No one can predict with certainty whether nuclear war will happen, but the honest answer is that the risk is higher now than at any point since the Cold War. The Bulletin of the Atomic Scientists set its Doomsday Clock to 89 seconds to midnight in January 2024, then moved it even closer in January 2026 to 85 seconds, the nearest it has ever been in the Clock’s eight-decade history. That doesn’t mean a nuclear exchange is imminent or inevitable. It means the guardrails that kept the world safe for decades are eroding, and several dangerous trends are running at the same time.
Where the World’s Nuclear Weapons Are Right Now
Nine countries possess nuclear weapons. As of early 2024, roughly 3,904 warheads were actively deployed on missiles and aircraft, an increase of about 60 from the year before. Around 2,100 of those sat on ballistic missiles in a state of high operational alert, meaning they could be launched within minutes. Nearly all of those high-alert warheads belong to the United States and Russia, but China is now believed to have placed some of its warheads on alert for the first time.
China’s nuclear expansion has been faster than Western analysts expected. The country has long maintained a no-first-use policy, pledging it would never launch a nuclear weapon unless one were used against it first. But its growing arsenal and shift toward keeping warheads ready for rapid launch signal a changing posture. The United States and Russia, meanwhile, each maintain arsenals large enough to end civilization several times over.
Why the Risk Is Elevated
Several factors are converging to push nuclear risk upward. The most concrete is the collapse of arms control. New START, the last remaining treaty limiting U.S. and Russian nuclear arsenals, was extended through February 4, 2026. Russia suspended its participation in the treaty in 2023, and no replacement agreement is being negotiated. When New START lapses, there will be no binding limits on the size of any country’s nuclear arsenal for the first time since the 1970s.
The war in Ukraine has kept nuclear threats in public discourse. Russia has repeatedly referenced its nuclear capability in the context of the conflict, and its official doctrine permits nuclear use if the state faces an existential threat. The United States has historically sized its nuclear forces around deterring and, if necessary, defeating Russia. With both nations in an adversarial posture and no active diplomatic channel on arms control, the margin for miscalculation is thinner than it has been in decades.
The U.S. is also in the middle of a massive modernization of its nuclear forces. The Sentinel program, a new intercontinental ballistic missile designed to replace the aging Minuteman III, carries an estimated price tag of $141 billion and is expected to deliver initial capability in the early 2030s, with the missiles serving through 2075. Russia and China are running their own modernization programs. This three-way arms competition is new territory; Cold War arms control only ever had to manage two players.
How Technology Makes Accidental Escalation More Likely
One of the most underappreciated risks isn’t a deliberate decision to launch. It’s a chain of errors, misreadings, or technical failures that spirals out of control. This nearly happened several times during the Cold War, most famously in 1983 when a Soviet satellite system falsely detected incoming American missiles. A single officer’s judgment call prevented a retaliatory strike.
Today’s risks are different and in some ways harder to manage. Artificial intelligence is compressing the time leaders have to make decisions during a crisis. AI-enhanced cyber capabilities could, deliberately or accidentally, degrade a nuclear state’s command-and-control systems. If a country believes it is about to lose the ability to launch its weapons, the pressure to use them before they’re destroyed creates a “use it or lose it” scenario. Research from West Point’s Modern War Institute highlights that AI-driven disinformation could also confuse intelligence assessments during a crisis, making it harder for leaders to distinguish a real attack from a false alarm.
The problem isn’t that any government wants a nuclear war. It’s that the systems designed to prevent one are being stressed by new technologies that move faster than human decision-making.
What Forecasters Actually Estimate
Putting a number on the probability of nuclear war is inherently uncertain, but forecasting platforms offer a useful window into how informed analysts weigh the risk. Metaculus, a well-regarded prediction aggregation site, has its community placing the probability of a nuclear weapon being detonated as an act of war before 2050 at roughly 23%. That figure encompasses everything from a single tactical weapon used in a regional conflict to a full-scale exchange, and it spans a 25-year window.
A 23% chance over several decades may sound low for any given year, but it’s disturbingly high for an event that could kill hundreds of millions of people. For perspective, if you had a 23% chance of being in a serious car accident over the next 25 years, you’d probably change how you drive.
What a Nuclear War Would Actually Do
The immediate destruction of nuclear weapons is well understood: a single modern warhead can flatten a city. What’s less intuitive is the global aftermath. Climate models from Penn State University show that even a regional nuclear war, involving a fraction of the world’s arsenals, would send roughly 5.5 million tons of soot into the upper atmosphere. That alone could reduce worldwide corn production by 7%.
A large-scale war between major powers would be catastrophically worse. Models project 165 million tons of soot entering the atmosphere, triggering a nuclear winter that would cause global temperatures to drop sharply and extinguish most agriculture. Annual corn yields could fall by 80%, with additional disruptions pushing the total loss to 87%. The resulting famine would likely kill far more people than the explosions themselves. These aren’t worst-case fantasies; they’re the outputs of peer-reviewed climate and agricultural models.
What You Can Realistically Do
For most people, the practical question isn’t whether nuclear war will happen but how to think about the risk in daily life. The probability in any single year remains low. The mechanisms that have prevented nuclear use for nearly 80 years, primarily the shared understanding that no one wins, still hold. But “low probability” and “zero probability” are very different things, and the trend lines are moving in the wrong direction.
If you want to be minimally prepared, the guidance from Ready.gov is straightforward. In the event of a nuclear detonation, the most important action is getting inside a solid structure, ideally a basement or the center of a large building, and staying there for at least 24 hours. Radiation levels drop dramatically in the first day. That single step, sheltering in place for 24 hours, is the most effective thing a civilian can do to reduce radiation exposure.
On a broader level, the experts who set the Doomsday Clock have called for three specific actions: limiting nuclear arsenals through new agreements, creating international guidelines for AI use in military systems, and forming multilateral frameworks to address emerging biological threats. These are policy decisions, not individual ones. But public pressure has influenced arms control before, most notably in the 1980s nuclear freeze movement that helped push the U.S. and Soviet Union toward the treaties now expiring.