Pricing Laser Fusion: A Live LCOE Calculator

Everyone who covers fusion eventually gets asked the same question: what would the electricity actually cost? Almost nobody answers it, because almost nobody can. There’s no operating laser fusion power plant, so there’s no bill to point to. What exists instead is a handful of assumptions about lasers, targets, and plants that don’t exist yet, wired together into a spreadsheet.

So I built the spreadsheet. Then I turned it into something you can play with.

Quick orientation before you start dragging sliders: the three buttons at the top aren’t three snapshots of the same thing at different points in time. They’re three different kinds of numbers. Today’s laser is what a real facility (NIF) has actually, physically done this week, with hardware that exists — proof the physics works, not a power plant. LLNL LIFE design is a real engineering blueprint for a future plant nobody has built yet, published by the same lab that runs NIF. Hawker’s optimistic case is a theoretical best case from an economics paper, not tied to any specific lab’s design at all — just “if every parameter lands near its best plausible value.” None of the three generates electricity today. They sit at three different distances from reality: proven physics, an unbuilt blueprint, and an aspirational floor.

Where the numbers come from

The math underneath is Nicholas Hawker’s 2020 model, published in the Royal Society’s Philosophical Transactions A. Hawker co-founded First Light Fusion, the same company whose tritium breeding numbers I wrote about last time, and he built this specifically to be technology-agnostic: fourteen parameters that describe any inertial fusion concept, laser-driven or otherwise, without assuming a particular reactor design. That’s the equation chain the calculator reruns live in your browser on every slider drag: driver energy through efficiency to target energy, through gain to fusion yield, through the blanket multiplier to thermal power, minus recirculating power, through thermal efficiency to net electric output, then capital and operating costs annualized against that output. I collapsed his year-by-year discounted cash flow into a standard capital-recovery-factor form (30-year plant life, fixed) so it updates instantly instead of re-solving on every keystroke. Plugging in his own published optimistic case reproduces his $24.6/MWh result to within a few percent, about as much confidence as a rewritten model deserves.

For the default view, I wanted a design point I could actually attribute to a real engineering study rather than a plausible-sounding guess, so it’s built on LLNL’s LIFE program, the fullest laser-IFE power plant costing exercise that’s been published. To be precise about what comes from where: the physics is LLNL’s published Market Entry Plant design point (2.2 MJ of laser light on target, gain around 60, 8.3 shots a second, 45% thermal efficiency, 18% driver wall-plug efficiency), and running it through Hawker’s equations reproduces their headline machine — about 1,100 MW of fusion power and 388 MWe net, within about 11% of the 437 MWe LLNL published. The 25-cents-a-target cost goal is also LLNL’s own stated number (current targets run thousands of dollars each, which is the whole ballgame for the fuel-cycle supply chain I wrote about last time). The plant-cost slider, though, is my calibration, not their data: I set it so the full model reproduces the widely reported 9.1 cents/kWh LIFE projection, and the calculator duly lands at $90.5/MWh. Everything else on that preset — driver lifetime, O&M, target-factory infrastructure — is Hawker’s generic constant, because LLNL never published costs at that granularity.

That calibration hides something worth saying out loud. In the peer-reviewed LIFE economics paper, Anklam and colleagues put first-of-a-kind capital cost at $4 billion to $6 billion. Take that number at face value and run it through Hawker’s plain-vanilla financing — 8% discount rate, 70% availability, no special treatment — and the same machine prices out at roughly $180 to $250 per MWh, two to three times the 9.1 cents LLNL reported. Neither side is wrong. LLNL’s figure assumes utility-style financing and a tenth-of-a-kind plant running at 92% availability; Hawker’s defaults assume a merchant plant borrowing at market rates. The entire gap between “fusion competitive with nuclear” and “fusion at three times nuclear” lives in the financing assumptions, before you touch a single physics slider. Keep that in mind every time a fusion company quotes you an LCOE.

LIFE itself dates to around 2011 and was eventually shelved. But LLNL is back in this business: after NIF’s ignition results, the lab stood up the Livermore Institute for Fusion Technology in 2025, and it just released GEM, a “Generalized Economics Model” that does more or less what this calculator does, for the same class of design (diode-pumped laser driver, indirect-drive targets, liquid-lithium chamber), built on the same LIFE lineage. GEM ships as a licensed Excel tool for industry rather than a public equation set, which is the practical reason I built this from Hawker’s math instead. But it’s a good sign that the lab everyone treats as the most credible source on laser-IFE costing is still actively working the same problem this calculator is toy-modeling.

Gain is not the same as a power plant

The “Today’s laser” preset is deliberately close to what the National Ignition Facility has actually demonstrated, and it’s worth clicking before anything else. NIF fired 2.05 MJ of laser light at a target in December 2022 and got 3.15 MJ of fusion energy back: gain greater than one, a genuine first, front-page news everywhere. The program has kept climbing since — the April 2025 record shot delivered 2.08 MJ to the target and got 8.6 MJ back, a target gain of 4.13, which is the gain the preset uses. What makes fewer headlines is that NIF draws roughly 300 MJ from the electrical grid per shot, because its lasers are only about 1% wall-plug efficient. Fusion energy out over grid energy in — the number that actually determines whether a plant makes electricity — was about 1/100 for the 2022 shot, and even the 2025 record only gets that to roughly 1/35.

That’s why the calculator flags this preset as “not net electric” instead of giving it a price. There’s no LCOE for a plant that’s a net energy sink; the question isn’t well-formed yet. Target gain is necessary but nowhere near sufficient, and conflating the two is the single most common misreading of fusion news.

The sensitivity panel is the actual point

The “what’s driving the price” panel at the bottom recomputes live: it tests each of the fourteen parameters 10% higher and 10% lower, one at a time, holding everything else fixed, and shows whichever direction actually lowers the price and by how much. (A parameter’s “cheap direction” isn’t fixed in general — shrinking driver energy, for instance, shrinks electrical output faster than it shrinks capital cost at some design points, so the panel checks both ways rather than assuming.) On the LLNL-grounded default, gain and driver energy top the list, because that Market Entry Plant design point is nowhere near Hawker’s high-gain, low-frequency optimum, so there’s real room to buy the LCOE down with better target physics. Drag gain up toward Hawker’s own optimistic value and watch that leverage shrink: once gain stops being the binding constraint, the panel reorders and the cost constants take over.

That reordering is the point, and it echoes something Hawker found running his model across the whole 14-parameter space rather than one design at a time: averaged over many candidate designs, the strongest correlations with LCOE weren’t the physics parameters that get all the press coverage. They were the discount rate, the plant cost constant, and the target cost constant. Gain mattered, but less on average than the interest rate on the debt used to build the thing.

That tracks with how every other capital-intensive energy technology has actually gotten cheap. Nuclear fission’s history is as much about financing risk and regulatory certainty as it is about reactor physics. Solar’s cost curve is as much about manufacturing scale-up as photovoltaic efficiency gains. If laser fusion ever gets built at scale, the discount rate a utility or a government is willing to offer it, and the confidence with which the tenth plant can be costed against the first, will probably matter as much as the gain curve on some target physicist’s whiteboard.

Which is a strange thing to say about a technology that still needs a genuine physics miracle to work at all. Both things are true at once: the miracle is necessary, and it isn’t the part that sets the price.

Sources: Hawker, “A simplified economic model for inertial fusion,” Phil. Trans. R. Soc. A (2021) · Anklam et al., “LIFE: The Case for Early Commercialization of Fusion Energy,” Fusion Science & Technology (2011) · LLNL, “Systems Modeling for the Laser Fusion-Fission Energy (LIFE) Power Plant” · Livermore Institute for Fusion Technology / GEM · LLNL on the December 2022 ignition shot · LLNL on the April 2025 record shot · Big Think on NIF’s overall energy balance · Lazard, Levelized Cost of Energy+ (June 2025)