Oklo Turned On a Reactor for $75 Million. Now It Wants to Sell Fuel to Everyone Else.
Groves-1 finished at a fraction of industry cost and Aurora-Idaho is going vertical — but the 2028 start-up quietly slid toward 2029, the Meta deal is not binding, and the stock is down 55% in 90 days.
OKLO · Earnings Call · 2026-11-06
A company sprinting while its stock collapses
Oklo's third quarter lands against a brutal tape. The shares have given up 55% over the last 90 trading days and sit 86.5% below the October 2025 peak of $174.14, leaving a market capitalization of roughly $4.4 billion. The long-term supply theme that once carried the whole advanced-nuclear complex has been unwinding — OKLO itself appears in the trailing-90-day decliner basket for that keyword alongside Samsung and AXTI. Even the industry's loudest secular prop, the AI data center build-out, is voting the other way: 58 negative tickers against 7 positive over the same window. And yet the operating story is, on its face, accelerating. Dig into what management actually said this quarter and there are three genuinely new things — a new revenue ambition, a timeline that moved the wrong way, and a customer relationship that got quietly softened — all sharing a call with a record-setting construction feat.Groves-1 is the real bull case, and nobody is pricing it
In August the company turned on Groves 1, and this quarter Jake Dewitte finally put a number on it: “we were able to do what we did for around $75 million all in” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-11-06 — roughly $65 million of build plus about $10 million of commercially sourced Framatome fuel. He contrasted that with the $300–500 million figures that defined isotope reactors when he entered the industry. Substantial construction took 229 days; first criticality came in under a year from groundbreaking. That is the single most important datapoint in the quarter, because it is the only hard evidence that Oklo's core thesis — that nuclear's cost structure is a choice, not a law of physics — is real rather than rhetorical. What Groves-1 actually bought, though, was not electricity. It bought an organization: safety bases, quality assurance, security programs, trained operators, DOE authorization experience. Dewitte is explicit that this is the transferable asset, and it is what the Aurora Idaho construction now draws on, where excavation is complete, structural steel is going vertical, and the preliminary documented safety analysis has been approved by DOE.The pivot: from fuel self-sufficiency to fuel merchant
For six quarters, fuel was framed defensively — an input Oklo had to control so it would not be the constraint. This quarter it became a product. Joseph Osha's question about third-party fuel revenue drew the clearest strategic statement the company has made: “our goal is to be the leading fuel fabricator for advanced fuel forms” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-11-06, spanning metallic fuel above LEU, oxide fuels, and next-generation nitride and zirconium-hydride forms — plus rented floor space for third parties to bring their own machines. The supporting pieces are all new. The fabrication capability at A3F is now being positioned for outside customers; there is an ARPA-E-aligned $6.8 million DOE selection to advance recycled fuel with Idaho National Laboratory; and the Nuclear Lifecycle Innovation Campus initiative — with more than $50 billion sitting in the nuclear waste fund at fiscal 2025 year-end — is the policy tailwind Dewitte says could unlock a recycling business at real scale. Timeline honesty matters here: asked for a first revenue date, he offered “some potential for early sort of services revenue that could come over the course of sort of the next 2 years” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-11-06. That is not near-term revenue; it is a call option on the fuel cycle.Two things management would rather you not notice
One: the clock moved. The 10-Q now frames Aurora-INL start-up as 2028 to 2029. Drew Scott asked directly whether that was a pushout; the answer ran long on electrical noise management, neutron detector calibration, and the need for schedule buffer rather than conceding the point — “having that extra capacity, that extra flexibility is critically important” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-11-06. It is a defensible engineering answer. It is also a different answer than the one Craig Bealmear gave in March, when he said “we're targeting a 2028 time line. We know it's an aggressive target” — Richard Bealmear, Chief Financial Officer · 2026-03-17. A year-over-year reading of the same promise has moved, and no CapEx number for the 75-megawatt unit has yet been provided — the second consecutive quarter that question went unanswered. Two: Meta got softer. In March, Jake called the Meta arrangement “a binding commitment to provide power and -- from us, for them to buy power from us” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-03-17. This quarter, asked whether the agreement is binding to build the reactor: “So it's not binding as is to build the reactor” — Jacob Dewitte, Co-Founder and Chief Executive Officer · 2026-11-06. Craig walked it back to binding elements around reservation capital and fuel-securing milestones. That is a meaningful distinction for a stock whose entire re-rating rested on a hyperscaler anchor.That is the paradox in one line: $3.1 billion of liquidity bought with roughly 58% share-count dilution since the SPAC, against a business whose operating loss reached -$73 million last quarter and whose capital expenditure ran to $94 million in a single quarter. The balance sheet is genuinely fortress-like — liabilities are just 2.5% of assets — which buys time, but time is not the same as a PPA.At September 30, we had $3.1 billion of cash, cash equivalents and marketable securities. ... The balance also reflects $2.3 billion raised through our ATM offerings, during the first 3 quarters of the year.