There are many reasons why fusion energy—derived from the fusion of atomic nuclei, rather than the fission used in today’s nuclear power plants—is desirable as a replacement for fossil fuels in a low-carbon economy: Fusion’s basic fuels are abundant and easily available to all nations. And compared with nuclear fission, it is intrinsically safer; the radiological waste is much less long-lived, and the risk of nuclear proliferation is much less. Fusion can be a dispatchable complement to daily and seasonally variable renewables such as solar and wind. With care and sufficient oversight (full disclosure: this is an area of the author’s research), its risks can be effectively managed.

To realize fusion’s potential, however, it must pass five different technical goal posts, namely power density, gain, plasma power handling, blankets, and magnets and drivers. Here is a simplified (that is, non-mathematical) description of these goal posts.

Power density. The first requirement of a fusion system is that it must project to a high power output at a reasonable volume— i.e., have a high enough power density. Nuclei (the cores of atoms) repel each other because they are positively charged. High temperatures are needed to overcome this repulsion, so that the nuclei can fuse together rapidly. This is why you hear that fusion requires what is known as a plasma—a gas so hot that the electrons and nuclei are torn away from each other.

Table 1 shows the different fusion fuels being pursued across the fusion landscape, their optimal temperatures in degrees Celsius, and their maximum power densities in megawatts per cubic meter, or MW/m³. (A megawatt, or a million watts, measures power, while m³ or cubic meters, measures the volume of the fusion system). Power density is evaluated at a pressure 10 times that of our atmosphere—a pressure of “10 atmospheres.” Beyond the question of temperature, if you can double the pressure and still stay at the optimal temperature, you can quadruple the fusion power density, so pressure matters a lot. But as nature would have it, pressure is a hard-won quantity in a fusion plasma.

Gain. Even with a system that promises to deliver a favorable power density, it must also have high “gain”—the fusion power produced by the fuel, compared with the power that is needed to keep it hot. This is obviously very important; if you have a reasonably sized fusion power plant that needs to consume all the power the plasma makes just to keep running, then it is useless for energy production. The critical parameter for fusion gain is the fuel pressure multiplied by the “energy confinement time,” or the amount of time that the energy in the fuel is held in before it leaks out. This “gain parameter” is measured in atmospheres of pressure times energy confinement time in seconds, or atmosphere-seconds. The kicker is that the reactions with lower power densities for a given pressure need proportionally higher values of the gain parameter. So the third column in Table 1 gets you twice: once for power density and once again for gain.

 Plasma power handling. The plasma power being lost from the fuel will eventually impinge on the inner surfaces of the container holding the plasma. The power deposited on these surfaces can be extraordinarily high. There are innovative approaches for mitigating this problem (full disclosure: this is also an area of the author’s research). But this issue is also very sensitive to the configuration of the fusion system, as will be seen when exploring the landscape of fusion. Plasma power can also be released in transient pulses in an otherwise steady system, which increases this challenge. Pulsed systems, like laser fusion, intrinsically have pulsed releases of energy.

Blankets. The fusion reaction with the highest power density for a given pressure, by far, is the reaction between the two isotopes of hydrogen, deuterium (D) and tritium (T). Deuterium is plentiful in seawater, but tritium needs to be produced in the fusion blanket which surrounds a fusion reactor, which is bombarded by neutrons emanating from the deuterium-tritium fusion reaction. Consequently, the easiest reaction from which to produce fusion energy has the most complex fusion blanket. This requires development of improved materials to resist damage from highly energetic neutrons, as well as systems for breeding tritium by the reaction of the neutrons with lithium, extracting the tritium from the blanket, and recycling it into the plasma. The quantity of tritium is considerable: about150 grams of tritium are burned per day in a plant producing perhaps up to 400 megawatts of electrical power.

Magnets and drivers. Outside the blanket, magnetic confinement fusion systems generally require large and powerful magnets to produce the magnetic fields that confine the plasma, and these magnets an area of focus for a number of fusion companies. Generally speaking, the higher the magnetic field the better, because, in an ideal world, doubling the magnetic field increases the plasma pressure that can be attained by a factor of four—and so the fusion power density by a factor of 16! On the other hand, the forces on the magnets are large, and there is a limit to the forces that can be sustained.

In addition to magnetic confinement, some researchers are pursuing inertial confinement fusion, which is pulsed by definition. In inertial confinement fusion, power must be delivered to a small target in a very short time. Lasers have been used for this to date, but to be used in a commercial power plant, lasers need higher efficiency and improved repetition rate. There are also concepts being pursued in which the power is delivered by pulsed electrical currents (think lightning bolts), and here again efficiency and repetition rate need development. An additional consideration for inertial confinement fusion: The tiny fusion targets need to be manufactured precisely, and then positioned precisely, compressed, and heated to fusion temperature—perhaps as often as 10 times per second.

Fusion’s wide research portfolio

 Each of the concepts being pursued to achieve the overall goal of commercial fusion energy has advantages and challenges. Some concepts use well-established physics, and some promise simpler or more cost-effective fusion systems. The concepts can be characterized as falling under two general approaches: magnetic confinement and inertial confinement.

To provide some context, the maximum gain parameter is reported for each concept (Wurzel and Hsu 2022) along with the cumulative private financing for the concept (Fusion Industry Association 2024). A US Department of Energy peer-reviewed competition was used as the basis to report the number of companies participating in a milestone-based public-private partnerships to develop the preliminary design of a fusion pilot plant (US Department of Energy 2023).

Magnetic confinement fusion

Tokamak. Plasma spirals along magnetic field lines, and if those magnetic field lines are curved around into a torus—or doughnut—configuration, then these magnetic field lines can be made to circulate the torus indefinitely without striking a material surface. Early on, researchers realized that to confine plasma in a torus, the magnetic field lines need to circle it not only in the long direction but also in the shorter direction. One way to cause the magnetic field to circulate in the shorter direction is to induce a strong electrical current, traveling the long way around the torus. This was tried at Los Alamos soon after the Manhattan project, in a small whimsically named device, the “Perhapsatron”—which proved wildly unstable, thrashing about and slamming into the chamber walls.

The international ITER project, under construction in southern France, uses a tokamak design, first studied by the Soviets, in which the magnetic field is much stronger the long way around the torus than the short way, giving it greatly enhanced stability compared with the “Perhapsatron.” ITER is designed to address each of the major goal posts for fusion energy. Key to this will be the high total energy stored in the ITER plasma and in its induced magnetic fields, and the long pulses available on ITER, up to about one hour. These will enable the demonstration of the sustainment of high power density and high gain over extended periods of time. These pulses will also enable the study of strategies for handling high steady and transient power loads, the performance of the blankets surrounding the plasma, and the stability of magnets over extended periods of time.

The highest gain parameter reported in large tokamaks is 1.3 atmosphere-seconds. The cumulative private financing for tokamaks is over $2.1 billion at two companies; the US Milestone-Based Fusion Development Program includes one company pursuing the tokamak concept.

Stellarator. While the tokamak was being developed in the Soviet Union, the United States was pursuing a concept called the stellarator. In this case the coils that form the magnetic fields are twisted or tilted in such a way as to coax the magnetic field lines around the torus in the short direction, as they mainly travel around the long way. Continuing the comparison with foods, the stellarator looks more like a cruller than a doughnut. The stellarator has been successful in providing additional stability compared with the tokamak. Results from stellarators have been promising, but they have not yet achieved the gain parameter of the large tokamaks. The stellarator gives every indication of being an optimal device for machine-learning based optimization, since the plasma’s characteristics are very much in the hands of the designer, and the possible designs are many.

The highest gain parameter reported in stellarators is 0.10 atmosphere-seconds. Cumulative private investment in stellarators is approximately $190 million at eight companies; the US Milestone-Based Fusion Development Program includes two companies pursuing the stellarator concept.

Spherical tokamak. The spherical tokamak is essentially a tokamak in which the torus has a relatively small hole through the middle. As foods go, this resembles more a cored apple than a doughnut, but it inherits the basic stability and confinement properties of the tokamak in a potentially more compact system. There are indications that the spherical tokamak will improve in gain with increasing temperature.

The highest gain reported in spherical tokamak experiments, is 7.7 10^-3, or 0.0077 atmosphere-seconds. The cumulative private investment in spherical tokamaks is about $770 million at three companies, and the US Milestone-Based Fusion Development Program includes one company pursuing the spherical tokamak concept.

Field-reversed configuration. The tokamak, spherical tokamak, and especially the stellarator take strong external control of the plasma, but at a cost in terms of complexity and expense of the magnet systems. An alternative is not to produce a magnetic field the long way around the torus altogether; the current in the plasma still sustains a magnetic field that goes around the torus the short way. This is called the field-reversed configuration. If you do not require a magnetic field the long way around the torus, you don’t need magnetic coils to link through the center of the torus. This allows for a great simplification of the magnet system. Furthermore, the heat from the edge of the plasma can escape the magnet coils altogether, so that the lost heat flux can be taken, in principle, on a much larger and more accessible area.

Such a configuration is quite attractive as a fusion concept. The problem is that, in theory and in practice, it is quite unstable. There are demonstrated techniques to stabilize the configuration, in part by driving rotation in the outer layers of the plasma, but it is not clear yet for how long a time this will work, and if it will be viable on the larger devices that may be required to attain the necessary gain parameter for a commercial power plant. The published values of achieved gain parameter for steady field-reversed configurations are far below those of tokamaks. This problem is compounded by the fact that the companies that are pursuing the field-reversed configuation (pulsed and steady) plan to use fusion reactions that require a much higher gain parameter than the deuterium-tritium reaction: deuterium-helium 3 and proton-boron (p¹¹b). On the other hand, if these fuels can be successfully brought to sufficient power density and gain, they provide the further attraction of simplified blankets with no need to breed tritium, and no need for advanced materials beyond those used in some fission reactors. Even if the system has to fall back on deuterium-tritium fusion, the magnet and blanket configuration is much simplified compared with tokamaks or stellarators.

The highest gain parameter reported is 3.7 10^-4, or 0.00037 atmosphere-seconds (Wurzel and Hsu did not report published results from repetitively pulsed field-reversed configurations). The cumulative private investment in these configurations is about $1.8 billion at three companies. The US public-private partnership includes no field-reversed configuration concepts—but it is unknown if the companies pursuing this concept applied to participate.

Flow-stabilized Z-pinch. The flow-stabilized Z-pinch dispenses with all magnetic field coils, and indeed the toroidal configuration altogether. It drives a strong electrical current from one electrode to another, making its own high magnetic field to confine the plasma, heating it by the resistive dissipation of the current. This is like the heating in your toaster—if your toaster’s coils were made of plasma and carried a current similar to a lightning bolt. This configuration is again stabilized by flows, but there are questions about how the flow will be maintained in power-plant conditions. Nonetheless, this approach is very favorable from the point of view of designing a fusion power plant—although the electrodes will suffer significant wear and tear. The absence of coils and lasers and the efficiency of the heating and compression process simplify the device dramatically. It presently has a very low gain parameter, but if this can be enhanced at higher currents, the ultimate power plant could be relatively small in unit size, with rather a simple blanket, even if it runs on deuterium-tritium.

The highest gain reported in Z-pinch experiments is 3.4 10^-4, or 0.00034 atmosphere-seconds. The cumulative private investment in Z-pinches is about $225 million at two companies, and the US Milestone-Based Fusion Development Program includes one company pursuing the flow-stabilized Z-pinch concept.

Magnetic mirror. Another alternative to wrapping magnetic field lines into a torus is to plug long, straight magnetic fields at each end. This can be achieved, to a degree, by increasing the magnetic field at the ends. The increasing field at the ends pushes plasma back towards the center. But plasma particles with their energy mostly directed along the field can still squeak through. This puts a limit to how long the heat can be contained, and so limits the gain parameter. There are techniques—such as making a long, simple mirror, with separate mirror cells at the two ends—that have mirror losses, but with much less plasma to lose. Furthermore, the advent of higher magnetic fields allows for the plugging at the ends to be increased. This configuration allows the heat flux from the plasma to escape the ends and permits a blanket with a simple geometry. The gain achieved to date is modest, but the theoretical basis for this system is relatively well developed.

The highest gain reported in magnetic mirror experiments, is 4.8 10^-6,  or 0.0000048 atmosphere-seconds. The cumulative private investment in magnetic mirrors is about $14 million at two companies, and the US Milestone-Based Fusion Development Program includes one company pursuing the magnetic mirror concept.

Inertial confinement fusion

Indirect-drive laser fusion.The fusion concept with the highest gain to date is indirect-drive laser fusion, in which 5.2 megajoules of fusion energy has been produced for 2.2 megajoules of input laser energy to the target, at the National Ignition Facility, or NIF. (One megajoule is equivalent to one million watts for one second.) This is indeed a great accomplishment. Indirect-drive laser fusion proceeds by committing laser energy to heating the inner wall of a small cylinder coated with heavy metal, gold in the case of the NIF experiments. The cylinder radiates x-rays, which rapidly compress a small diamond capsule containing a thin layer of frozen deuterium-tritium, with a small mass of deuterium-tritium gas in the center. As the capsule is driven inwards it compresses the deuterium-tritium gas, heating it to fusion temperature—while the original frozen deuterium-tritium shell is much denser and colder.

The success of the NIF experiment was that this central, small mass “hot spot” achieved the necessary temperature and gain to produce enough fusion energy that a burn could propagate from the hot spot into the much larger mass of colder fuel. This process gave the dramatic result of a net target gain during the brief time that the fuel was held together by its own inertia.

To build upon this success to commercial fusion energy will still require passing the goal posts previously discussed. It currently takes weeks to months to manufacture the very precise targets used in NIF, and hours to days to align them in front of the very precisely aimed laser beams. In a fusion power system, this will need to take place about 10 times per second to achieve the required time-averaged power density. Because of the intrinsic inefficiencies of lasers, the overall gain must be increased from 2.4 to 80 or so, requiring a much smaller fraction of hot-spot fuel compared with frozen fuel. Power handling will be challenging, since the shock to the walls of the fusion chamber containing the target will likely ablate wall material, which must be cleared away in time for the next target to be injected into the system.

The blanket system for such a laser-driven system, on the other hand, may be simpler than for magnetic fusion, since it is not contained in a magnet. The tritium handling systems may also be simpler, as typically more of the injected fuel is burned in laser-fusion systems than in magnetic fusion systems. The laser drivers, outside of the blanket, need to have efficiency in the range of 15 percent or higher, and need to be able to work repeatedly about 10 times per second. The very precisely manufactured targets, their delivery system, and the precise engagement of the laser beams required for commercialization of such a power plant will be challenging to achieve.

The gain parameter does not play exactly the same role in inertial confinement fusion as in magnetic confinement fusion, but it is a key measure of the approach to the propagating burn that was achieved on NIF.No private investment was reported specifically for indirect-drive laser fusion, and the US Milestone-Based Fusion Development Program includes no companies pursuing indirect-drive laser fusion. It is unknown if companies pursuing this concept applied to participate.

Direct-Drive Laser Fusion. A variant on the indirect-drive approach used by the NIF is to target the deuterium-tritium capsule directly with the laser beams. This has an immediate advantage: Energy is not lost by the production of x-rays on the cylinder walls that do not happen to impinge on the target capsule. This, in turn, reduces the ratio of colder fuel to the hot spot that is needed for sufficient gain, easing the requirements on the implosion. However, there is a trade-off: The lasers, hitting the capsule directly, must be very precisely distributed, since x-rays from an external cylinder will not smooth their impact. Furthermore, the lasers directly impinging on the surface of the capsule can interact with the plasma blowing off its surface in ways that are disadvantageous. In other aspects, direct-drive laser fusion largely inherits the advantages and challenges of indirect-drive laser fusion, including the positive success in demonstrating a propagating burn.

Cumulative private investment in direct-drive laser fusion is about $110 million at three companies, while the US Milestone-Based Fusion Development Program includes one company pursuing the direct-drive laser fusion approach.

Fast-ignition laser fusion. In this concept, lasers are used to compress the cold fuel, but there is no requirement for a core hot spot, nor for very precise symmetry. Instead, a very short and powerful separate laser pulse impinges on the surface of the compressed fuel, heating a small volume rapidly to fusion temperature. This externally created hot spot drives the propagating burn in the bulk of the fuel. There has been less research done in this area than in direct-drive laser fusion, and much less than in indirect drive. But it promises to provide an alternative way to produce a low-mass hot spot to propagate the fusion burn.

Cumulative private investment in fast-ignition laser fusion is about $310 million at two companies, while the US Milestone-Based Fusion Development Program includes one company pursuing the fast-ignition laser fusion approach.

Magnetized-liner inertial fusion. In this approach, a small, cylindrical metallic shell is pre-loaded with a magnetic field and plasma. Then a powerful current is driven through the shell, causing it to compress rapidly—but not as rapidly as in laser fusion. Since this process is slower, it is necessary to provide some of the confinement of the plasma using the preloaded, and subsequently compressed, magnetic field, making this process inherit some of the characteristics of both inertial and magnetic fusion. Like with the Z-pinch, this configuration is unstable, but rather than stabilize it with flows, the approach is to choose conditions where the instability grows slowly enough that the fusion yield occurs before the instability disrupts the plasma. Because of the efficiency of the heating method compared with laser-driven compression, this system might—in principle—be able to provide the necessary gain without a hot-spot/cold-fuel configuration as used for laser fusion. However, projections do call for frozen deuterium-tritium fuel between the plasma and the metallic liner, to support a propagating burn.

The highest gain reported in magnetized linear fusion experiments is an impressive 0.483 atmosphere-seconds. The cumulative private investment in magnetized-liner inertial fusion is about $18 million at one company, and the US Milestone-Based Fusion Development Program includes no companies pursuing magnetized-liner inertial fusion, but it is unknown if companies pursuing this concept applied to participate.

The future?

This is an exciting time for fusion energy. There is strong market pull, driven by concerns about climate change, and a wide range of concepts under investigation, constituting strong technology push. Some concepts have a stronger scientific basis, and some have a more attractive potential for commercialization. Partnership between private enterprise and publicly funded research institutions, including both universities and national laboratories, will be needed to drive fusion development forward more rapidly than in the past. Rapid progress will require both a sound basis in science and technology and a keen drive for a practical new, low-carbon energy source that can effectively complement variable wind and solar power to combat climate change.