Xcimer Energy Is Building the World’s Largest Laser. Here’s the Systems Engineering Problem That Comes With It.
When the National Ignition Facility achieved scientific breakeven in December 2022 — the moment a fusion target released more energy than the laser delivered to it — it ended a sixty-year debate about whether laser-inertial fusion was physically possible. It opened a different and harder debate: whether it could ever be commercially viable.
The laser system that accomplished this at NIF cost approximately $3.5 billion. It operates at roughly three shots per day. A commercial power plant needs to fire at rates closer to ten shots per second. The gap between those two numbers is not an engineering optimization problem. It is closer to a complete system reinvention.
Xcimer Energy, founded in Denver in 2022, is attempting exactly that reinvention. The company was founded by MIT plasma physicists Conner Galloway and Alexander Valys, who believe the path to commercial fusion is not to refine NIF’s solid-state laser architecture but to replace it entirely with excimer laser technology — specifically krypton fluoride, or KrF, gas lasers driven by pulsed electron beams. The physics of why this matters is inseparable from the engineering challenge it creates.
Why Excimer Lasers Change the Economics
NIF’s Nd:glass solid-state lasers are exquisitely precise instruments. They are also, for the purposes of commercial power generation, essentially unusable. The fundamental problem is thermal: solid-state laser gain media accumulates heat with every shot. At NIF’s three-shots-per-day cadence, there is time to let the glass cool. At ten shots per second, the medium would destroy itself.
Excimer lasers — lasers using excited-state molecular compounds as the gain medium — operate on a fundamentally different principle. KrF excimer lasers use a gas medium that is replenished and re-excited electrically, shot after shot. The gain medium doesn’t accumulate heat the same way because it isn’t solid. This is not a marginal improvement in thermal management. It is a categorical difference in how the energy system scales.
There are additional advantages. KrF lasers produce shorter wavelengths of light than the frequency-tripled Nd:glass beams at NIF, which is favorable for target coupling efficiency. Xcimer’s analysis suggests excimer-based systems can deliver equivalent drive energy to a fusion target at a fraction of the capital cost of a comparable solid-state system. That cost difference is what makes the commercial calculation potentially viable.
The catch — and it is a substantial one — is that no one has built a high-energy, high-repetition-rate electron-beam driven excimer laser at the scale Xcimer is targeting. The technical knowledge exists in scattered form across decades of defense research programs, some of which was declassified and some of which was simply abandoned when funding priorities shifted. Reconstructing and extending that knowledge base, and validating it experimentally, is the operational work Xcimer is doing right now.
What 2025 Actually Demonstrated
In 2025, Xcimer completed a milestone that received less public attention than it deserved: the world’s first private-sector electron-beam excimer laser, and the first such system built anywhere in approximately twenty years. In the same year, the company achieved the longest pulse length ever recorded for a KrF laser.
These are not incidental achievements. Pulse length in a KrF system is directly related to the delivered energy per shot, which is directly related to whether the laser can drive a fusion target to ignition conditions. Longer pulses mean more energy, and Xcimer’s result extended the known performance envelope of the technology. That this was done by a startup operating outside of a national laboratory setting, with a team assembled in roughly three years, is worth registering.
The Phoenix prototype laser — Xcimer’s intermediate-scale test system — is on schedule for completion in 2026. Phoenix is not a power plant component. It is a development and validation platform: a system designed to demonstrate that the underlying laser physics and pulsed-power engineering work as modeled at increasing scale before committing to the capital requirements of the next phase.
That next phase is Vulcan, a facility targeting engineering breakeven — the point where the fusion system produces enough energy to power its own operation — by 2031. Vulcan is currently in site selection. The scale of infrastructure required to support it is significant: high-voltage pulsed power systems, laser bay geometry that accommodates converging beam lines from multiple directions, target chamber engineering, tritium handling, and the beginnings of a thermal energy recovery system. Each of these domains has its own engineering community, its own standards bodies, its own modeling conventions, and its own failure modes.
The Systems Engineering Problem at the Core
Fusion devices are, by engineering classification, complex systems. But Xcimer’s program has a particular property that distinguishes it from most complex systems engineering programs: the physics of the system is not fully characterized in advance. The models improve through experimental iteration. Requirements that are set today based on physics models may need revision when Phoenix data comes in. That revision will propagate.
Consider a concrete example. The KrF laser gain medium must be excited by an electron beam delivered through a foil — a thin metallic membrane that separates the electron beam diode, operating in vacuum, from the gas medium at higher pressure. The foil must be thin enough to pass the electron beam with acceptable energy loss, strong enough to survive the pressure differential, and robust enough to survive repeated shots. These constraints interact with laser pulse energy requirements, which interact with target drive specifications, which interact with fusion yield models, which in turn constrain the minimum laser energy needed to achieve ignition. Change the foil material and you change the electron beam parameters. Change the electron beam parameters and you change the pulse shape. Change the pulse shape and you may change the target physics.
This is not an unusual situation in advanced physics programs. It is, however, one that exposes the limitations of managing requirements in traditional document-centric systems. An interface control document that captures the foil specification as a table entry does not automatically flag that a materials substitution has implications for the plasma physics model three levels up the requirement hierarchy. A human engineer tracking this manually across dozens of subsystem interfaces will, eventually, miss something.
The degree of bidirectional dependency across Xcimer’s technical domains — laser physics, pulsed power, optics and beam transport, target physics, nuclear engineering, materials science, thermal systems — means that requirements management is not a documentation overhead function. It is a core engineering activity that determines whether the development program proceeds coherently or accumulates hidden technical debt that surfaces at integration.
What Rigorous Systems Engineering Looks Like at This Scale
Programs like Xcimer’s require requirements management infrastructure that treats the system as a graph, not a document. Each requirement has upstream parents in physics models or design constraints, downstream children in hardware specifications, and lateral dependencies on requirements in adjacent subsystems. When a model is updated, every requirement that derives from it needs to be evaluated. When a test result contradicts a design assumption, the blast radius of that contradiction needs to be visible immediately.
This is not a philosophical preference. It is an operational necessity when a program is simultaneously developing novel physics, novel hardware, and the experimental infrastructure to validate both, under funding pressure and schedule constraints that do not permit extensive rework cycles.
Modern requirements management tools vary significantly in their ability to represent these structures. Legacy systems built around hierarchical document frameworks — IBM DOORS and its successors, Polarion in its traditional configuration — can represent requirements and link them, but they were architected around the assumption that requirements flow downward from system to subsystem and are largely stable once set. The feedback loops in a physics-intensive development program violate that assumption repeatedly.
Tools built around connected graph models, where requirements, test results, design artifacts, and physical interfaces are nodes in a traversable network rather than entries in a document, are better suited to this mode of engineering. Flow Engineering, which is built specifically for hardware and systems engineering teams working on first-of-kind programs, takes this approach — representing system architecture as a live model where cross-domain dependencies are explicit and changes propagate visibly across requirement hierarchies. For a program like Xcimer’s, where a laser physics revision on Tuesday might have structural implications for pulsed power specifications by Wednesday, that kind of connected visibility is the difference between managed complexity and accumulated surprise.
The practical implication is not that fusion startups need specific software. It is that the systems engineering methodology needs to match the epistemological character of the program — iterative, model-driven, with physics uncertainty treated as a first-class input to the requirements process rather than a nuisance to be resolved before requirements management begins.
An Honest Assessment
Xcimer is attempting something that has never been done commercially and that national programs with vastly greater resources have not yet accomplished at operational scale. The 2025 laser milestones are genuine and significant, but they validate physics and component-level engineering at a scale that is still orders of magnitude below what Vulcan will require. The 2031 engineering breakeven target is credible on paper but depends on a development program that must go right across multiple domains simultaneously.
The $120M raised to date is substantial for a fusion startup. It is not large relative to the capital requirements of the full program. Xcimer will need continued investment, and continued investment will require continued experimental progress. That creates a tight coupling between the technical program and the funding environment that is characteristic of breakthrough energy ventures and does not have easy mitigations.
What Xcimer has that many fusion programs do not is a physics approach with a validated proof of concept at NIF, a specific and defensible argument for why their laser architecture is cheaper and more scalable, and a team with the right combination of plasma physics depth and startup execution experience. The milestones they have hit are the milestones they said they would hit.
Whether the systems engineering infrastructure scales to match the physics ambition is the question that will determine whether Phoenix delivers the data Xcimer needs and whether Vulcan is a facility that gets built. That question is less dramatic than the fusion physics, but it is equally determinative of whether the company succeeds.
Building the world’s largest laser is a physics problem and a materials problem and a pulsed power problem. It is also, at its foundation, a requirements management problem. The teams that have learned this lesson from prior large-scale physics programs did not learn it easily.