Harvard's Matthew Bunn on the factors driving the risk of nuclear war — and how we shift them toward greater safety, for ourselves and for future generations.
What does a one-in-a-thousand chance look like over a thousand years?
In eighty years, no nuclear weapon has been used in war — a record almost no one in 1945 would have predicted. But Harvard’s Matthew Bunn argues that luck has played an uncomfortably large part, and that even a small annual risk of nuclear war, compounded over the spans of time Long Now thinks in, adds up to near-certainty unless the risk keeps falling or we stop running it altogether. Speaking at the Foundry in Cambridge the day after returning from unofficial U.S.–Russian talks in Armenia, the Harvard Kennedy School professor laid out how a nuclear war could actually begin, why the danger has grown, and what can be done about it.
Matthew Bunn
Matthew Bunn is the James R. Schlesinger Professor of the Practice of Energy, National Security, and Foreign Policy at Harvard Kennedy School, and Co-Principal Investigator of the Belfer Center's Project on Managing the Atom. His research focuses on nuclear theft and terrorism, nuclear proliferation and measures to control it, and the future of nuclear energy.
Speakers
Event Summary
What does a one-in-a-thousand chance look like over a thousand years?
In eighty years, no nuclear weapon has been used in war — a record almost no one in 1945 would have predicted. But Harvard’s Matthew Bunn argues that luck has played an uncomfortably large part, and that even a small annual risk of nuclear war, compounded over the spans of time Long Now thinks in, adds up to near-certainty unless the risk keeps falling or we stop running it altogether. Speaking at the Foundry in Cambridge the day after returning from unofficial U.S.–Russian talks in Armenia, the Harvard Kennedy School professor laid out how a nuclear war could actually begin, why the danger has grown, and what can be done about it.
Bunn began with Nagasaki in September 1945 — one bomb, one airplane, the heart of a city gone — and then brought it home. A typical Russian warhead detonated over the Massachusetts State House would produce a fireball reaching some 1,400 feet in every direction, flatten most buildings within two miles, and very likely set the city burning. There are more than 12,000 such weapons in the world today, held by nine states, and almost all are far larger than the Hiroshima and Nagasaki bombs. Burning cities could also loft enough smoke to cool the planet: some studies suggest even a regional war between India and Pakistan could cut harvests enough to starve on the order of a billion people.
Deterrence makes a deliberate decision to start a nuclear war hard to imagine, Bunn acknowledged — but the history of near misses is sobering, and he is co-editing a book that examines them systematically. Kennedy put the odds of nuclear war during the Cuban Missile Crisis at one in three; a Soviet submarine officer, Vasili Arkhipov, talked his captain out of firing a nuclear torpedo that the U.S. Navy didn’t know was aboard. In 1961, over Berlin, the president was seriously briefed on a plan for a nuclear first strike. In 2022, U.S. intelligence estimated a 50-50 chance that Putin would use nuclear weapons if Ukraine’s counteroffensive threatened Crimea. Bunn offered a framework for reducing the risk at each step of the path — from peace to crisis, from crisis to armed conflict, and from conflict to nuclear use — and noted that every motivation for use, from preemptive to coercive, has been planned or considered at some point in the nuclear age, alongside the dangers of unauthorized use, false alarms, and accidents.
Why is the risk higher now — the highest since the Cuban Missile Crisis, as Bunn argued in an editorial in Science? U.S.–Russian official dialogue has nearly vanished, and the last treaty limiting their nuclear forces expired in February with no replacement under negotiation. China has more than tripled its arsenal, with no strategic dialogue underway with Washington. New technologies — AI, cyber, counter-space weapons, systems that mix conventional and nuclear roles — are reshaping the picture; Bunn cited a report of an AI system that wrongly concluded a Chinese ship was carrying nuclear equipment, nearly prompting U.S. forces to board it. He also surveyed the hot spots: Iran, where he judges the chance of a bomb within a decade has risen rather than fallen; North Korea, with scores of weapons; South Asia; and the lingering threat of nuclear terrorism.
Then the good news: more than four-fifths of all the nuclear weapons that ever existed have been dismantled. Fewer than 5% of the world’s states have nuclear weapons — the same as 35 years ago, despite the Soviet collapse and the black-market network. More than half of the states that started weapons programs abandoned them, and more than half of the states that once held bomb-usable material have given it up — much of it through a program Bunn himself proposed. The gaping holes in fences and doors propped open with textbooks that he once saw with his own eyes are gone.
What can we as individuals do? His answer: get informed, get involved with organizations across the political spectrum, and — for those with the means — help fund the work, as nongovernmental funding for nuclear policy has fallen by roughly four-fifths in twenty years. A back-of-the-envelope estimate suggests that even a modest reduction in risk would repay such investment several times over in lives saved, and Bunn noted that many past successes began as ideas from outside government. He closed on the importance of judgment at the top: Kennedy rejected his advisors’ unanimous call for airstrikes and invasion in 1962, unaware that Soviet tactical nuclear weapons were already in Cuba. The lesson, Kennedy said, was to always give an adversary a choice other than humiliating defeat or nuclear war.
In a lively Q&A moderated by Long Now Boston president Gary Oberbrunner, Bunn took aim at the premise of the film A House of Dynamite, recalled advising a Hollywood producer on an alien beam that would disable nuclear weapons (strengthen the strong force, he offered — then realized the sun would probably explode), weighed Pakistan’s new defense pacts with Saudi Arabia and Turkey, shared what he could from the Armenia dialogue, explained how the “damage limitation” doctrine drives a likely U.S. warhead buildup to cover China’s new missile silos, considered Europe’s security rethink in the face of Russian pressure and doubts about American reliability, and told of a Vatican meeting on AI and nuclear weapons held, in July, inside a greenhouse.
Full slides of this talk are available: https://matthewbunn.scholars.harvard.edu/sites/g/files/omnuum3466/files/2026-09/Preventing%20Nuclear%20War%202026.pdf
Professor Bunn will also be appearing on a new podcast on nuclear dangers with Pultizer-Prize winning journalist Jane Perlez. First episode drops Oct. 6.