Why Oklo's Earnings Miss Has Not Broken the Nuclear AI Power Thesis, September 2026

Oklo's September 5 filings confirmed another pre-revenue quarter, but a reported 1.2 GW Meta nuclear deal, Trump administration backing, and $33 million in capital expenditure — up from zero in fiscal 2024 — have kept the investment thesis intact ahead of next week's regulatory catalysts.

The filing of Oklo's 10-Q and a pair of 8-Ks on September 5 confirmed what the market had been bracing for: the Aurora microreactor developer remains pre-revenue, with another development-stage quarter of losses on the books. Yet the stock — which had shed roughly 28% in a single month as recently as July 2026 amid investor frustration over a lack of tangible operational milestones — has largely held its ground, and observers on retail platforms are now pivoting toward next week's potential catalysts rather than dwelling on the shortfall. The divergence between the accounting reality and the prevailing market narrative captures the central tension in the nuclear AI power trade: the thesis is being priced on pipeline and policy, not on current cash flows.

The commercial scaffolding underpinning that thesis sharpened materially over August. On August 7, Meta announced a landmark 6.6 GW nuclear procurement spanning Oklo, Vistra, and TerraPower, locking in hyperscaler demand at a scale that no single SMR developer can currently supply alone. Oklo's portion is reported at approximately 1.2 GW — the company's largest disclosed offtake commitment and the clearest signal to date that the hyperscaler-to-SMR supply chain is moving from letters of intent toward binding contractual structure. The Trump administration reinforced the policy backdrop on July 30 by designating Oklo and X-Energy as preferred partners for accelerating nuclear power for AI data centers; EPA Administrator Lee Zeldin publicly touted SMR construction as a cornerstone of U.S. AI infrastructure on August 8. Together, those policy and demand signals have built a narrative resilient enough to survive repeated quarterly disappointments.

Oklo's balance sheet tells the story of a company only now beginning to translate ambition into physical infrastructure. Capital expenditure stood at zero in fiscal year 2024 and jumped to $33 million by the period ending March 31, 2026 — a 9,333% year-over-year increase that, while striking in percentage terms, still places Oklo sixth out of seven peers in absolute capex within the power category compiled by Slicast from SEC XBRL filings, with Constellation Energy leading by a wide margin. That capex ramp is the most concrete operational signal in Oklo's recent disclosures: the company has begun spending on property, plant, and equipment rather than engineering studies alone. The trajectory matters because Oklo's Aurora design — a compact, fast-spectrum reactor designed to use fuel recycled from existing nuclear waste — requires substantial site preparation and fuel qualification work before any commercial unit can enter service.

Headwinds are building from two directions at once. On the regulatory front, a September 2 report indicated that PJM Interconnection's grid-access ruling will govern the timeline for Oklo's Ohio project; any delay or conditional approval could push back the commercial operation date for the data-center-adjacent nuclear supply that hyperscalers are counting on. On the public-opinion front, a late-August analysis flagged growing data center backlash as a source of fresh regulatory and community resistance for both Oklo and NuScale — a risk that had been largely discounted during the nuclear AI euphoria of early 2026 but which can materially complicate siting approvals. The July selloff demonstrated how quickly sentiment can reverse: at its trough, Oklo shares had lost roughly 28% in about a month, driven by investor impatience with a timeline that has consistently expanded rather than compressed.

Three concrete signals will determine whether next week marks a turning point or another deferral. First, any regulatory update on Aurora's licensing process — whether through forthcoming 8-K disclosures or third-party reporting — would clarify the single largest binary risk in Oklo's path to first revenue. Second, the specifics of PJM's Ohio grid-access ruling will reveal whether the Meta deal's reported 1.2 GW tranche can hit its implied delivery window or faces slippage. Third, Oklo's Q3 filings will show whether the $33 million capex level established through March 2026 is being sustained or accelerated — a continued ramp would signal operational momentum, while a plateau would invite scrutiny of execution pace. The structural case for nuclear as the AI buildout's long-term power solution — underpinned by real hyperscaler capital commitments, a supportive federal posture, and the physics of carbon-free baseload — remains intact. Whether Oklo specifically can convert that structural case into operational results before investor patience thins again is the question the next several weeks of disclosures will begin to answer.