Fusion ignition has been achieved. The race to commercial power plants is now a race to build capacitor banks at a scale and performance level that does not yet exist.
In December 2022, the National Ignition Facility at Lawrence Livermore National Laboratory made history: 192 laser beams compressed a tiny fuel capsule and produced more fusion energy than the laser delivered. NIF proved the physics. It also revealed the engineering bottleneck. NIF’s capacitor banks store roughly 400 MJ just to produce those laser photons, and the facility fires about once per day. The path from ignition to a power plant runs directly through the capacitor bank.
Three years later, roughly $2 billion in private capital has converged around inertial confinement fusion (ICF). Companies like Pacific Fusion, Fuse Energy Technologies, Xcimer Energy, Inertia Enterprises, and Focused Energy are designing machines that will require tens of thousands to hundreds of thousands of capacitors per facility. An IEEE pre-roadmapping paper co-authored by researchers from Sandia, LLNL, the Naval Research Laboratory, General Atomics, and the University of New Mexico identifies energy storage, high-voltage switching, and advanced dielectrics as the key technologies that currently limit the advancement of fusion power.
The capacitor is the component manufactured in the highest volume in any fusion machine, and the dielectric film inside it is the material manufactured in the highest volume inside every capacitor. The fusion story is typically told through lasers, plasmas, and magnetic fields. That framing is incomplete. Understanding the capacitor supply chain challenge reveals why materials science has become the gating factor for commercial fusion energy.
The Architecture Shift That Multiplied Capacitor Demand
The Impedance-Matched Marx Generator (IMG) represents a fundamental shift in pulsed power architecture. Traditional Marx generators charge capacitors in parallel and discharge them in series to multiply voltage. Sandia’s Z Machine, the world’s most powerful X-ray source, uses this approach. But traditional designs require multiple pulse-compression stages between the capacitor bank and the target, each dissipating energy. The Z Machine delivers only about 8% of its stored energy to the imploding liner.
IMGs eliminate most of those compression stages. Their building block is the “brick”: two small capacitors and a single switch that fire sequentially, launching a coherent electromagnetic wave along an internal transmission line. The result is roughly 90% energy delivery efficiency and 6× lower stored energy requirements. The same architecture that improves efficiency also multiplies the number of capacitors each machine needs.
The Z Machine uses 2,160 large capacitors. IMG-based machines use tens of thousands of smaller ones. Pacific Fusion’s 156-module demonstration system contains approximately 50,000. Sandia’s conceptual Jupiter machine specifies 352,800. The architecture that makes commercial fusion plausible also multiplies component demand by two to three orders of magnitude.
The Machines Creating That Demand
The ICF landscape now spans multiple architectures, each generating substantial capacitor requirements.
Pacific Fusion has raised approximately $900 million and is constructing the largest IMG-based machine yet: 156 pulser modules arranged spherically around a fusion chamber. In February 2026, Pacific Fusion and Sandia demonstrated self-premagnetizing MagLIF targets that eliminate the need for external coils. The company targets net facility energy gain by 2030, with each machine requiring approximately 50,000 capacitors that must survive millions of charge-discharge cycles.
Fuse Energy Technologies built TITAN, a 14-stage IMG whose 6-stage testbed achieved 330 GW peak power. Fuse plans Z-Star (approximately 2027): 16 TITAN modules delivering 12+ MA for MagLIF experiments. Each scale-up multiplies the capacitor count and tightens requirements on shot life and thermal endurance.
Xcimer Energy ($111M raised; DOE Milestone Program awardee) pursues a different driver architecture using electron-beam-pumped excimer lasers, in which Marx generators charge pulse-forming lines. Xcimer completed its first private-sector KrF laser in June 2025 and targets a 400 MWe pilot plant by 2035.
Inertia Enterprises ($450M Series A, February 2026) is building a 10 MJ, 10 Hz diode-pumped laser system with roughly 1,000 modular laser units, the most direct commercialization of NIF’s proven physics. Co-founded by NIF ignition lead designer Annie Kritcher, the company has licensed nearly 200 LLNL patents.
Focused Energy ($175M+ raised; DOE Milestone Program awardee) pursues proton fast ignition, separating compression and ignition phases to reduce laser energy requirements by approximately 4× compared to NIF. The company has partnered with Amplitude (France) in a $40M laser development agreement.
The pattern is consistent: whether the driver is a Z-pinch, an excimer laser, or a diode-pumped solid-state laser, the energy that reaches the fusion target originates in a capacitor bank. The capacitor is the universal component.
Why the Supply Chain Is Not Ready
The qualified U.S. manufacturer base for high-voltage pulsed power capacitors is concentrated in a small number of firms, principally General Atomics Electromagnetic Systems. Internationally, ITOPP/ALCEN (France) and API Capacitors (UK) serve the market. General Atomics’ largest production contract was 2,500 units for the Z Refurbishment. The jump to tens of thousands of capacitors per facility represents a manufacturing scale-up that the existing base is not configured to deliver.
The material supply chain compounds the problem. Most capacitor-grade polypropylene film is sourced from China, leaving fusion developers exposed to geopolitical disruption and long lead times on their highest-volume component. The dominant dielectric, biaxially oriented polypropylene (BOPP), stores approximately 2.4 J/cc with a practical temperature ceiling around 85°C. Those characteristics impose hard limits on capacitor bank size, cooling requirements, repetition rate, and cost per shot.
The IEEE pre-roadmapping paper (Curry et al., 2025) quantifies the gap. For a facility firing 200 shots per year, current capacitor lifetimes of roughly 17,000 shots are adequate. For Hz-rate commercial operation requiring millions of shots over a multi-decade plant life, energy storage costs must decrease by 5–10×, and component lifetimes must increase by roughly 1,000×. The paper explicitly identifies advanced dielectrics capable of operating at hundreds of kilovolts under repetitive discharge as a critical development need.
The Dielectric Gap
BOPP has been the standard dielectric for capacitors in pulsed power applications for decades. It offers high breakdown strength, good self-healing characteristics, and well-understood manufacturing processes. For the Z Machine’s mission profile of 200 shots per year, BOPP-based capacitors perform reliably.
The limitations emerge when commercial fusion demands are applied. BOPP’s temperature ceiling of approximately 85°C requires active cooling infrastructure that adds volume, weight, and cost. Energy density at the film level is constrained, meaning capacitor banks must be physically large to store sufficient energy. And the shot life that works for laboratory repetition rates is orders of magnitude short of what a commercial power plant requires.
The National Academies Press has identified two primary pathways to improvement: increasing the operating field or increasing the dielectric constant while maintaining breakdown strength. The report pointed to high-dielectric-constant polymer films and nanolayered structures as potential routes to 2–4× improvements in energy density that would meaningfully change capacitor bank architecture.
For fusion program directors and procurement managers, the dielectric gap translates directly into system-level constraints: larger facilities, more cooling, higher capital costs, shorter maintenance intervals, and more frequent capacitor replacement across tens of thousands of units.
Closing that gap requires more than a better film. It requires an integrated development pathway that connects advanced dielectric materials to capacitor hardware validated under real operating conditions. To address this gap, Peak Nano and E&P Technologies announced a strategic partnership to co-develop next-generation high-energy-density capacitors for fusion drivers, pulsed power systems, and aerospace applications. E&P Technologies brings direct program experience from Xcimer Energy and Blue Origin, combining capacitor design, automated manufacturing, and rigorous qualification in a single integrated workflow. The partnership pairs Peak Nano’s NanoPlex™ HDC dielectric platform with E&P’s precision capacitor manufacturing to produce fusion-grade hardware through a collaborative design, rapid prototyping, and U.S.-based production pipeline.
Why This Matters Now
The timeline for major component procurement decisions is compressing. Pacific Fusion targets net facility gain by 2030. Fuse Energy plans Z-Star operations around 2027. Xcimer targets multi-kilojoule laser shots in 2026. Inertia Enterprises targets pilot plant construction around 2030. Component orders for these machines will be placed within the next two to three years.
Federal policy is beginning to respond. FY26 appropriations language has directed continued support for U.S.-based production of nanolayer capacitor film to reduce reliance on foreign suppliers. The proposed Fusion Manufacturing Parity Act and the expansion of 45X advanced manufacturing tax credits signal that Washington sees domestic component manufacturing as a prerequisite for fusion commercialization. But policy signals alone don’t produce hardware.
Whether the first commercial fusion power plant uses Z-pinch compression, excimer lasers, diode-pumped solid-state lasers, or an approach not yet at this scale, the capacitor banks at its foundation will fire billions of times over a multi-decade operational life. The dielectric film inside those capacitors will determine the energy density, thermal envelope, shot life, and ultimately the economics of the plant. The machines being designed today will define the performance requirements for those films. The qualified supply of advanced dielectric materials, manufactured in the United States from allied-nation sources, will determine whether those requirements can be met on the timelines the industry is targeting.
The fusion industry has spent years proving the physics. The next phase is proving the supply chain.
