General Fusion Group is a pre-revenue fusion energy company that reached Nasdaq in the summer of this year, when it merged with Spring Valley Acquisition Corp. III and paired the deal with a large PIPE, the capital event that defines the investment. The shares now trade below the PIPE unit price, and that gap frames the whole question of what the market is actually paying for.
The defining recent development is the June plasma heating result from the Lawson Machine 26 demonstration, in which the lithium liner compressed the plasma to roughly 8.4 million degrees Celsius. That reading is the first measured proof that the magnetized target liner produces real heating, and it sits well short of the first milestone, which is why the result matters and at the same time why the stock still carries so much of the downside in its price.
The tension sits in the capital structure. The PIPE investors hold multiple voting shares that accrue value at a double-digit percentage each year, and the deal stacks a twelve-and-a-half million share earnout plus a further twenty-five million PIPE warrant shares behind them. Every new tranche of equity issued to fund the next physics milestone ranks ahead of the public shares, so the stock prices in dilution before any physics success.
The near-term catalyst is the LM26 program itself, and the first gate it has to clear is the one the June result already approached. Closing that gap, and then pushing on toward the second gate, is the whole near-term story. All of that has to happen while the F-1 registration opens the door for the early holders to sell into a float that is about to expand.
General Fusion Group Ltd. is the Nasdaq-listed successor to two entities. The public shell, Spring Valley Acquisition Corp. III, was a Cayman blank check company. The operating business, General Fusion Inc., was a British Columbia company founded in 2002 that spent two decades developing magnetized target fusion, an approach in which liquid lithium is compressed against a plasma target. The Cayman shell continued into British Columbia, changed its name, and absorbed the operating company, leaving a Canadian-domiciled parent headquartered in Richmond.
The strategic posture is a demonstration-to-plant path rather than a commercialization path. The company runs the Lawson Machine 26, a fusion demonstration machine operating at about half the diameter of its planned commercial design, to validate the physics milestones of the program. In parallel it evaluates first-of-a-kind plant sites and builds the customer and regulatory framework that a net-energy plant would need when the technology matures. The company holds no customers, generates no revenue, and treats the government of Canada and the Business Development Bank of Canada as anchors of its capital base.
The ownership base is a tell of what the market thinks this is. Alyeska Master Fund holds about a quarter on a 63.5 million share basis, BDC Capital Inc. about a sixth, and Segra New Energy Opportunities about a tenth, with PenderFund and the Spring Valley sponsor rounding out the five percent holders. That is a blend of venture-grade fusion investors, a Canadian crown institution, and the SPAC sponsor, all holding a stock whose value is a function of one physics program.
The listing itself is an event, not a starting line. The shares began trading on Nasdaq under the symbol GFUZ on the closing date, with the warrants listed as GFUZW, and the company filed a shell report on Form 20-F. An F-1 to register the resale of the PIPE and sponsor holdings followed a few weeks later, and the first quarter as a public company is also the first quarter in which the early backers are permitted to sell.
Magnetized target fusion is General Fusion's product and its moat, and it is a distinct answer to a hard problem. The physics question is whether a liner of liquid lithium, driven inward against a plasma target, can compress the target to the temperature and density that sustain fusion reactions. The engineering answer General Fusion builds around is that the lithium liner is both the compression medium and the tritium source, so the machine's blanket and its ignition path are the same liquid. That coupling is the core intellectual property, and it is why the company frames MTF as engineering-driven rather than a pure physics pursuit.
The Lawson Machine 26 is the proof vehicle. It began operating in 2025 and runs the full compression cycle at about half the diameter of the intended commercial machine, which lets the team validate the liner, the magnetic fields, and the plasma formation at a scale where failures are cheap. The program has three named milestones in sequence, from a first plasma-temperature gate to a second, higher-temperature gate, with the Lawson criterion, the product of temperature, density, and confinement time, as the final mark of a self-sustaining burn. Each milestone is a discrete, reportable event, which is what makes the program investable at all.
The June 2026 result is the mechanism that moves the thesis. By compressing the plasma with the lithium liner, the team produced meaningful heating to an electron temperature of about 8.4 million degrees Celsius. That reading is 72% of the way to the first milestone, and the consequence for shareholders is that the company's central claim has shifted from modeled to measured. The gap to the first milestone is the same order of physics problem, and clearing it converts the demonstration machine from a prototype into a validated platform for the larger step.
The moat has two layers and a gap. The first layer is the two-decade head start, a longer public record of liner-driven compression than most of its magnetized target peers. The second layer is the Canadian government stake, which brings patient capital and regulatory familiarity. The gap is integration: the filing is explicit that the MTF systems have not yet been integrated and demonstrated at power-plant-relevant scale, so the moat today is a research moat, not a manufacturing one.
The pre-merger financials describe a company one quarter away from running out of money. At the end of 2025, General Fusion held $49.1 million in cash, a thin balance against the program it had to fund. That cash sat against a working capital deficit of $27.3 million. The accumulated deficit stood at $332.0 million, the product of two decades of financing a machine that has never produced electricity. The business combination changed the balance sheet in one transaction. The closing date was July 10, 2026. At that point the company had received about $123.4 million in net proceeds, a figure that combined the PIPE after transaction costs with the balance of the Spring Valley trust account, which had been drained to under $20 million after the redemptions. The merger therefore landed with far more cash than the trust alone implied.
The operating loss for 2025 was $31.4 million, a meaningful improvement from the prior year. The prior-year figure stood at $57.8 million, a figure that reflected the full ramp of the demonstration program. Operating cash burn ran to $24.0 million for the year. That compares with $29.1 million in 2024. Research and development was the largest single line, with government assistance offsetting part of the operating cost, a dependency the Canadian crown anchor makes explicit.
The income statement carries a layer of accounting noise that flatters nothing. The pro forma net loss for the first half of 2026 was $80.0 million, but most of that was a non-cash revaluation of the government contribution liability. The cash-relevant number is the operating burn, a run rate near $55 million a year that the new cash comfortably covers for at least 12 months.
Three structural items define the forward financial picture. First, the PIPE multiple voting shares accrue value at a double-digit percentage until conversion, a charge the company booked as a deemed dividend in 2025 and again for the first half of 2026, and it grows the longer the preferred stack runs. Second, the earnout shares, warrants, and options are all liability-classified, so their fair value moves through profit or loss with every share price move. Third, the F-1 opened in early September 2026, registering the resale of the early holdings, which sets up the supply overhang the market has already begun to discount.
The outlook is a sequence of physics gates, each of which consumes the capital the merger raised. The named variable here is the LM26 milestone track: the June 2026 result already measured, then the first temperature gate and the second, with the Lawson criterion as the endpoint that marks a burn. The company's management discussion says the current cash covers at least 12 months of planned operations, which maps directly onto the time it needs to reach the first gate and begin the engineering work on seals, valves, and heat exchange for the first-of-a-kind plant.
A second named variable is the capital gap between the first gate and a plant. The pro forma burn for the first half of 2026 ran near $55 million annualized on operating lines alone, and the company is explicit that additional financing is required as the program advances. The March 2026 amendment to the Strategic Innovation Fund contribution agreement, which added five million Canadian currency units against warrant consideration, shows the pattern: every tranche of patient capital arrives with dilution attached, and the recurring test is whether the market funds the next gate at a price that does not punish the existing holders.
A third variable is the F-1 overhang. The registration covers resale of the multiple voting shares, the PIPE warrants, the sponsor and public warrants, and the earnout stack, an aggregate of about 109.9 million shares issuable on exercise or conversion against 53.1 million shares outstanding before any of it. The selling shareholders, not the company, control the timing, and the first two quarters as a public company are when the lock-ups and the registration rights collide: the stock has to rise to make the physics case credible, but the rise is exactly what the early holders are positioned to sell.
A fourth variable is the multiple voting share structure itself. The PIPE shares convert at a floating price and carry a sub-10% blocker for the largest holder, which means the lead investor can sit on its conversion while the public float absorbs the warrant exercise, and the double-digit accrual means the preferred stack grows in value even if the stock stalls. The governance dimension adds friction on top, because the company disclosed material weaknesses in internal control over financial reporting, the board was reconstituted on the closing date with seven new directors, and the management team has limited experience operating a Nasdaq-listed company, the kind of friction that slows disclosure and gives the overhang more room to work.
The downside case is a physics stall combined with a capital structure that punishes the public holder for it. If the LM26 program fails to reach the first milestone, the company still carries a burn that the merger cash covers only for a limited number of quarters, and the next financing arrives at whatever price the market tolerates. The multiple voting shares, the earnout stack, and the warrant stack all dilute further with each round, and the accrual keeps the preferred value growing through the stall, so the share price ends up tracking the cash balance rather than the physics, with the F-1 overhang turning a falling stock into a falling stock with a guaranteed seller base.
A second downside is structural. The sub-10% blocker and the floating conversion price give the PIPE investors the option to hold their position while the public float absorbs the overhang, and the earnout shares are liability-classified, so a sharp price move reprices them through the income statement and can trigger scrutiny even though the company carries no traditional debt covenants. The material weaknesses in internal control add a compliance risk on top, in a first year as a foreign private issuer filing on a reduced cadence, and the combined effect is a stock whose worst case is set by its own cap table rather than by its science.
The counterargument to the bear case is direct, and it is worth stating plainly. The June plasma heating result is not a press-release number; it is a measured electron temperature produced by the exact mechanism the company built the thesis on, sitting at 72% of the first milestone in a program with no comparable public peer at its stage. The capital base, led by Alyeska and anchored by the Business Development Bank of Canada, is the kind that survives a two-year physics program, and the government of Canada has already extended the Strategic Innovation Fund contribution rather than walk away, which means the funding runway is better than almost any of its cohort and the physics is the only thing standing between the stock and its next milestone.
The honest resolution is that both cases are true at once, and they are resolved by the same number: the electron temperature. Until the first milestone is reported, the overhang and the dilution dominate the price action, and once it is reported, the overhang becomes a supply issue the market can absorb and the stock re-rates on the second gate. The risk is not that the bear case wins; it is that the window between the two is long enough for the capital structure to erode the public position before the physics can defend it.
Valuing a pre-revenue fusion company requires a framework that prices the option, not the earnings. The right frame is three layers: the cash, the dilution stack, and the milestone probability. The stock price sits at $8.24 per share. Against 53.1 million shares outstanding, that implies a market cap near $438 million. That number overstates the claim on the company, because it sits in front of roughly 109.9 million issuable shares across the preferred, warrant, and earnout stacks and behind a large pro forma cash balance.
The cash layer is the only one with a hard number. Net cash after the merger is near $123.4 million, set against a pro forma accumulated deficit of $444.7 million that no amount of physics can erase. The burn runs near $55 million a year on operating lines, which buys roughly two years of the LM26 program at the current pace. That is the true asset the public shares own before any physics happens, and a bear valuation that strips the market cap down to the net cash divided by the fully diluted count puts a floor under the stock that the overhang alone cannot break as long as the company keeps filing and the burn stays on plan.
The dilution layer is the discount. The PIPE multiple voting shares rank ahead of the public float. The PIPE warrant tranche accounts for 25.3 million shares. The SVS warrant tranche sits at 11.8 million shares, while the 12.5 million earnout shares complete the stack. On top of the share count, the double-digit accrual means the preferred grows through a flat tape, so the discount has two moving parts. The market has already begun to price that discount, because the transaction value embedded in the exchange ratio was set by the PIPE investors at $10.20 per unit rather than by the public market, and the public has since marked the shares down to $8.24, a gap the overhang has not yet finished widening.
The milestone layer is where the scenarios diverge, and the divergence is wide enough to matter. A bear case holds the physics at 0.72 keV through two more financing rounds, and the fully diluted value lands in the low single digits per share, where the net cash roughly equals the diluted claim. A base case clears the first gate on schedule, funds the second-gate work from the existing cash, and holds the stock in the $6.00 to $9.00 band, where the market pays for the demonstration without pricing in the plant. A bull case reaches the second gate or a credible path to the Lawson criterion, which re-rates the option value and carries the stock back toward the PIPE level. Past the deal anchor, the first earnout trigger at $15.00 is where the Class A earnout shares begin to convert and the overhang partially resolves into holders rather than sellers.
General Fusion is the purest expression of the fusion investment case on Nasdaq, and it is also the purest expression of its risk. The company has now put a measured number where a model used to be: 0.72 keV of plasma heating from the lithium liner, two-thirds of the way to the first milestone in a program that defines the stock. That is the strongest possible position for a pre-revenue physics company, and it is the reason the PIPE investors paid $10.20 while the public trades at $8.24, a gap that is a bet on the timing of the next gate rather than a verdict on the science.
The judgment on the structure is less forgiving, and it deserves to be stated without the numbers doing the talking. The multiple voting shares, the double-digit accrual, the sub-10% blocker, and the liability-classified earnout stack are a capital design that transfers optionality from the public holder to the PIPE group in every scenario except a fast bull case. The F-1, which opened in early September 2026, gives the early backers a registered exit into a float that is about to triple, and that is not a flaw in the physics but a feature of the deal that the public shares are paying for in real time.
The resolution comes down to one variable, and the reader should hold the position against it: the electron temperature on the next LM26 report. If the first gate arrives on schedule, the overhang becomes absorbable, the net cash buys the second-gate work, and the stock has a clear path back to the $10.20 anchor. If it slips, the dilution and the accrual compound, the overhang gets a larger discount to sell into, and the net cash becomes the only thing standing between the share price and the fully diluted floor near $3.00. In either case the mechanism is the same, and the market is pricing the probability of the first outcome against the cost of the second.
This is a stock that belongs in a portfolio sized for a two-year physics hold, where the position is underwritten by the cash balance rather than the multiple. The downside is real, bounded, and priced into the structure; the upside is real, unbounded, and gated by a single number. The honest read of GFUZ at $8.24 is that the public holder is long the physics with a short the capital structure, and the two legs only align if the next milestone clears on time.