Oklo Has Massive Potential Due to Artificial Intelligence, but the Stock Is Sinking. Here's What I'd Do

Oklo shares have lost roughly 50% in 2026 after rising 238% in 2025, as investor enthusiasm over its potential role in supplying AI data centers has faded. The company generated $153 million of losses over the past 12 months, more than double its $74 million loss in 2024, and may be years away from meaningful revenue or profitability. The article advises investors to keep Oklo on a watch list rather than buy now, citing substantial execution, demand, and financing risk.

Analysis

OKLO remains a duration-sensitive development asset rather than a clean AI-power proxy: absent operating cash flow, valuation is driven by licensing milestones, customer contract credibility, and repeated access to equity capital. The relevant downside is not merely further operating losses; it is dilution if the company must fund construction before obtaining financeable, long-dated power purchase agreements. A risk-off move in long-duration technology or a rise in real yields can compress the equity value disproportionately within days to weeks.

The better read-through is that data-center power scarcity benefits incumbents able to monetize capacity now or within a defined construction cycle. CEG, VST and NRG have existing generation fleets and can capture tightening power-market pricing, while GE Vernova (GEV) benefits from the equipment backlog associated with gas and grid buildout. SMR developers—including OKLO—compete for a limited pool of nuclear-qualified labor, fuel-cycle capacity, regulatory attention and customer deposits; bottlenecks raise project costs and favor better-capitalized or already-operating alternatives.

Over the next 1-3 months, monitor whether OKLO converts announced interest into binding, creditworthy PPAs with disclosed terms, interconnection status and customer-funded development capital. Over 6-18 months, the key falsifier for the bearish relative thesis is a funded project reaching a clear regulatory/construction pathway without material share issuance; that would shift the market from option value to a more financeable project pipeline. Consensus may be too dismissive of the strategic value of firm clean power, but the equity can still be structurally unattractive if that value accrues primarily to customers and project lenders rather than common shareholders.

Key Decisions for Investors

• Avoid adding directional OKLO exposure ahead of independently verifiable PPA, licensing and funding disclosures; treat any sharp rally on non-binding AI/data-center announcements as a potential trim or short-term short setup, with a stop on a fully funded, creditworthy contracted project.
• Express data-center power scarcity via a 3-6 month long CEG or VST / short OKLO relative-value position, sized modestly for high OKLO borrow and squeeze risk. The thesis earns if near-term power-market cash flows are rerated while pre-revenue nuclear optionality continues to dilute; exit if OKLO secures non-dilutive construction financing and binding contracted economics.
• For a diversified infrastructure expression, favor GEV over SMR developers for the next 6-18 months: grid and generation equipment orders monetize the buildout earlier and carry less binary licensing risk. Reassess if backlog conversion weakens or power-demand forecasts are revised down materially.
• Set an event alert around OKLO quarterly cash burn, share-count growth, customer deposits, NRC milestones and interconnection agreements. A material acceleration in dilution or absence of executable contracts after multiple reporting periods supports the relative short; disclosed project-level returns and committed financing invalidate it.