Each experiment at the National Ignition Facility (NIF) in California—a “shot”—lasts just a few billionths of a second. A lot happens in that brief moment, however: 192 laser beams, totalling some 500trn watts, converge in the machine’s target chamber and dump their energy onto a gold cylinder, which is just a few centimetres long. Inside the cylinder is a peppercorn-size diamond sphere filled with a mixture of deuterium and tritium, heavy isotopes of hydrogen.

As the sphere absorbs the laser’s energy, its outer layers rapidly ablate away. That creates a shock wave travelling at 300km per second that implodes the sphere’s insides. As the atoms of deuterium and tritium are pushed together at billions of times atmospheric pressure, their temperatures exceeding 100m°C, they start fusing into helium, releasing vast amounts of energy.

This is the kit you need to be able to re-create a nuclear-weapon explosion without actually setting off a bomb. NIF was conceived in the 1990s, a few years after America decided to stop testing its nuclear arsenal in underground explosive tests. Without these tests, the people responsible for the country’s nuclear deterrent still needed ways to guarantee the safety of their warheads as they sat in storage and, most important, instil confidence that they would perform as intended, if they were ever called upon.

The facilities that America’s nuclear establishment developed to answer that challenge eventually included NIF, the world’s most powerful laser, and El Capitan, its fastest and most capable supercomputer. Both have become central to a renewed mission for America’s nuclear-weapons labs, as they upgrade their existing bombs and, for the first time in decades, design brand new ones.

Maintaining nuclear weapons takes an army of scientists and engineers. NIF is part of the Lawrence Livermore National Laboratory near San Francisco, set up in 1952 as a rival to the Los Alamos National Laboratory in New Mexico. It was at Los Alamos that the first nuclear bombs were built less than a decade before. “We were developing this advanced technology in a very classified environment,” says Kim Budil, Livermore’s boss. “It was really important to bring scientific rigour, peer review and competition to that technology race.” The two labs purposefully pursue different designs for weapons and, though they sometimes collaborate, refer to each other as “competimates”.

Livermore and Los Alamos design the “physics packages” in America’s warheads, which is to say the nuclear bits of the nuclear bombs. A third institution, Sandia National Laboratories, adds the non-nuclear components (such as triggers, batteries, sensors and radiation-hardened electronics) and integrates the devices made by the two physics labs with the delivery systems (eg, missiles) that turn them into robust, deployable weapons. All told, the three labs of the National Nuclear Security Administration (NNSA) employ tens of thousands of scientists and engineers. All three granted The Economist rare access to their researchers and some of their facilities.

When Livermore opened, one of its primary goals was to accelerate the development of hydrogen, or thermonuclear, bombs. Unlike the fission bombs that had been developed in the Manhattan Project, which released energy by splitting atoms of heavy elements (uranium and plutonium), thermonuclear bombs were designed to release energy by fusing atoms of deuterium and tritium, some of the lightest in existence. (These bombs are called thermonuclear because they have two stages: first, a fission bomb made of plutonium which creates an intense burst of heat; that then ignites a second stage in which the fusion occurs.)

Thermonuclear technology opened the door to more powerful but also more compact weapons. In the 1950s, when the US Navy decided to create a sea-based nuclear deterrent, Livermore was assigned the task of miniaturising nuclear bombs so that they could be affixed to missiles that fit inside submarines. It took them less than four years to come up with Polaris, a missile system an order of magnitude smaller than anything that had come before and which Dr Budil proudly describes as “the single most important technology change in the history of nuclear weapons.”

Small, compact thermonuclear devices became the workhorse of both the American and the Soviet nuclear arsenals as they were expanded during the cold war. Fortunately, none of these weapons was ever used in anger and, decades after being built, thousands remain in their stockpiles.

One of the biggest tasks occupying the scientists today at the Los Alamos, Livermore and Sandia labs is to keep a close watch on those warheads. “A nuclear weapon sitting on the shelf is sort of like a chemistry experiment cooking along year after year,” says Dr Budil. “Things are changing. Radioactive materials decay over time. Polymer materials degrade.”

Every year a few devices are taken apart and thoroughly examined. More extreme testing also happens. Microscopic samples of material are placed inside NIF’s target chamber, where they can be imaged by X-rays while experiencing the equivalent of a nuclear blast. At Sandia, the Z machine is another way to approximate the core of a nuclear blast, but using intense electromagnetic fields rather than lasers. At Los Alamos, by contrast, the non-nuclear parts of the weapons are blasted by shock waves from the conventional explosives that are used to initiate a nuclear bomb.

All that experimental work is used to better understand the properties of materials that go into bombs. And, alongside the thousand or so full-scale nuclear-weapons tests carried out before 1992, the data are also used to build better computer simulations of nuclear blasts. These are now so good that Thom Mason, director of Los Alamos, reckons that scientists have a better understanding of how nuclear weapons work today than they did during the explosive-testing era. “The modern scientific tools really outstrip significantly anything that we had in the 1990s,” he says.

Exactly how much better is demonstrated at Livermore’s computing centre, a few minutes’ walk from NIF. In January, scientists and government officials gathered there to unveil the NNSA’s latest (and now the world’s most powerful) supercomputer—El Capitan. This machine can run a quintillion (10¹⁸) floating-point operations (a measure of calculations) per second. That is around 100m times faster than a typical laptop, and makes it only the third ever exascale computer (“exa” being the measurement prefix for 1 followed by 18 zeros). Its roughly 90 refrigerator-size racks of processors are densely packed over the same space as a couple of tennis courts.

The supercomputer is part of the Advanced Simulation and Computing (ASC) programme, started in 1995, alongside NIF, as part of America’s response to its moratorium on nuclear-weapons testing. One of its first goals, set for the turn of the millennium, was to assemble the hardware and software required to run a three-dimensional simulation of a weapon system.

Scientists overcame the enormous challenges using the parallel-computing architecture that was becoming possible at the time. This meant splitting up a simulation into small chunks that could be run simultaneously across the central-processing units (CPUs) and graphics-processing units (GPUs) found in high-end computers. It still took months to run a single simulation. “On El Capitan, we’re now estimating we could be able to run upwards of 200 of those in a day,” says Rob Neely, Livermore’s associate director for weapon simulation and computing. And all that at much higher resolution too.

Look closer at the processors and something else becomes apparent. Instead of CPUs and GPUs, El Capitan uses specialised chips developed for Livermore by Advanced Micro Devices, a chip designer, called accelerated-processing units (APUs). Typically GPUs and CPUs will have their own storage and memory and the communication between them, known as the bus, can become a bottleneck to a system’s speed. Each APU is, instead, a single piece of silicon with sections (“chiplets”) that individually operate as CPUs or GPUs, allowing them to share memory and storage. “It’s the only architecture in the world right now that we know of that’s doing it this way,” says Dr Neely.

The density and architecture of those APUs give El Capitan its edge over machines that might, on paper, have more raw computing power. At Los Alamos, the simulations are also being deployed for a new task—designing a new weapon from scratch. The W93, as it is called, will eventually be used on ballistic missiles deployed by the US Navy’s new Columbia-class submarines. It is the first new weapon in the American nuclear arsenal since the 1980s and, with explosive tests off-limits, Los Alamos will need to run simulations from the very start of the design process. El Capitan will allow scientists to optimise the design, says Dr Neely.

The W93 is emblematic of the renewed energy at Los Alamos. “Our budget has roughly doubled over the past five or six years,” says Dr Mason. That means thousands more scientists, modernised facilities and a restored ability to make plutonium pits, a core element of modern thermonuclear bombs. And, in contrast to many other areas of scientific research in America today, the budget for the NNSA is not expecting any cuts in federal funding.

All this is a response to what Dr Mason calls the “fourth age” of nuclear weapons. The first was the invention of nuclear bombs during the Manhattan Project; the second was the cold-war race to build up nuclear arsenals; and the third age was the period after the fall of the Soviet Union during which it was thought that nuclear deterrence would have a declining role in world affairs. The fourth nuclear age is a worrying time featuring the breakdown of arms control, Russia’s threats of nuclear use, China’s rapid build-up and tensions among other nuclear powers such as India and Pakistan. There is also uncertainty over new and would-be nuclear powers, and the risk that America’s allies could develop their own nuclear weapons as they lose faith in its protective umbrella. “It’s clear that deterrence is, once again, pretty important,” says Dr Mason.

Though the primary purpose of the labs at Los Alamos and Livermore is never in doubt, their scientists are keen to point out that these facilities can do much more than national-security work. NIF, for example, is a leading laboratory in the attempt to create power from nuclear fusion.

In December 2022 NIF made good on the “I” in its name and became the first site in the world to achieve ignition—releasing more energy from fusion than had been used to get it going. Since then the scientists there have achieved ignition on eight more occasions, gradually increasing the energy yielded each time.

Mark Herrmann, programme director for weapons physics at Livermore and a former director of NIF, is well aware that it will take a lot more work to turn these breakthroughs into a viable source of energy. For a start, the lasers themselves have to get a lot more energy-efficient and the fusion reactions would need to happen dozens of times per second (rather than just a dozen times per week). Although more engineering work is needed, says Dr Herrmann, “There are no scientific obstacles to those things happening.”

It’s the weapons, though, that these labs exist for. And their terrifying power is never far from the minds and motivations of the scientists involved. When asked how he and his colleagues feel in their role developing nuclear bombs, Dr Mason points to the (albeit occasionally uneasy) geopolitical order that has been maintained as a result of people’s fear of their power. “If the weapons we design are never used,” he says, “we will have been successful.”■