring, now suspended, that would have let anyone follow it. Fordow is the most dramatic piece of the picture and, taken by itself, the least conclusive.

And then there is the part of the answer that does not depend on the buildings at all. Destroying centrifuges sets back the machinery of enrichment. It does not touch the material already enriched, and by mid-June Iran held roughly 440 kilograms of 60 percent uranium, the stockpile that is the actual proliferation risk. After the strikes, more than 400 kilograms of that material fell outside IAEA monitoring, its whereabouts unresolved, and on 2 July Iran suspended cooperation with the inspectors who might have tracked it. A strike can bury a facility and still leave the most dangerous thing in the program mobile, unaccounted for, and in the hands of a state that has just decided the inspectors are the enemy. Any verdict on Fordow that stops at the concrete has answered the smaller question and skipped the larger one.

Reading a strike no one can see

The reason to spend this many words on a question that cannot be closed is that the way it was answered in public is the exact failure Blackwing exists to correct. Within seventy-two hours, the most consequential technical assessment of the decade had been reduced to a contest between the word "obliterated" and the phrase "a few months," each backed by an institution and an incentive, neither backed by anything the outside world could verify. The instruments that could speak, the seismographs and the satellites, were quoted selectively by whichever side found them convenient, and the one instrument that mattered most, the IAEA's continuity of knowledge over Iran's uranium, was going dark at exactly the moment it was needed.

The discipline that produces a trustworthy account here is not access to a secret. It is the willingness to separate the corroborated from the assumed and to say out loud where the evidence stops. The aim is corroborated: read the imagery. The near-term disablement is well-supported: read the subsidence and the impact pattern. The claim of permanent, total destruction is not something any outside instrument can prove, and neither is its opposite, so an honest record files both as assertions beyond the evidence and reports the defensible middle as a range. And the fate of the enriched uranium, the part that actually governs how fast Iran could rebuild a weapon's worth of material, is marked unresolved, because it is.

The quietly decisive fact in the whole episode is not a crater; it is the loss of continuity of knowledge. The IAEA's ability to say what Iran holds, and where, rests on an unbroken chain of inspections, seals and cameras. Break the chain and even a flawless strike leaves the world unable to verify what it accomplished, and Iran's 2 July suspension broke it. From that day, the honest answer to how much uranium survived and where it now sits stopped being "we are not sure" and became "we can no longer check." A strike that destroys machines but ends the monitoring can leave the world less certain about Iran's capabilities than it was before the first bomb fell. That paradox is the part the loud verdicts skipped, and it is the part that will still matter when the craters are old news.

That is a less satisfying story than "we destroyed it," and less alarming than "it was all for nothing." It has the single advantage of being what the data supports. In a war whose defining question turned on a hall no camera has ever seen, the most useful thing a record can do is refuse to pretend it saw it.

Every figure in this briefing traces to a primary source you can open, and the full event record is published as open data so you can check the reconstruction against the sources yourself. The live map places the Fordow strike in the sequence of the twelve days; the intelligence page carries the seismic, connectivity and market instruments still pointed at the same ground; and the full timeline lists every sourced event behind this analysis. If you can show, with a source, that any part of this reading is wrong, we will correct it. On a question this important, that willingness is not a weakness of the record. It is the only reason to keep one.