Feb. 24, 2026 XCIMER ENERGY CORPORATION Commercialization of laser fusion energy Conner Galloway, Co-founder & CEO, Xcimer Energy Inc. Alexander Valys, Co-founder & CTO, Xcimer Energy Inc. Dr. Dirk Sutter, Principal Laser Scientist, TRUMPF Laser SE Executive Summary 2 Key Fusion Energy Challenges 4 Challenge 1: Fusion Plasma Confinement and Performance 4 An Introduction to Laser-inertial Fusion 4 Challenge 2: Materials, Chamber Survivability, and Byproduct Streams 7 Challenge 3: Cost and Economics 9 Xcimer’s Approach to Addressing Fusion’s Key Challenges 12 Xcimer’s Novel Laser Architecture 12 Xcimer Laser Cost and Schedule 18 Xcimer’s Hybrid Direct-Drive Target 20 Xcimer’s Chamber Design 22 Roadmap to Deployment of Laser Inertial Fusion Energy 25 Next Steps 28 Humanity has attempted to harness fusion energy for decades. Governments, universities, and private companies are pursuing multiple different technical approaches to commercialize fusion energy. But to date, the only fusion plasma confinement concept that has been experimentally demonstrated to generate more fusion energy than the incident energy to heat and confine a fusion plasma is laser-driven, hotspot-ignited inertial confinement fusion. This approach uses a high-energy laser to compress, heat, and ignite a millimeter-scale fuel capsule. In December 2022, the National Ignition Facility (NIF) at Lawrence Livermore National Laboratory (LLNL) in the USA used a 2 megajoule (MJ) laser to produce 3.6 MJ of fusion yield from this approach, making laser-inertial fusion the first and only fusion concept to exceed scientific breakeven. In April 2025 the NIF set a new record of 8.6 MJ of yield, for a gain (Qsci) exceeding 4. While the results from the NIF are an impressive scientific achievement that represents decades of hard work from scientists and engineers at LLNL and other US (and international) labs and universities, the NIF was never designed for fusion power production applications. There are several obstacles in the way of simply extending the NIF architecture to a commercially competitive fusion power plant. First, greater fusion performance is still required to generate enough surplus energy to sell commercially; a laser-efficiency × scientific-gain product of around 10 (vs. the 0.005×4≈0.02 as has been achieved on the NIF) will be required. Second, a fusion chamber (including laser windows and beam delivery) which can survive sustained exposure to hundreds of megawatts of fusion power, without damage requiring continual and expensive replacement, must be developed. Finally, the capital cost and maintenance costs of efficient high-energy lasers must be substantially reduced. Technologies derived from those deployed on the NIF, such as using laser diodes to pump solid-state lasers to improve laser efficiency, will require lengthy and capital-intensive development of production capacity and supply chains to scale to the laser energies needed for a laser-fusion power plant. Even with significant investment in such development, it is likely that diode-pumped solid-state lasers (DPSSLs) will not achieve costs below approximately $700 - $1,000 / joule-on-target. This cost presents economic challenges even if the other obstacles are overcome. Xcimer Energy in Denver, Colorado USA was founded in 2022 with the challenging goal of overcoming these obstacles. Xcimer is developing a laser technology based on large, efficient, low-cost deep-UV (248 nm) krypton-fluoride (KrF) excimer lasers, and nonlinear optics in gas media for beam manipulation. Together, these technologies remove much of the expensive and delicate solid optics, crystals, and laser glass needed in solid-state laser systems (such as the NIF) as well as obviate the need for expensive laser diodes. The Xcimer approach offers the potential to achieve significantly lower laser costs (under $100 /J) and higher energies (over 10 MJ) than other laser-fusion concepts, as well as reducing time-to-market. Furthermore, by bypassing optical fluence limits on solid-state optics, Xcimer can reduce the number of fusion chamber penetrations needed to just two, allowing the use of a thick-liquid layer that protects structural components of the chamber from fusion neutrons and eliminates the need for first-wall replacement. Together, these innovations provide a path to deploy economically competitive fusion power plants, while bypassing many of the technology development risks and uncertainties associated with other concepts. TRUMPF was founded in Stuttgart, Germany in 1923 and today is one of the world’s leading manufacturers of industrial laser systems and components, including laser diodes such as would be suitable for laser fusion applications. In 1985, TRUMPF introduced its first kilowatt CO2 laser and has since been the market and technology leader for industrial high-power gas lasers. Since the beginning of this millennium, TRUMPF has commercialized complex, high-power pulsed CO2 gas lasers which track and engage a molten tin droplet injected into vacuum at 50,000 times per second for today’s most advanced EUV lithography – a major engineering achievement which has enabled the continuation of Moore’s Law. Through these efforts and others TRUMPF has significant experience with the development, industrialization, mass production, and cost reduction of laser and optical technologies. TRUMPF also has a significant interest in providing laser-fusion technology towards realization of the first power plants. So far, due to their key competence in diode-pumped solid-state lasers, TRUMPF has mostly been engaged with DPSSL approaches for fusion. However, based on its expertise in gas lasers, the potentially disruptive approach of Xcimer has recently caught TRUMPF’s attention. In the subsequent sections, we summarize the major challenges that stand in the way of commercializing fusion energy, and laser-inertial fusion in particular. We then describe Xcimer’s approach to addressing these challenges, the advantages this approach will provide if successful, and the roadmap for resolving the extant technical risks and uncertainty in the Xcimer approach. Key Fusion Energy Challenges Many fusion concepts have been studied, but not all have a feasible path to commercial deployment. Any credible approach to commercial fusion energy must address three challenges: 1. fusion plasma confinement and performance, 2. materials, chamber survivability, and activated byproduct streams, and 3. cost and economics. Challenge 1: Fusion Plasma Confinement and Performance The hardest thing about fusion is fusion itself. Plasma confinement sufficient for ignition and breakeven is the highest risk element of every fusion approach. The Lawson triple product (nTτ) and the scientific gain (Qsci) are the two universal metrics of fusion performance across all concepts. The Lawson triple product is the product of plasma density, temperature, and confinement time. For a DT plasma to ignite or produce gain, the triple product must exceed a threshold curve as illustrated in Fig. 1. Scientific gain is an arguably even more important metric, and is defined as the fusion energy produced by a fusion plasma relative to the energy required to produce and heat the plasma. It is no surprise that the two best-performing confinement approaches (laser-inertial fusion and tokamak magnetic fusion) in terms of the Lawson metric are also the best performing in terms of Qsci, with the NIF having demonstrated Qsci of 4.13 in April 2025 and the Joint European Torus (JET) tokamak having achieved Qsci of 0.67 in 1997. The third best performing approach to date is magnetic liner fusion (MagLIF) pulsed power, with a DT-equivalent Qsci on the Z-Machine of under 0.01 (these shots utilized DD fuel and reported Qsci extrapolated to the performance expected if DT was used). Any viable fusion approach must reach the Lawson threshold and Qsci > 1. The entire history of fusion development over the past 60 years shows how hard this is and the many, often unforeseen challenges that arise. Thus, in seeking to make commercial fusion a reality, it is natural for some to focus on the only approach which has achieved these goals to date: laser-driven, hotspot-ignited inertial fusion as demonstrated by the NIF. An Introduction to Laser-inertial Fusion In laser-inertial fusion a millimeter-scale spherical fuel capsule is energized by a high-energy laser. The inert outer layer is blown off (ablated), producing a force which compresses, heats, and ultimately ignites the fusion fuel within, as illustrated in Fig. 2. NIF uses an approach called indirect-drive, in which the laser energy is not delivered directly to the fuel capsule, but is instead delivered to a cylindrical enclosure referred to as a hohlraum around the capsule, as depicted in Fig. 3. The walls of the hohlraum become superheated and re-emit this energy as x-rays, and it is this field of x-rays which then ablates and implodes the fuel capsule. This contrasts to the direct-drive approach, such as utilized on the OMEGA laser at the Laboratory for Laser Energetics, where the laser is aimed directly at the capsule. Indirect drive is less efficient than direct drive, as much of the laser energy is “wasted” heating the walls of the hohlraum or through x-rays that leave through the laser entrance holes, but indirect drive is more tolerant of high-order spatial non-uniformity in the laser beams themselves. On NIF’s most recent record shot in April 2025, 2 MJ of laser energy in a 15 ns pulse heated the walls of the hohlraum, and about 250 kJ of the re-emitted x-ray energy was absorbed by the capsule. The capsule ignited and yielded 8.6 MJ of fusion energy. This resulted in a scientific gain Qsci of 8.6 / 2.08 = 4.13. The capsule gain – the fusion yield relative to energy absorbed by the capsule – was approximately Qc = 8.6 / 0.25 = 34. The significant difference between capsule gain and scientific gain on NIF is due to the low efficiency of coupling laser energy to the capsule (roughly 12%) inherent to indirect drive. The scaling of inertial fusion performance with laser energy is driven by several competing factors. Capsule gain Qc is higher for larger capsules, but larger capsules also require more laser energy to drive their implosion. When both effects are considered, the capsule gain scales roughly with the ⅔ power of energy coupled to the capsule, Qc ∝ Ec^(2/3). Larger fuel capsules have other advantages, including the need for lower fuel compression. The compression required scales roughly as the inverse square root of energy, C ∝ 1/Ec^(1/2), and the lower compression requirement of larger capsules results in lower hydrodynamic instability and more robust and reliable fusion performance. The flip side of this scaling is that smaller capsules require higher compressions, which are more hydrodynamically unstable leading to inconsistent performance and sharply reduced gain. As the capsule size and driver energy are reduced, these instabilities inhibit ignition and the gain drops much faster than a ⅔ power law. The NIF is near the threshold of this transition. There is another historical data point that supports robust performance at higher energies. In the 1980s, Los Alamos and Lawrence Livermore National Laboratories in the US sought to determine the minimum energy required to achieve robust ignition and gain in a program called Halite-Centurion. In this program, x-rays from underground nuclear tests drove representative ICF fuel capsules, which put to rest questions about the basic feasibility of achieving high gain. While the details are classified, these tests led the national labs and National Academy of Sciences (NAS) to recommend building a 10 MJ laser. The NIF ultimately was built at 2 MJ, not because this was determined to be optimal, but because a 10 MJ laser was considered too costly and too large a step relative to the preceding NOVA facility at 30-40 kJ. The recent success of the NIF and the historical record of the Halite-Centurion program suggests a clear strategy: utilize the same type of fuel capsule that worked on the NIF, but do so with a larger capsule in order to achieve higher gain and more reliable performance, and seek to couple energy to the capsule from the laser as efficiently as possible. For example, using the ⅔ power-law scaling described above, coupling 10 MJ to a fuel capsule vs. the 250 kJ as on the NIF, when accounting for differences between x-ray and laser coupling, margin, and reliability considerations, suggests a capsule gain Qc of over 200 could be readily achievable with a 10 MJ laser that directly illuminates a correspondingly larger fuel capsule. With a wall-plug laser efficiency of at least 5%, this system would produce a “wall-plug” gain of over 10 – more than sufficient for a commercial power plant. A cost-effective laser system that could deliver these parameters, with sufficient symmetry to uniformly drive a fuel capsule at high coupling efficiency (i.e. utilizing direct-drive), would enable a highly viable approach that leverages the most proven implosion, confinement, and ignition physics in fusion: laser-driven hotspot-ignited inertial fusion. Fusion performance alone is not enough for a commercially viable power plant. A burning fusion plasma creates a very harsh environment. Even if a given approach overcomes the plasma confinement challenge, without a viable, low-activation, safe, and economic chamber concept to capture energy from the fusion plasma there may still be no path to commercial viability. Many different fusion chamber approaches have been proposed over the decades, which can be evaluated for commercial viability based on a few high-level considerations. A burning DT fusion plasma produces energy in three primary forms: 14 MeV neutrons, x-ray radiation, and ion debris. All of these are damaging to solid materials, and a fusion chamber must be designed to survive sustained exposure to all three. A majority (70%-80%) of the fusion energy is released in 14 MeV neutrons, which both damage materials and induce radioactivity. The remainder of the energy is released in X-rays and ion debris, both of which can ablate, shock, and erode solid materials over time. Assuming sufficient plasma performance is achieved, the primary goal of a fusion chamber is to safely capture this fusion energy and convert it into heat with minimal damage to solid structures, minimal activation and debris mass, and minimal maintenance downtime. There are two general classes of chamber designs: solid first walls and liquid first walls. Here, we will refer to the “first wall” as the material directly exposed to the fusion plasma, and the “first structural wall” as the closest solid material to the fusion plasma. The major challenge with a solid first wall is that 14 MeV neutrons damage and embrittle structural materials. This is measured by displacements per atom (dpa), the average number of times a single atom is knocked out of its lattice position by neutron collisions. There are currently no commercially available structural materials that can withstand more than 20 to 30 dpa from 14 MeV neutrons. The typical tokamak or dry-wall IFE concept accumulates roughly 10 to 20 dpa per year, and therefore any fusion system utilizing a solid first wall would have to replace all exposed structural elements every 1 to 2 years, unless new materials are developed that can extend the lifetime beyond this limit – a major materials science challenge. Furthermore, solid structures directly exposed to the 14 MeV fusion neutrons become highly radio-activated, producing intense gamma fields that may interfere with or outright preclude maintenance activities for several days to months after shutdown. Only after a suitable “cooling” period can radiation-hardened remote-handling robots safely approach the chamber to begin component removal and replacement. In systems that require it, frequent first-wall and blanket replacement over the plant’s lifetime will generate large volumes of activated solid waste, significantly increasing downtime, maintenance complexity, and operating cost. In some systems, portions of these waste streams may exceed low-level waste classifications, further complicating handling and disposal requirements. Testing and qualifying materials for these conditions will require construction of a new fusion-prototypical neutron source because no existing facility can reproduce the neutron spectrum, flux, and helium production rates experienced in a fusion dry wall chamber. A liquid first wall made out of low-Z materials, such as FLiBe or FLiNaK molten salts or molten lithium, thick enough (at least tens of centimeters) to moderate fusion neutrons, can eliminate these challenges with plasma-facing structural components. Effective “thick-liquid-wall” protection of these components can eliminate the need for first structural wall replacement and can significantly reduce activated waste and the severity level of worse-case accident scenarios. Furthermore, with sufficient liquid protection, current commercially available steels can be utilized as reactor components, eliminating the need for a new prototypical neutron source materials qualification facility. In addition, use of a thick-liquid wall can in principle enable almost complete capture of 14 MeV neutrons leading to high tritium breeding ratios without lithium enrichment required. Magnetic fusion systems such as a tokamak or stellarator cannot in general utilize thick liquid first walls, as introduction of impurities and/or droplets from the molten salt or liquid metal would create challenges such as plasma radiative cooling, fuel dilution, and disruption triggers. Laser inertial fusion systems, in principle, can utilize a thick-liquid wall. This is because the plasma confinement and burn in laser inertial fusion systems occurs in the millimeter-scale fuel capsule that is meters away from the first wall material and is not directly affected by vapor in the fusion chamber. However, conventional solid-state laser fusion architectures like those derived from the NIF are not compatible with thick-liquid walls due to a large number of laser beams and high shot repetition rate. Due to optical damage, the NIF and other solid-state architecture concepts require hundreds of laser beams to penetrate the fusion chamber, which makes it very challenging to engineer gaps in a thick liquid flow for each beam while also protecting the chamber structure around all of the beam ports. Additionally, at target shot rates higher than approximately one shot per second (1 Hz), a liquid wall does not have enough time to recover to a quiescent state to allow target injection and laser propagation for the subsequent shot. This is why the LIFE inertial fusion energy reactor concept developed by Lawrence Livermore National Laboratory and many partners in the late 2000s with the goal of designing a power plant around the NIF architecture, with hundreds of beams at a 10 to 16 Hz shot rate, must utilize a solid first chamber wall that requires replacement every year, leading to significant operational and maintenance costs. Pure pulsed-power approaches (e.g., Sandia’s Z-machine, MagLiF, etc.) in principle can also utilize a thick-liquid wall, but they still face significant containment, materials, and activation challenges. This is because pulsed power approaches must couple electrical energy directly to a target with a physical connection, generating over 4 orders of magnitude more debris throughput as compared to laser inertial fusion power plant. This debris generates more mechanical stress in any containment structure and results in significantly more activated waste, as far more material is exposed directly to 14 MeV neutrons. So, even if fusion performance challenges can be overcome and demonstrated, many fusion approaches still face significant material science, engineering, and architectural challenges to deploying a practical, economic, reliable, and safe power plant. Even if an approach can be developed that achieves meaningful fusion performance and overcomes the above challenges of maintenance and reliability in a power plant environment, the final obstacle to commercial viability is cost. In order to meaningfully change the trajectory of humanity’s energy use, fusion concepts must not merely work; they must achieve similar economics to the comparative mechanical simplicity of a natural gas turbine, a wind turbine or a solar panel. Even with the most aggressive assumptions about economies of scale and mass production, there are some concepts for which this is difficult to imagine ever being possible. While we could speculate about the costs of other fusion concepts, our focus here is laser-inertial fusion. Solid-state laser architectures and the DPSSL architectures required for adequate efficiency are the most well studied and well understood for laser fusion, with decades of development both in government and industry. However, the nature of solid-state laser architectures presents several challenges in making them commercially competitive for fusion energy: • Solid-state architectures must utilize a very large total aperture throughout the system to avoid damaging physical optics with their own laser light. A large aperture at the fusion chamber interface presents additional difficulties, as the laser windows must survive both laser-induced damage as well as damage produced by the harsh nuclear environment. • Architectural constraints (high beam counts, high repetition rates) lead to inherently complex chamber and plant schemes, which preclude the use of thick-liquid walls and require frequent chamber replacement, materials development, and increased waste disposal and maintenance costs. • Components such as laser gain media, frequency-conversion crystals, optics, and laser diodes have high upfront capital cost and will require significant and long-running capital investment in supply chains to produce these components in the quantities needed for a first pilot plant, let alone the deployment of numerous commercial plants. Optics Damage Optics damage is the most significant constraint for solid-state laser systems. The NIF system, for example, comprises 192 identical beamlines each providing a maximum energy of about 11.5 kJ. The peak laser fluence in the final optics and windows into the target chamber is about 15 J/cm2 (at 351 nm) with average fluence of about 8 J/cm2, and a total final optical aperture into the NIF chamber of 30 m2. At a fluence of 15 J/cm2, the NIF damages its optics on every full-power shot. As a result, after each shot a camera is placed at chamber center to inspect all optical surfaces in all 192 beamlines and identify damage sites. Damaged optics are removed and reprocessed through a sophisticated refurbishment loop onsite before they can be used again. While very expensive on a per-shot basis, the refurbishment loop is much less expensive than buying new optics. The NIF system could have been designed with lower peak and average fluences, but this would have required more beamlines, more optical area, and significantly higher upfront capital cost. As such the refurbishment loop is critical to maintaining a cadence of about 400 target shots per year, allowing the NIF to serve its national security missions. An optics refurbishment loop such as utilized on the NIF is impractical for a power plant operating at typically envisioned 10 Hz repetition rates (or even 1 Hz). To reduce downtime due to optics damage, a power plant would likely have to operate at lower peak fluence, roughly 4 J/cm2 or less, in order to maximize optics lifetime. Scaling this architecture up to 10 MJ and reducing peak fluence to this threshold would require a total aperture of about 300 m2 and significantly increase already high beamline and optics costs. Furthermore, it would require a larger target chamber than the current 10-meter-diameter NIF chamber, which would still need to be replaced every couple years, further driving costs. Finally, the 300 m2 of windows would be exposed to high energy neutrons, and it is unclear whether coatings or substrate material can be developed to allow long window lifetime in this environment. Laser Efficiency Achieving sufficient laser efficiency presents a further cost challenge. In the NIF, Nd:glass amplifier slabs are pumped by flashlamps that are inexpensive, but highly electrically inefficient, as only a small fraction of their broadband optical output is absorbed by the laser glass. This is the main contributor to the poor overall wall-plug efficiency of the NIF laser, at about 0.5%. A viable laser inertial fusion power plant needs much higher efficiencies – at a minimum roughly 5%. Thus, a solid-state laser fusion system must utilize laser diodes to pump the laser glass as opposed to flashlamps – hence the common acronym DPSSL for a Diode-Pumped Solid-State Laser. Diodes emit at the precise wavelength needed to pump the glass and can be up to 30 times more efficient than flashlamps. However, laser diodes present substantial cost challenges. Diodes today are not produced anywhere close to the volumes needed for a MJ-scale inertial fusion system. Even with a significant investment in mass production, which would take years to build out the required factories and production capabilities, there is a floor to how low DPSSL costs can be driven. Cost Estimates of Diode-Pumped Solid-State Lasers (DPSSLs) for Inertial Fusion Energy A DPSSL consists of a few primary components: laser pump diodes, laser glass gain media, frequency conversion crystals, and then a host of other components such as optics, electronics, cooling systems, power supplies, etc. For the high-level discussion here, we’ll focus mainly on the first three. A first important fact about DPSSLs for inertial fusion energy is that they need to produce powerful, intense nanosecond-scale laser pulses, whereas laser diodes are limited in their peak output power and are actually better suited for continuous operation. A second important fact is that laser glass gain media can only store energy in its excited state for approximately 0.3 milliseconds before that energy is spontaneously radiated away and wasted. As a result, a DPSSL works as follows: the laser diodes are energized and emit optical energy into the laser glass at a pulse length on the order of the excited state lifetime, long enough for energy to be stored up in the excited state of the glass. Then a fast, nanosecond-scale laser pulse is propagated through the glass which extracts a fraction of the energy now stored in that excited state. After laser energy extraction from the glass, the output laser pulse passes through large frequency conversion crystals that convert the wavelength of the laser light from infrared (about 1 micron) to ultraviolet (351 nm). Through this whole process, an overall wall-plug to UV efficiency of approximately 15% is expected to be achievable at scale in commercial systems. The laser diodes only have the timescale of the excited state lifetime (~0.3 milliseconds or less) to deliver all the required diode pump energy to the gain medium, which means they must run at very high peak power and are utilized at a very low overall duty cycle. This timescale also determines the conversion rate between the total diode power and final laser output energy. For example, 1 Watt of diode power over 0.3 milliseconds energizes the laser glass with 0.3 millijoules of energy. At this timescale the efficiency of diode pump energy to UV output is expected to be approximately 20%. Thus, 1 Watt of diode pump power would convert to approximately [1 W]×[0.3 ms]×[0.2] = 0.06 millijoules (mJ) of UV laser output, and therefore using these numbers a 10 MJ DPSSL would require a total diode pump power of approximately 170 gigawatts (GW). The efficiency could be improved by going to higher pump power over a shorter duration, but this would further increase laser pump diode costs. Laser diodes are packaged in “bars,” with today’s industrial diode bars considered for IFE capable of producing 0.5 to 1 kW of pump power. Here we will assume a 1 kW diode bar unit. The current cost of stacked diode bars is around $0.30 to $0.40 / watt. Thus, a 10 MJ DPSSL would require approximately $50B worth of laser diodes alone, not counting any other component of the laser. This is why the NIF is pumped with flashlamps, and not laser diodes, though we note that even with inexpensive flashlamps, and at 2 MJ rather than 10 MJ, the NIF facility cost approximately $7B in today's dollars. In considering how low diode production costs could be driven, one must consider not just the cost of manufacturing the diode chip itself, but also the associated cost of mounting and packaging the diodes on appropriate heat sinks and related driver electronics. For instance, consider a backplane-cooled 50 kW diode stack consisting of 50 of the 1 kW bars mentioned above. Assuming a bare diode bar cost of $5, which is demanding for modern large-scale 6” manufacturing and processing capabilities, results in a cost of $250 for the complete unpackaged stack. Including the costs of the entire mounting and packaging process with CTE-matching submounts, microchannel backplane coolers, AuSn solder, FAC lenses, etc., results in a total material costs of approximately $500/stack (not including labor). Considering all these costs plus the addition of labor, depreciation costs for high-volume robotic production lines, and certain assumptions of yield and other production costs, results in an estimated cost base of about $1000 / stack, or about 0.02 $/W (consistent with the asymptote of other analyses) of diode pump power. It is the authors’ assessment that this figure represents an absolute floor in cost for laser diodes, and billions of dollars of investment in diode supply chains over several years would be needed to achieve it. One can utilize this information to estimate the long-term best-case cost of a DPSSL driver for IFE. Assuming 0.02 $/W is achieved, a 10 MJ system would require $3.3B of laser diodes. This corresponds to a laser driver cost contribution of approximately $330 / J (hardware cost per UV laser output) for the diodes alone. This figure ignores the costs of other components of the laser such as the laser glass and frequency conversion crystals, which have significant supply chain challenges in themselves. Typical estimates are that laser diodes will contribute approximately 33% to 50% of the total cost of a DPSSL inertial fusion energy laser when the diode cost has reached its asymptote. Therefore, assuming that adequate investment has also occurred in other supply chains to produce the laser glass, crystals, and other components, the long-term best-case cost for a DPSSL driver is about $700 - $1,000 / joule. The resulting best-case, long-term cost floor for a 10 MJ solid-state laser system for IFE would be in the range of $7B to $10B. Xcimer’s Approach to Addressing Fusion’s Key Challenges Xcimer is developing an architecture for a laser-fusion power plant that aims to address the above three challenges of fusion performance, materials and activated byproduct streams, and system cost. There are three critical aspects of the Xcimer architecture: • An affordable, two-beam deep-UV (248 nm) inertial fusion energy laser system that is not constrained by laser damage limitations on solid optics and windows interfacing with a target chamber, allowing for a much smaller total aperture (< 1 m2) and much lower optical count than solid-state architectures. This system can scale to energies over 10 MJ on-target with sufficient wall-plug efficiency (> 5%), and most critically, can achieve all the parameters needed for commercial operation at substantially lower cost than a DPSSL, under $100 / joule at scale. • An inertial fusion target and fuel capsule design utilizing two-beam direct-drive deep-UV illumination, that leverages the implosion, plasma confinement, and ignition physics proven on the NIF, but scaled to the larger size enabled by a 10 MJ-class laser. At this size, this fuel capsule offers the highest confidence in fusion plasma confinement and performance of any fusion approach. Notably, sufficient gain and fusion yield are produced to enable operating a commercial plant at a repetition rate of 0.25 - 1 Hz, as required to allow a thick-liquid wall. • A fusion chamber and power plant based on the HYLIFE thick-liquid-wall concept first proposed by LLNL, which utilizes two-beam illumination and a repetition rate under 1 Hz to enable a flow of molten salt providing near-total protection of chamber components from 14 MeV neutron damage. Xcimer’s Novel Laser Architecture Solid-state lasers have been pursued for inertial fusion since the 1970s and widely used in multiple industrial, scientific, and other applications. Also in the 1970s, a different kind of laser was developed – the excimer laser. Rather than glass, most excimer lasers use a mixture of noble gases with a small (< 1%) concentration of a halogen as the laser gain medium. Excimers were initially developed for defense applications, and were a component of some proposed “Star Wars” missile defense systems. These first excimer lasers were pumped by relativistic electron beams (e-beams) produced by high-voltage Marx generators, as depicted in Fig. 4. Later in the late 1980s and 1990s, discharge-pumped systems were developed that do not require high voltage but are limited in energy output per pulse. These discharge-pumped systems were commercialized and deployed worldwide as the primary light sources for the semiconductor lithography industry for 40 years. Electron beams remain ideal for pumping large volumes of gas needed to produce high-energy pulses. High-energy e-beam pumped excimer lasers were studied for inertial fusion and offered several advantages, such as utilizing a lower-cost gas gain media that can’t be damaged, and a native deep UV wavelength (248 nm for KrF and 193 nm for ArF) that does not require frequency conversion crystals. Most significantly, excimer lasers can be built at much lower costs, as they require few advanced technologies, utilize abundant low-cost material inputs, and can be produced with relatively simple and easy-to-scale fabrication processes. The very first excimer laser was built by Paul Hoff at Lawrence Livermore National Laboratory (LLNL) in 1972. Multiple subsequent e-beam pumped KrF lasers were developed and built for fusion applications, including the RAPIER system at LLNL, the Aurora system at Los Alamos National Laboratory (LANL), and more recently the Electra system at the Naval Research laboratory (NRL), all shown in Fig. 5. The Aurora laser amplifier produced approximately the same energy (11 kJ) at 248 nm that a NIF beamline produces today at 351 nm. And most significantly, the 750 J Electra laser funded under the High Average Power Laser (HAPL) program demonstrated not only that 7% wall-plug efficiency could be achieved, but also operated continuously for 10 hours at 2.5 Hz repetition rate, solving several engineering challenges related to sustained repetitive operation for an electron-beam pumped system. At this time, Electra still holds the record for the highest UV pulse energy for a laser system operating at sustained IFE-relevant repetition rates. However, excimer lasers have two marked disadvantages with respect to inertial fusion energy. First, they are ½ to ⅔ of the wall-plug efficiency of DPSSLs (but still over an order of magnitude more efficient than flashlamp-pumped solid-state lasers such as the NIF). This is more than made up for by their cost-effectiveness and ability to scale to high energy levels (>10 MJ), enabling more than sufficient target/capsule gain to overcome the reduced efficiency. Second, while they can produce high energy at relatively low cost, they can only produce these energies effectively at long pulse lengths – hundreds of nanoseconds or longer – while pulse lengths of several nanoseconds are required to drive inertial fusion fuel capsules. Thus, the use of an excimer driver for inertial fusion requires a pulse compression scheme to convert the low-cost, long-pulse laser output into a short, high-intensity pulse. The conceptually simplest pulsed compression scheme is angular multiplexing, using a series of mirrors at varying distances to stack in time a sequence of short pulses such that they pass through an excimer laser amplifier back-to-back, but arrive at a target simultaneously. This method of compression is currently implemented on the NIKE KrF laser at NRL. However, this requires a significant number of optics, is not compatible with two-beam target illumination, and cannot scale to 10+ MJ. Nonlinear optical (NLO) schemes utilizing Stimulated Brillouin Scattering (SBS) or Stimulated Raman Scattering (SRS) for laser light temporal pulse compression were pursued in the early 1970s and 1980s as alternatives by multiple groups, including at LLNL, LANL, Imperial College, with the LLNL Energy and Technology Review reference providing a great explanation of both angular multiplexing and nonlinear optical compression schemes, including a hybrid version. In both SBS and SRS, two separate laser beams are coupled via interaction with the medium through which they propagate. Fig. 6. shows how this can be utilized for pulse compression, where a low-energy short-duration ‘seed’ beam counterpropagates relative to a high-energy long-duration ‘pump’ beam in a gas. As the beams cross each other and overlap, energy is transferred from the pump beam to the seed beam: as the seed beam sweeps across the pump, it is amplified, drawing energy from the pump. After the beams have passed each other, a majority of the energy that was in the pump pulse traveling to the left is now in the amplified seed pulse travelling to the right. Thus, the backward SBS NLO process acts like a gas mirror where the laser energy initially in the long-duration pump is both reflected and compressed in time simultaneously. Reflecting laser energy in the backward direction is not the only way a NLO gas mirror can operate. Several applications have utilized SRS in a configuration known as Raman beam combining, where multiple high energy pump beams cross a seed beam in a near forward direction, also depicted in Fig. 6. Here, the same energy transfer process from pump beam to seed beam occurs, but the seed beam is the same duration as the pump and points nearly in the same direction. This gas mirror configuration can concentrate laser energy spatially while achieving very high beam quality. Today, Xcimer is continuing where this past work on excimer lasers and NLO gas mirrors left off, and adopting a laser approach leveraging inexpensive, microsecond-scale excimer lasers that direct output to NLO gas mirrors initially for beam combination and then for pulse compression. While various concepts have been proposed by other organizations, in the public literature this architecture is closest in spirit to the Prometheus concept developed by McDonnell Douglas. Xcimer power plants will utilize up to 100 modular and stackable electron-beam-pumped krypton-fluoride (KrF) “Argos” excimer laser amplifiers, each with an output energy well over 100 kJ and a pulse length of over 1 μs, to generate hundreds of beams of low-cost laser energy at 248 nm. All of these beams will be aimed at a Raman beam combiner – the first of three nonlinear optical (NLO) gas mirrors. This process is depicted at an abstract level in Fig. 7. In the Raman beam combiner, the seed beam is injected through hundreds of overlapping pump beams at a small angle, so this mirror only changes the direction of the energy by a small amount, as depicted in Fig. 7. However, the “reflected” energy in the amplified seed beam becomes highly spatially concentrated, with a beam fluence well over 1,000 J/cm2, vs. the pump beams at about only 10 J/cm2 at the output of the Argos KrF modules. The spatially concentrated beam is then reflected in a second noble-gas NLO mirror, but this time utilizing SBS with a seed beam that is travelling in the opposite direction as the pump beam and with a much shorter duration of tens of nanoseconds. In this configuration, laser energy is both reflected 180 degrees and compressed in time, as the seed beam is approximately 50 times shorter in duration than the pump beam. Thus the reflected laser energy is much more intense. Finally, the third NLO mirror, also utilizing the SBS process in a noble gas, reflects laser energy again while compressing it further in time to a few nanoseconds, but only at a 170 degree near-backward angle. This final beam passes through a fast-acting vacuum shutter valve and engages a fusion fuel capsule in a fusion target. The shutter is a key element of the architecture, allowing the high-fluence beam to pass from gas to vacuum without touching a solid window. This is feasible due to the very small aperture size at this point of only a couple tens of centimeters. In summary, the key feature of this architecture is a method to compress μs-long, high-energy, low-cost KrF laser pulses to nanosecond scale with the efficiency and beam quality required for inertial confinement fusion via SBS after Raman beam combining. The final SBS power amplification for light on target occurs in 1 atm gas after a final optic, which allows 10^4 amplification to >1,600 J/cm2 while preserving phase, avoiding other non-linear propagation effects, and remaining below breakdown threshold, while the final physical optic sees less than 0.2 J/cm2 and is physically far removed from the fusion target plasma. The beams exit the SBS amplification region directly to the target chamber through a windowless vacuum transition. This bypasses all final optics and window damage issues and allows delivery of high driver energies (10 MJ or more) from a very small solid angle (< 10^-3 sr), with high standoff distance (50 m) and near-diffraction-limited beam quality. Xcimer estimates that the laser architecture described above can be constructed with total costs of approximately $100 to $120 / joule of laser light on-target in first-of-a-kind (FOAK) systems, and costs of $60 to $80 / joule in nth-of-a-kind (NOAK) systems, roughly an order of magnitude lower than the long-term DPSSL costs estimated above at $700 - $1,000 / joule. This cost advantage is derived from multiple factors: • Electron-beam (e-beam) pumped excimer lasers scale well to large sizes and high output energies, as fundamentally they consist of a low-cost gain medium (a gas mixture) contained in a mechanically simple pressure vessel, volumetrically pumped by an e-beam. The larger the laser gas cavity and higher the e-beam voltage, the more efficiently electron energy can be deposited in the gas. • The e-beam diodes used to pump the gas mixture in excimer laser amplifiers are simple high-voltage metal and insulator structures and bulk textile electron emitter surfaces meters in scale, not optical microelectronics devices, and are energized electrically by high-voltage Marx generators. • The materials in an excimer laser, including the e-beam cathode and Marx generator, are primarily steel, aluminum, and plastics – all commodities economical to procure and produce. Manufacturing these elements involves conventional casting, fastening, surface treatment, and assembly techniques of meter-scale metal parts and assemblies with physical scale and complexity comparable to automotive components. • The use of nonlinear optical processes for beam manipulation at high fluence removes a significant amount of expensive optics that would be subject to damage, and removes essentially all of the windows and optics exposed to the fusion environment relative to a conventional DPSSL IFE architecture. Table: High-Level Comparison Between Conventional DPSSL and Excimer-NLO Architecture DPSSL Architecture | Excimer-NLO Architecture - Laser pump diodes | Electron-beam cathode, Marx generator - Solid-state/glass laser gain media | Gas laser gain media - Optics and frequency conversion crystals | Significantly fewer optics, no conversion crystals The Marx generator and electron beam cathodes serve a similar function as laser pump diodes do for a DPSSL. Furthermore, similar to a DPSSL, the pump source in a commercial Xcimer IFE system (Marx and e-beam cathode) represents about half the total cost, so it’s useful to understand the function and cost of the Marx and e-beam pump source when comparing to laser pump diodes. A Marx generator consists of multiple “stages” that contain high-voltage capacitors. The stages are slowly (hundreds of milliseconds to a second) charged in parallel and then rapidly (in roughly 10 ns) connected in series through high-voltage switches. Once the stages are connected in series, their voltages add and the combined output is dumped at constant power over a microsecond into an impedance-matched electron beam cathode, with the resulting electron beam ultimately delivering the energy from the capacitors into krypton-fluorine laser gas. The electron excitation of the KrF laser gas is conceptually analogous to photons from the laser pump diodes exciting the Nd:glass laser gain medium in a DPSSL. The capacitors are the most expensive single element of the Marx generator, and are characterized by a cost per joule of electrical energy storage capacity ($/joule) vs. a cost of optical power rating ($/watt) as for a laser diode. Current high-voltage capacitor prices are roughly $10 / joule, and these units are not produced in volumes anywhere close to that needed for a commercial power plant. However, Xcimer has opened a proprietary capacitor manufacturing plant in Tucson, AZ, USA and is already producing capacitors in-house for current and future laser systems, at a significant cost savings. Cost estimates for volume production at the 3 MJ (stored) level for the first full-scale excimer amplifier module are approximately $0.85 / joule, and as volumes increase, this can be further driven down to below $0.40 / joule. Cost estimates for volume production of Xcimer’s e-beam cathodes and Marx generators is approximately $3.6 / joule stored for initial full-scale systems. This latter estimate is lower by a factor of ~1.7 compared to estimates for system costs in pulsed power fusion demonstrators, as the current density and voltage requirements to pump Xcimer’s KrF lasers over a microsecond are not nearly as stringent. For a 7% wall-plug efficient Xcimer laser system, $3.6 per joule stored corresponds to approximately $51 dollars per joule of UV laser light on target for Xcimer’s laser pump source (e-beam and Marx generator). Another cost driver in most laser systems are physical optics. Xcimer’s laser architecture replaces physical optics subject to damaging short-pulse high-intensity light with nonlinear optical interactions in gas, by first Raman beam combining at long pulse length to very high fluence and subsequently utilizing SBS NLO gas mirrors to reflect and compress the high-fluence light. As a result, the vast majority of physical optics in Xcimer’s system are positioned before the beam combiner. Thus each 50 cm x 50 cm beam from the Argos KrF laser amplifiers sees only three large 50 cm x 50 cm physical optics: a laser window and two turning array flat mirrors that direct each beam to the beam combiner. At pulse lengths above a microsecond, these optics can handle 8 to 10 joules/cm2 of UV light without damage. Furthermore, the quality of the optics can be relatively low, as the beam quality of the Raman NLO-combined and reflected light is not affected by the beam quality of the pumps. At an estimated average cost of approximately $55,000 per each of these large optics for first-of-a-kind systems, the total physical optics cost comes out to approximately $12 / joule of UV light on target. When including all of the other laser components such as laser chambers, gases and gas handling, NLO compression hardware, vacuum shutter, and front-end seed lasers, and accounting for remaining uncertainty, Xcimer’s estimate for a first-of-a-kind commercial laser system is $100 to $120 per joule on target. The cost of Nth-of-a-kind systems can likely be further reduced to $60 - $80 / joule with optimizations to efficiency and higher-volume manufacturing. Table 1: Breakdown of contributions to Xcimer's cost estimates for initial, first-of-a-kind (FOAK) laser systems Laser Component | Cost / joule-on-target in UV Laser Pump Source: - Pulsed Energy Storage Capacitors: $10 - Marx Generator: $24 - Electron Beam components: $17 Subtotal: $51 Laser Gain Media, Optics, Structures, Other: - Laser Chamber & Gas Systems: $19 - Laser Output Windows & Optics: $12 - Seed Lasers / Nonlinear Optical Systems: $23 - Control, Diagnostics, Other: $4 Subtotal: $58 Total cost estimate for FOAK excimer-NLO system: $109 / joule Total cost estimate for DPSSL architecture with most aggressive assumptions: $700 - $1,000 / joule The same factors that contribute to the favorable economics of Xcimer’s laser architecture also enable delivery of these systems on rapid timescales and with lower upfront investments. Because excimer lasers use relatively simple technologies that don’t require substantial capital investment for initial production, excimer laser systems at 100 kJ scale can be brought online within the next 2 years, and MJ-scale systems within the next 5 years. Xcimer’s technology offers the potential of faster timelines and lower capital costs as compared to what will be required to scale DPSSL supply chains to similar capacities. As discussed above, the only type of plasma confinement and ignition method that has surpassed scientific breakeven to date is laser-driven hotspot-ignited inertial fusion, where an inert ablator layer surrounding a spherical shell of deuterium-tritium fuel is rapidly heated and vaporized, creating a reactive force that implodes, heats, and ignites the fuel. Such a fuel capsule on the NIF has achieved a capsule gain of 34, where 8.6 MJ of fusion energy has been produced from only 250 kJ of absorbed ablator energy. For two reasons, the most direct path to high gain is to field a larger fuel capsule utilizing the same type of implosion and ignition mechanism. First, capsule gain scales with absorbed energy via a ⅔ power law, and so a larger capsule absorbing more energy can achieve higher capsule gain. Second, robustness against hydrodynamic instabilities, manufacturing nonuniformities, and implosion asymmetries is significantly increased at larger capsule size, which is critical for reliable commercial operation. Therefore, Xcimer’s approach is the following: • Field a NIF-style fuel capsule coupling approximately 10 MJ of energy, providing the ability to achieve capsule gains over 200. • Couple a majority of the laser energy directly to the fuel capsule, avoiding the inefficiency associated with indirect-drive. • Utilize only two laser beams, which is key to minimizing the number of chamber penetrations and enabling use of a thick-liquid-wall fusion blanket to mitigate neutron damage and activation issues. While the physics of direct drive has been demonstrated on the OMEGA laser facility at the Laboratory for Laser Energetics, the OMEGA laser utilizes 60 overlapping beams. Xcimer’s key challenge is achieving a sufficiently symmetric implosion with only two-beam illumination. This can be achieved if the spatial intensity pattern of the laser beam itself has a ring around the beam of higher intensity as compared to the middle, as shown in Fig. 9. This allows for higher laser intensity around the equator of the capsule as opposed to the pole, which can create sufficiently uniform absorption in the ablator to drive a spherical implosion. Xcimer has partnered with leading institutions around the design and modeling of this ‘hybrid’ direct-drive target concept, including the Laboratory for Laser Energetics, Los Alamos National Laboratory, and General Atomics, with a jointly published paper in 2024. The key enabling feature of two-beam direct drive is that the SBS amplification process of the low-energy seed beam that redirects the pump energy in Xcimer’s final NLO gas mirror preserves phase. This allows 1. The spatial intensity pattern of the light on the fuel capsule described above to be controlled by imparting phase on the low-energy seed beam via phase plates, and 2. Phase errors and aberrations introduced by the vacuum shutter to be controlled with adaptive optics which, in addition to the very low indices of refraction of the SBS gas medium, allows high control of unwanted intensity nonuniformity at the capsule. Demonstrating at scale that an SBS NLO gas mirror preserves phase and can be operated below other nonlinear thresholds that could inhibit operation, such as beam filamentation, are key milestones of Xcimer’s Phoenix prototype laser system. The materials, chamber survivability, and byproduct stream challenges discussed above makes clear the ideal scenario for a fusion reactor leading to minimum operating cost, complexity, and licensing challenges: • Utilize a very low-activation low-Z metal/molten salt thick-liquid (eg., Li, FLiBe, FLiNaK) first wall to protect all reactor components with at least tens of centimeters of liquid material. This has the added advantage of capturing all of the fusion energy directly in the liquid, which bypasses challenges with divertors in magnetic fusion devices, as well as enabling high tritium breeding ratios with natural lithium (~1.2 with FLiBe). • Produce as little vaporized mass as possible from fusion energy release with a minimum amount of activatable material. Of all fusion approaches considered to date with high scientific maturity, the HYLIFE concept comes closest to achieving these two goals. HYLIFE stands for High-Yield-Lithium-Injection-Fusion-Energy, and it was put forward by LLNL in the mid to late 1970s. (See Fig. 10) In this approach a low activation liquid (molten lithium in the first concept) rains down in a showerhead configuration, and an inertial fusion target is injected through small gaps in the flow and illuminated by two beams when at chamber center. This first concept considered either laser or ion beams and approximately 1 shot per second repetition rate with over 1 gigajoule of yield per shot and virtually complete capture of all neutrons in the thick-liquid wall. This “HYLIFE-1” first design concept focused on the reactor, with little discussion of how to build a laser or ion beam driver to enable it. A primary challenge with the driver was providing multi-megajoule energy levels with just two beams; this would put severe requirements on final optics for a conventional laser and final magnetic guides for an ion beam. By the 1990s, no driver existed that could enable the HYLIFE-1 two-beam illumination geometry, so the design evolved to HYLIFE-2, which envisioned utilizing 120 beams with lower total beam energy and a higher repetition rate of 6 Hz. This change put severe challenges on the liquid flow and chamber clearing, leading to oscillating jet designs to sweep out liquid droplets and force fast recondensation of vaporized liquid, all while allowing gaps for 120 beams through liquid jets. Ultimately, by the mid-2000s the HYLIFE concept was abandoned by LLNL, which subsequently pursued the LIFE dry-wall concept. Xcimer’s laser architecture can provide a driver that can enable the original two-beam, low repetition rate concept, driving a spherical implosion with 2-sided illumination from a very small final aperture, as described above. A depiction of an Xcimer power plant capable of producing hundreds of MWe to over 1 GWe is shown in Fig. 14, with an evolution of the chamber design looking back to the original HYLIFE-1 concept. Xcimer is building on a significant amount of past work on the HYLIFE concept, which was extensively investigated by LLNL and UC Berkeley over the years. Detailed simulations have been performed showing gigajoule fusion yields can be contained by a thick-liquid wall, and jet experiments have been implemented with water and oils as substitute for hot FLiBe showing laminar jets can be generated, which are required to prevent droplet and spray from blocking laser beampaths. Xcimer has confirmed much of this past work with further detailed simulations, with analysis indicating that less than 10 kg of FLiBe will be vaporized from a few-GJ fusion burst and that this vapor can readily vent through FLiBe jets with pressure loadings on chamber structures well below stress limits. Additionally, analysis indicates the total tritium inventory will be less than 200 grams in the GWe-scale commercial system (and under 150 grams in Xcimer’s 400 MWe “Athena” pilot plant), leading to a safer operating profile and lower licensing costs. Finally, analysis indicates that FLiNaK can be used in commercial plants in place of FLiBe, avoiding beryllium supply chains. This is due to Xcimer’s large fuel capsules providing sufficient neutron multiplication from (n,2n) reactions in the deuterium coupled with near-complete capture of fusion neutrons by the liquid wall, with an estimated TBR of 1.05. The first Athena pilot plant will certainly use FLiBe providing a TBR of approximately 1.2. However, many challenges lay ahead and development is still needed across several areas, for example, FLiBe pump and nozzle technology and redox control to prevent corrosion. Xcimer will be taking on these hydraulic and chemical challenges in lieu of the material science and nuclear challenges associated with dry-wall chambers. In summary, many fusion approaches still face significant material science, engineering, and architectural challenges to deploying a practical, economic, reliable, and safe power plant. Xcimer’s two-beam laser architecture is uniquely compatible with a thick-liquid wall, enabling better neutron capture, excellent tritium breeding, reduced material damage and activation, and exceptionally low tritium inventory, all of which collectively offer a more commercially viable fusion-chamber pathway. Roadmap Summary: - Phoenix: 1-2 kJ | Q2 2026 | Denver, CO, USA - Anvil: 200 kJ | 2028 | Denver, CO, USA (planned) - Vulcan (high-yield facility): 4 MJ (initial), 12 MJ (upgrade) | 2031 (initial) | TBD - Athena: 8 MJ on-target, 400 MWe output | 2035 | TBD Fusion is challenging, and humanity is still many years away from commercially deploying fusion power. Xcimer’s goal is to minimize risk on the back-end of the development roadmaps as well as ensure its commercial systems will have highly competitive economics. This is accomplished by adopting the most proven confinement and ignition method, which has already been demonstrated to scale to commercially relevant fusion gains, while utilizing almost complete thick-liquid-wall protection to minimize fusion chamber materials challenges, maintenance complexity, and waste streams. As compared to other approaches, this moves the majority of overall risk forward to the development of a laser driver that can enable the above, which is where Xcimer’s key innovations lie. Xcimer has a four-phase roadmap to retire key risks, mass produce excimer laser systems, and drive this technology to commercialization: 1 Phoenix. The Phoenix laser prototype system is under construction now at Xcimer’s facility in Denver, CO, with completion on track for early Q2 2026. Phoenix is a 75,000 square-foot facility with three goals: 1) demonstrating and validating the operation of excimer lasers at long pulse lengths (microsecond-scale) on the “LPK” laser platform, 2) developing, demonstrating, and beginning vertical integration of a new Marx pulsed power technology that will drive all future Xcimer KrF laser modules, and 3) demonstrating and validating the ability to use Stimulated Brillouin Scattering in an NLO gas mirror for pulse compression at scales relevant to inertial fusion. The largest excimer amplifier in the Phoenix facility, the "KJC," utilizes Xcimer's new Marx technology (a Marx is shown in Fig. 8) and will produce output energies approaching 2 kJ at 248 nm. Phoenix will soon be the largest and highest-energy use of SBS worldwide, and validate the key scientific and engineering aspects of Xcimer’s laser architecture. 2 Anvil. Xcimer’s next laser facility is codenamed Anvil. Anvil will begin with the construction and demonstration of Argos, a full-scale commercial excimer amplifier module with an output energy well over 100 kJ. The output from Argos will be used to drive a demonstrator IFE beamline, delivering 100 kJ in a single beamline to a target chamber and diagnostic suite. An identical beamline will then be constructed 180 degrees opposed to provide a 2-sided geometry with 200 kJ on-target. Anvil will demonstrate Xcimer’s laser architecture at full commercial scale and allow experimental validation of laser-target coupling for Xcimer’s unique two-beam direct-drive approach and other engineering uncertainties. Anvil will also support component lifetime development and test stands. An upgrade plan will demonstrate the pulse repetition rates needed for IFE, as well as a prototype target injector system. Anvil is expected to be completed in 2028, with the above-mentioned upgrades complete by the end of the decade. 3 Vulcan. Vulcan will be the world’s highest-energy and most powerful laser system, using stacks of Argos modules to drive two high-fluence NLO compression beamlines. Initially delivering 4 MJ to target from two sides, Vulcan will be upgradeable to 12 MJ. With initial operations commencing by the end of 2030, Vulcan’s goal is to achieve wall-plug breakeven by the end of 2031. After breakeven is demonstrated, Vulcan will continue to serve as a testbed and validation platform for future commercial target designs, as well as support scientific and national security missions. 4 Athena. Athena will be the first operational laser-fusion pilot power plant, producing about 400 MW electric by repetitively firing the laser and igniting a fuel capsule just under once per second, capturing the resulting thermal energy in the molten salt blanket, and then ultimately generating steam. With sufficient funding for parallel development of needed industrial-scale technologies (e.g. FLiBe pumps), Athena can be operational by the mid-2030s. Today, laser-inertial fusion is the most experimentally validated form of fusion energy, thanks to the successful demonstration and accomplishments at the National Ignition Facility. And, diode-pumped solid state lasers are the most well-understood and established technology for building a laser capable of igniting an inertial fusion fusion fuel capsule with sufficient laser efficiency. Yet even better fusion performance than demonstrated by the solid-state NIF is still needed for a commercially viable laser fusion power plant. A laser fusion power plant also needs to solve the “first-wall problem,” to avoid destroying its own structure during operation, and the optical damage problem, to avoid destroying its own laser optics. Finally, even if all these problems can be solved, the cost of building a suitable laser system using a solid state laser architecture may be prohibitive, with a capital cost floor in the range of $700 to $1,000 per joule of laser light on-target even with very aggressive assumptions and investment in scaling supply chains. Xcimer is developing a combination of technologies and a roadmap to overcome these challenges. • A larger fuel capsule driven with up to 12 MJ of laser energy, to achieve commercially relevant performance. This enables commercially relevant gain (>200) from a larger, more robust, less sensitive fuel capsule using the same ignition mechanism as utilized on the NIF. These large capsules are less sensitive to non-uniform drive, and can be symmetrically imploded with just two high-energy laser beams, 180 degrees apart. • A thick-liquid-wall chamber to solve the first-wall problem. The two-sided, high-brightness laser geometry allows the use of a thick-liquid wall to shield the structural wall of the chamber from fusion neutrons, allowing Xcimer to avoid first-wall replacement and the ensuing capital cost, maintenance, and availability consequences. • Achieve an overall laser hardware cost under $100 per joule. By replacing diode-pumped solid-state architectures with gas-excimer lasers and nonlinear optical systems, Xcimer “replaces glass with gas” and thus removes the most expensive and complex elements from the IFE laser system. Xcimer’s laser technology offers the potential for both lower costs and faster time-to-market than the dominant technology, and would significantly improve the economics of construction of a laser-fusion power plant. • Operate final optics well below laser damage thresholds. By performing final seed pulse amplification via nonlinear optical interaction in a gas medium after the final optic, Xcimer avoids any optical element in the system subject to damaging fluences, thus avoiding the need for an optical refurbishment or replacement system or the need to go to very large (many tens to hundreds of square meters) total aperture into the target chamber. • Achieve an overall laser wall-plug efficiency of 5% to 7%. An Nth of a kind system producing 250 target gain (Qsci) with a 7% laser efficiency will have a recirculating power fraction in the range of 11% to 13%. The basic elements of Xcimer’s laser architecture are not new; the system leverages ideas and systems which have been proposed and studied for decades. But by combining these concepts in a new way, Xcimer offers the potential to rapidly develop and deploy a technology that bypasses some of the most significant problems and risks associated with fusion. Xcimer’s path is not without risk itself. Most significantly, Xcimer must demonstrate that this laser architecture, never before built at MJ-scale, can deliver on the performance, cost and other advantages as outlined in this paper. And beyond the laser, there remain years of work and challenges to engineer an integrated power plant that can realize these advantages in a reliable, working system that can be scaled across the world. But if these challenges can be met, the benefits will be worth the effort. The authors wish to thank Dr. Hagen Zimer, CEO of Laser Technology at TRUMPF, for his support and attention in engaging on these topics.