A third world war would almost certainly involve the threat of nuclear weapons, and the risk of their actual use would be higher than at any point since 1945. Whether the conflict stayed conventional or crossed the nuclear threshold depends on who is fighting, what they stand to lose, and how quickly events spiral beyond anyone’s control. The short answer: a major war between nuclear-armed powers could go nuclear, but it is not guaranteed to, and the path from conventional fighting to nuclear exchange is more complicated than most people assume.
Why Nuclear Weapons Would Be Part of the Picture
Nine countries currently possess nuclear weapons: the United States, Russia, the United Kingdom, France, China, India, Pakistan, North Korea, and Israel. Together they hold roughly 12,241 warheads as of early 2025, with about 3,912 deployed on operational forces and around 2,100 kept on high alert, ready to launch on short notice. The US and Russia alone account for nearly 90 percent of the global total.
Any large-scale war involving even one of these states would be shaped by the existence of those arsenals, even if no warhead is ever launched. Nuclear weapons define the boundaries of what each side is willing to risk. They influence which targets get hit, which allies get drawn in, and how far any offensive push can go before the other side starts signaling that it might escalate.
The Steps Between Conventional War and Nuclear Use
Conflicts don’t jump from peacetime to nuclear war in a single step. Military strategists have long mapped out “escalation ladders,” sequences of increasingly severe actions that move from diplomatic threats through conventional warfare to chemical, biological, tactical nuclear, and finally strategic nuclear strikes. The Cold War theorist Herman Kahn famously outlined a 44-step ladder ranging from political crises to all-out nuclear exchange, and modern versions of this concept still guide military planning.
The critical insight is that several major thresholds exist along the way. Geographic escalation (the war spreading to new regions), escalation in the type of weapons used, and escalation in the kinds of targets hit (military bases versus cities) are all distinct steps. Each threshold gives both sides a chance to pause, negotiate, or back down. The fact that these thresholds exist is one reason nuclear weapons haven’t been used in 80 years of conflict, even when nuclear-armed states have fought conventional wars or backed opposing sides in proxy conflicts.
But thresholds can also collapse quickly. If one side is losing badly, if communication breaks down, or if a leader misreads an incoming conventional strike as a nuclear one, the ladder can be climbed in hours rather than weeks.
Russia’s Approach to Nuclear Escalation
Russia’s military doctrine is the most relevant case study because it explicitly incorporates nuclear weapons into its war-planning at multiple levels. According to analysis from Lawrence Livermore National Laboratory’s Center for Global Security Research, Russia’s escalation framework moves through three phases: conventional strikes on military targets, conventional destruction of both military and civilian infrastructure, and then the use of smaller, “nonstrategic” nuclear weapons against critical economic, military, and political targets, potentially followed by strategic nuclear strikes.
A key element of this approach is what analysts describe as “dosing and calibrating” damage to sober an opponent without provoking full-scale nuclear retaliation. Russia’s theory of victory assumes it would control the pace of escalation, including through preemptive options. The doctrine also reflects a willingness to lean more heavily on nuclear weapons if conventional forces suffer significant losses on the battlefield, a scenario that has become more discussed since Russia’s difficulties in Ukraine.
This doesn’t mean Russia would automatically launch nuclear weapons in a major war. It means Russian military planners have built nuclear use into their playbook as a deliberate option, not just a last resort.
China’s Different Stance
China takes a fundamentally different public position. Since its first nuclear test in 1964, Beijing has maintained a no-first-use policy, pledging not to use nuclear weapons first under any circumstances. In January 2022, the leaders of all five original nuclear-weapon states (the US, Russia, the UK, France, and China) jointly affirmed that “a nuclear war cannot be won and must never be fought.”
China is, however, rapidly expanding its arsenal, growing from an estimated 500 to roughly 600 warheads over the past year alone. Some Western analysts question whether Beijing’s no-first-use pledge would hold if the country faced an existential threat, such as a conventional attack on its mainland or its nuclear forces. The gap between declaratory policy and actual behavior under extreme pressure is one of the great unknowns in nuclear strategy.
Tactical Nuclear Weapons Lower the Bar
One of the factors that makes nuclear use more plausible in a modern conflict is the existence of tactical nuclear weapons, smaller warheads designed for battlefield use rather than the destruction of entire cities. These weapons range from a fraction of a kiloton to about 50 kilotons, compared with strategic warheads that start around 100 kilotons and can exceed a megaton.
Their smaller size makes them more “usable” in military thinking. A commander facing a massive enemy advance might see a tactical nuclear strike on a troop concentration or supply hub as a way to change the course of a battle without triggering full-scale nuclear war. That logic is dangerous precisely because it treats nuclear use as a manageable step on the escalation ladder rather than a line that, once crossed, changes everything. There is no guarantee that the other side would view a “small” nuclear strike as limited.
How AI and Speed Increase the Risk
Modern technology has introduced new ways a conventional conflict could go nuclear by accident. Artificial intelligence is increasingly integrated into early warning systems and military decision-making, and the speed of AI-driven analysis can outpace human judgment. Research from West Point’s Modern War Institute warns that without carefully designed human override mechanisms, AI’s speed raises the risk of inadvertent escalation “initiated at algorithmic speed.”
The risks go beyond speed. Sophisticated AI could potentially generate convincing false signals: splicing fake imagery into satellite feeds, feeding malicious data into automated detection systems, or creating deepfakes that mimic key individuals in a country’s nuclear chain of command. Public commitments to keep humans “in the loop” on nuclear decisions lack clear definitions, and it remains uncertain whether human oversight applies to the assessment of early warning information that feeds into launch decisions.
In a fast-moving crisis, a false alarm processed by an AI system could compress the time a leader has to decide whether an incoming strike is real from roughly 30 minutes (for a traditional ballistic missile) to something far shorter, increasing the chance of a catastrophic mistake.
What a Nuclear Exchange Would Actually Mean
If nuclear weapons were used, even on a limited scale, the consequences would be staggering. A single 1-kiloton detonation, among the smallest possible, produces severe damage roughly a quarter mile from the blast point. Within that zone, virtually all buildings collapse and nearly all people die quickly. A 10-kiloton weapon extends that severe damage radius to about half a mile. A 1-megaton strategic warhead causes severe destruction beyond two miles from the blast, with burns, radiation, and infrastructure damage extending much further.
But the immediate blast is only the beginning. The most consequential long-term effect of a large-scale nuclear war would be agricultural collapse. Research published in 2025 in the journal Medicine, Conflict and Survival found that just 2 percent of the world’s current nuclear weapons, detonated over cities, would produce enough smoke and soot to cause ice-age temperatures globally. This would put more than 2 billion people at risk of starvation within two years. A war involving a substantial fraction of the global arsenal would threaten the vast majority of the human population and risk the extinction of many species, including potentially humans themselves.
The mechanism behind this is straightforward: burning cities would inject millions of tons of black soot into the upper atmosphere, blocking sunlight for a decade or more. Crops would fail worldwide, not just in the countries that were attacked. This means that even nations far from the conflict zone would face famine on a scale with no historical precedent.
Could a World War Stay Conventional?
It’s possible, but only under specific conditions. If a major war remained limited in geography, if neither side’s homeland was directly threatened, and if both sides maintained clear communication about their red lines, the conflict could theoretically stay below the nuclear threshold. The Korean War and the wars in Vietnam and Afghanistan were all fought with nuclear-armed powers on opposing sides without nuclear use.
The difference in a true world war scenario is scale. A conflict large enough to be called World War 3 would likely involve direct fighting between nuclear-armed states, territorial threats to homelands, and the kind of existential stakes that make nuclear doctrines relevant. The more one side faces decisive defeat, the more tempting nuclear weapons become as a way to change the outcome or force a ceasefire. Russia’s doctrine explicitly accounts for this. The US maintains what strategists call “calculated ambiguity,” deliberately leaving open the question of exactly when it might use nuclear weapons, which is itself a form of deterrence.
The honest answer is that no one knows where the line is. Deterrence works until it doesn’t, and the margin for error in a conflict between great powers armed with thousands of warheads is terrifyingly thin.