Balancing act: a ballistic missile on parade in Moscow, May 2014 .

Four months into his presidency, at a summit in Prague, Barack Obama pledged to take “concrete steps toward a world without nuclear weapons.” Yet nearly eight years later, he presides over a program to modernize the entire U.S. nuclear arsenal at a cost of $35 billion a year through the next decade and beyond. To those who accuse him of hypocrisy, Obama has said that he always regarded a nuclear-free world as a long-term goal, unlikely to be met in his lifetime, much less his time in office—and that his modernization program is designed not to build more or more deadly nuclear weapons but rather to maintain and secure the arsenal the United States has now.

This claim is true, by and large, but it leaves open a bigger question: Does the United States need the arsenal it has now? Obama seems to be mulling this very question as his tenure winds down. In a June 6 speech to the Arms Control Association, his deputy national security adviser, Ben Rhodes, noted that “the modernization plan was put together in a different budget environ­ment, with a different Congress,” and that the president “will continue to review these plans as he considers how to hand the baton off to his successor.” In one sense, Rhodes was merely repeating the concern that Robert Work, the deputy secretary of defense, had expressed back in February—that the nuclear plan’s price tag would force tradeoffs in an era of budget constraints and that if this meant cuts in conven­tional forces, then that would be “very, very, very problematic.” But other officials have said that the review Rhodes men­tioned is propelled not only by budgetary dilemmas but by questions of strategy and history, too.

Rhodes’ statement set off alarm bells in certain corridors of Congress. In a June 16 letter, Senators John McCain and Bob Corker, both Republicans, reminded Obama that during the debate over the New Strategic Arms Reduction Treaty (New START) in 2010, he had promised to modernize or replace all three legs of the nuclear triad—the land-based inter­continental ballistic missiles ICBMs), the submarine-launched ballistic missiles (SLBMs), and the long-range bombers—in exchange for Senate ratification. They warned him not to backpedal on this commitment.

 And so a quarter century after the end of the Cold War, the United States stands on the precipice of another nuclear debate. In the 1980s, nuclear weapons dominated discussions of national security affairs to a degree that special­ists under the age of, say, 50 would find baffling. The arsenals of both sides had grown to such staggering levels, and the chance of a real war between the two super­powers had so diminished, that the nuclear arms race entered a realm of almost pure abstraction, in which such recondite (and substantively meaningless) measures as “missile throw-weight ratios” became tokens of competition and conflict.

Notwithstanding the tensions between the United States and Russia in the era of Vladimir Putin, this sort of contest has long been abandoned. Tabulations of each side’s nuclear arsenal, which were once parsed with scholastic flair, are now hard to come by. No one serious would dream of presenting such statistics as a measure of the “balance of power,” how­ever that phrase might be defined. So it’s an ideal time—before the renewed debate is taken over by baroque abstractionists—to ask some basic questions. What does the United States need nuclear weapons for? And how many, of what sort, are enough? 


Public discussion of these questions has always been disingenuous. President John F. Kennedy’s defense secretary, Robert McNamara, devised a formula for “finite deterrence,” a concept popu­larized as “mutual assured destruction,” or MAD: if, after a Soviet first strike, enough U.S. weapons survived to destroy the Soviet Union’s 200 largest cities in a retaliatory blow, then that would be enough to deter the Russians from con­templating a first strike to begin with. The damage done by any additional weapons, McNamara argued, would be so marginal as to be superfluous. In fact, this formula was only the secretary’s way of capping the military’s appetite. (The Joint Chiefs of Staff wanted 10,000 ICBMs; McNamara held them to 1,000.) Even in McNamara’s day, the U.S. missiles were never aimed at Soviet cities or population centers per se; they were always aimed mainly at Soviet military targets. Still, the warheads and bombs were so enormous at the time—many delivering an explosive punch of well over a megaton—that tens or hun­dreds of millions of people would have been killed anyway, not to mention the millions more around the world who would have died from radioactive fallout.

The first coordinated U.S. nuclear war plan, known as the SIOP (for Single Integrated Operational Plan), was drawn up at the Strategic Air Command (SAC) in Omaha, Nebraska, in 1960, just before Kennedy was elected. It supplied a rationale for as many bombs and missiles as the military desired. Every remotely valuable facility in the Soviet Union (and in communist China and Eastern Europe) was designated a target, and officers in SAC’s Joint Strategic Target Planning Staff decreed that several particularly valuable targets had to be destroyed with a 90 percent probability, others with a 98 percent probability. Under these rules, several weapons would have to be fired at those targets—and thus the military would have to buy several times as many weapons as might seem reasonable at first glance.


What does the United States need nuclear weapons for? And how many, of what sort, are enough?

In 1961, just after the start of Kennedy’s term, McNamara revised the SIOP to give the president an option to launch “limited” strikes, just against Soviet strategic military targets (ICBMs, submarine pens, and bomber bases), avoiding cities. Still, SAC’s requirements remained enormous. And as the Soviets built up their nuclear arsenal through the 1960s, largely in response to the U.S. buildup, the requirements grew propor­tionately. McNamara’s 1,000-ICBM limit remained in place, so the U.S. military developed missiles tipped with several warheads, each of which could be flung at a separate target. These were known as MIRVs, for “multiple independently targetable reentry vehicles.” When Soviet and American ICBMs were “MIRVed,” they became at once the most lethal weapons and, because of that, the most vulnerable. The sheer existence of these weapons created a new sort of instability: in a crisis, each side might have an incen­tive to launch a first strike with its ICBMs, if just to preempt the other side’s launching a first strike with its ICBMs.

This situation, which theorists dubbed “crisis instability,” spawned a small library of nuclear-exchange sce­narios, replete with deceptively precise calculations. They all envisioned a U.S. president and a Soviet premier firing hundreds or thousands of nuclear war­heads at each other’s country, killing tens or hundreds of millions of citizens, all while maintaining the perspicacity to lob missiles back and forth (assuming, absurdly, that surveillance satellites and data-processing computers would still function well enough to assess the damage)—and that, through this curious chess match, one side or the other would achieve some sort of victory. In retrospect, these books and articles (many of them reviewed and published in journals such as this one) seem bizarre, if not insane.

When the Cold War ended, so did this strange discourse. And so did the nuclear arms race. The U.S. nuclear stockpile, which had peaked in 1967 at an astounding 31,255 weapons, had already been wound down to 19,008 by 1991, mainly due to the dismantlement of tactical nukes in Western Europe and South Korea. This number was cut by half over the next decade and by half again in the decade after. Some of these cuts stemmed from nuclear arms control accords beginning in the early 1970s. But to a much larger extent, they occurred because, starting under President George H. W. Bush and continuing with President Bill Clinton, civilian analysts in the Pentagon wrested control of the SIOP for the first time since McNamara’s revision in 1961.

A cruise missile maneuvers during a Nuclear Weapons System Evaluation Program in Utah, September 2014.
A cruise missile maneuvers during a Nuclear Weapons System Evaluation Program in Utah, September 2014.

Taking a close look at SAC’s beyond-top-secret list of targets and how many weapons were aimed at each, the civilian analysts concluded that it was all wildly inflated: many facilities on the list didn’t need to be targets, and many of the targets didn’t need to be hit with so many weapons. As a result, according to Uncommon Cause, a new book by retired General George Lee Butler, former SAC commander, the nuclear requirements were slashed from 10,000 weapons to 5,888. (The civilians regarded even this many as excessive, but political com­promise prevailed.) The actual number of weapons came down almost propor­tionately and has continued to slide since, although much less steeply.

The most recent (and fairly modest) reductions were the result of the New START treaty. Obama hoped to reach a follow-on accord but never did, a failure that he has attributed to Putin’s return to the Russian presidency. Yet even if Obama’s negotiating partner, the more moderate Dmitry Medvedev, had re­mained Russia’s president, it’s doubtful the two could have concluded a New START II. Obama said at the time that a second treaty should impose cuts not just on long-range missiles but also on short- and medium-range weapons. Yet given the United States’ superiority in conventional arms, no Russian leader would likely have gone down that path. When the Soviet Union and its Warsaw Pact allies outgunned NATO along the border between East and West Germany, U.S. presidents saw nuclear weapons—including missiles, tactical bombs, and even nuclear artillery—as an offset to the military imbalance. Russian leaders similarly see nuclear weapons as offsets today. (The United States maintains just 184 nuclear bombs in Europe, kept in storage and capable of being loaded onto tactical fighter jets, but Russia is estimated to have more than 2,000, many of them deployed.)

Even if Obama’s negotiating partner, the more moderate Dmitry Medvedev, had re­mained Russia’s president, it’s doubtful the two could have concluded a New START II.

Once the nuclear talks hit a dead end, should Obama have proceeded with unilateral reductions? In the realm of nuclear weapons, there is no need, after all, to match an adversary missile for missile, warhead for warhead, or kiloton for kiloton. Presidents must determine what missions they want the nuclear arsenal to accomplish and ensure that, even under pessimistic assumptions, the military has enough weapons to do so. If the calculations indicate that they need, say, 1,000 nuclear weapons, it shouldn’t matter whether Russia or some other hostile country has two, three, or ten times as many. But that’s not how the question has ever played out politically.


Today, the United States has 440 ICBMs, 288 SLBMs (on 14 submarines), and 113 strategic bombers—loaded, all told, with 2,070 nuclear bombs and war­heads, with another 2,508 held in storage as a hedge against a total break­down in international relations and a resumption of a serious arms race. Russia has 307 ICBMs, 176 SLBMs (on 11 submarines), and 70 bombers—with an estimated total of 2,600 bombs and warheads and about 2,400 more in storage. China has 143 ICBMs, 48 SLBMs (on four submarines), and three bombers with the range to hit the western United States—with a total of roughly 180 bombs and warheads. (These figures come from Hans Kristensen and Robert Norris, who have compiled the most complete—and, sources tell me, most reliable—unclassified tabulation of the world’s nuclear arsenals.) 

The Chinese have never played the nuclear arms race game; they follow a policy of “minimum deterrence,” holding just enough to keep two enemies at bay, Russia and the United States. (There are, however, reports of a modest recent buildup, perhaps in response to U.S. progress on a missile defense system.) The Russians, on the other hand, are in the midst of an active modernization program. They have retained 46 of their old MIRVed ICBMs and are developing a replacement for two-thirds of those.

Compared with the Russians, the Americans aren’t doing much beyond replacing old missiles and bombers with new ones—almost none of which will have greater destructive power than the old ones—and Washington isn’t in a rush to do even that. The United States got rid of its MIRVed ICBMs long ago, and a new model (armed with just one warhead each), called the Ground-Based Strategic Deterrent, is not expected to debut until 2028. A new submarine, the SSBN-X, won’t undergo sea trials until 2031. A new aircraft called the Long-Range Strike Bomber is scheduled for the late 2020s, as is the Long-Range Stand-Off weapon, an upgrade of the current air-launched cruise missile.


Ballistic missile submarine USS Tennessee returns to base in Georgia, February 2013.
Ballistic missile submarine USS Tennessee returns to base in Georgia, February 2013.

Even so, the U.S. modernization plan, if carried out in full, will be very expensive: according to the Congressional Budget Office, it will cost about $60 bil­lion for 642 new ICBMs (400 of which will be deployed in silos), $100 billion for 12 new nuclear-missile-carrying submarines, $55 billion for 100 new bombers, $30 billion for 1,000 new cruise missiles, and $50 billion for faster, more flexible, and more secure command-and-control systems—to say nothing of the roughly $80 billion over the next decade to maintain the nuclear laboratories.

The original reason for the nuclear triad was entirely bureaucratic. The army was building ICBMs, the navy was building submarines, and the air force was building bombers. (Eventually, the air force won the contract for ICBMs, leaving the army to field short- and medium-range nukes in Western Europe and South Korea, as well as conduct R & D on antiballistic missiles to defend the continental United States. These army projects have since largely evaporated.) Yet as the arms race took off in the 1960s, the triad took on a strategic rationale. ICBMs were the most powerful and accurate of the three legs, the ideal tool for promptly attacking blast-hardened Soviet missile silos. SLBMs were much less powerful and less accurate, but they were the most secure, since they were loaded onto submarines, which prowled the oceans undetected. If deterrence was defined as the survivability of a second-strike force, SLBMs were crucial. Bombers were slower, taking hours, not minutes, to reach their targets, but because of that, their pilots could linger outside Soviet borders awaiting instructions or even be called back to base if a crisis was resolved.

The case for land-based ICBMs today is extremely weak and has been since 1990, when the U.S. Navy started deploying Trident II missiles on submarines. Unlike earlier SLBMs, the Trident II is accurate enough to destroy blast-hardened missile silos. In other words, one of the ICBM’s unique properties—its ability to hit blast-hardened targets quickly—is no longer unique. Meanwhile, its other unique property—its vulner­ability to an adversary’s first strike—is all too enduring. Even by the esoteric logic of nuclear strategists, then, ICBMs make the United States less secure, with no compensating advantages.

During the Cold War, a case was made for ICBMs on the grounds that the United States needed to keep up with whatever the Soviets were doing, if just to demonstrate “resolve” and “credibility.” A dubious notion at the time, it makes no sense whatever today. So what is the rationale for preserving, much less modernizing, land-based ICBMs? If they are inherently destabilizing, why shouldn’t they be dismantled? 

There is no need to match an adversary missile for missile, warhead for warhead, or kiloton for kiloton.

The air force and some civilian strategists do have a new rationale. They call it “the sponge theory.” Without land-based ICBMs to soak up Russian missiles, the logic goes, there would be only six strategic targets in the continental United States: the three bomber bases (in Louisiana, Missouri, and North Dakota), two submarine ports (in Bangor, Washington, and Kings Bay, Georgia), and the National Command Authority (otherwise known as Washington, D.C.). The Russians could launch an attack on those six targets with just two MIRVed missiles, or possibly even one. An American president might not strike back, knowing that the Russians had thousands of warheads remaining and that, therefore, if the United States retaliated, Russia would retaliate further. (In the 1970s, some hawkish strategists outlined a similar scenario, which they called “deterring our deterrent.”) On the other hand, if the United States maintained its 400 ICBMs, so the theory goes, the Russians would have to fire at least 400 warheads to destroy them. By any definition, this would constitute a major attack, thus prompting certain, massive retaliation—the prospect of which would deter them from launching an attack in the first place.

A USAF B2 Spirit Bomber taxis before take-off from Whiteman Airforce Base, Missouri, October 2001.

This is a strange theory. It assumes that a U.S. president would tolerate a nuclear attack that, despite its supposedly limited scope, would kill hundreds of thousands, possibly millions, of civilians and not strike back with any of the hundreds of missiles poised on invulnerable sub­marines. But let’s posit that the theory has some validity—that the military needs to have more than six targets on U.S. soil to act as sponges to absorb a Russian nuclear attack. Even so, it’s a huge stretch to contend that the United States needs 400 ICBMs—400 sponges—for that purpose. How many would be needed? A dozen? Two dozen? Almost certainly no more than that. Nor would those missiles (again, assuming the theory makes a lick of sense) have to be up-to-date models. In other words, a strong case can be made that the United States does not need a new land-based ICBM at all.

Does it need a new long-range manned bomber that can penetrate Russian air defenses? The case for this, too, is far from persuasive. With a range of 1,500 miles, air-launched cruise missiles can be fired from bombers loitering well outside the reach of Russian air defenses. Does this mean a new air-launched cruise missile is a good idea? It might be, except that the air force’s Long-Range Stand-Off missile is not simply a replacement. This is the one new weapon in the U.S. mod­ernization plan that is a completely new design. It has a longer range and better accuracy, neither of which is controversial, but it’s also built to carry either a conventional or a nuclear war­head, and that feature has raised concerns. Former Defense Secretary William Perry has argued that if the Long-Range Stand-Off weapon were launched, the leader of the country being attacked would have no way of knowing which kind of warhead the missile was carrying. Assuming the worst (as leaders tend to do in wartime), he or she might assume it was carrying a nuke and respond accordingly.

Deterrence doesn’t require a lot of nuclear weapons: in some cases, a mere handful; in others, a couple of hundred.

As for a new class of nuclear-missile-carrying submarines, objections here are hard to muster. The current Ohio-class submarines are between 20 and 40 years old. At some point, subs get too old to risk plunging them into the ocean for another voyage. Does the navy need 12 new submarines? Would, as some suggest, eight to ten be enough? Any calculus has to take into account that some subs have to be in the Pacific Ocean, some have to be in the Atlantic Ocean, and some have to be in port for refueling and repairs. If one can safely assume that three are enough for each location, then nine should be fine. If that’s considered excessive, then eight would do; if that’s deemed too risky, then ten would do. The rationale for 12 is hard to parse.

Finally, the program for more rugged command-and-control systems is unassailable, given the growing vul­ner­ability of satellite communi­cations, either to a direct attack or to hacking. If the United States is stuck with nuclear weapons, the president should at least appear to have the ability to launch them, or withhold them, preferably in a controlled way.


The fact is that short of a transformation in world politics, which doesn’t seem remotely in the offing, we are—all of us—stuck with nuclear weapons. They do serve a function: they tend to deter aggression, not just nuclear attacks but conventional invasions, as well. During a mid-1960s border dispute, the Kremlin considered invading a patch of China but held back because China had a handful of nuclear weapons. Israelis fear an Iranian bomb not so much because they think the supreme leader might suddenly launch nuclear missiles at Tel Aviv (Israel is estimated to have roughly 200 nuclear weapons and could easily incin­erate Tehran in response) but because a pocketful of nukes could provide cover to other forms of Iranian aggression.

A possible parallel: if Israel hadn’t destroyed Iraq’s Osirak nuclear reactor in 1981, and if Saddam Hussein had gone on to build a few atomic or hydrogen bombs, it’s hard to imagine so many Western and Arab nations joining the coalition to oust the Iraqi army from Kuwait a decade later. The United States might have offered to extend its nuclear umbrella to the coalition partners, pledging to retaliate against an Iraqi nuclear attack, but the Arab countries, in particular, might not have trusted it. In the 1960s, British and French leaders, questioning the credibility of NATO, asked whether the United States would risk New York to protect London or Paris; Egyptian and Saudi rulers would rightly have doubted whether it would do the same for Cairo or Riyadh. But as these scenarios illustrate, deterrence doesn’t require a lot of nuclear weapons: in some cases, a mere handful; in others, a couple of hundred.


An earlier version of this article mistakenly stated that no civilian officials had scrutinized the U.S. nuclear war plan since the early 1990s. In fact, such a review was carried out under President Barack Obama, although he seems to have ignored its conclusions. It has been corrected and updated.