Small Modular Reactors and the AI Power Problem
AI data centers face an electricity problem that solar, wind, and batteries alone cannot solve. Training large AI models requires constant, reliable power measured in hundreds of megawatts or gigawatts, running 24 hours a day regardless of weather. Nuclear energy provides this baseload power with zero carbon emissions. But traditional nuclear plants cost $10-20 billion, take 10+ years to build, and generate 1+ gigawatt each, far more than a single data center site needs.
Small modular reactors (SMRs) promise a different approach: factory-manufactured units producing 50-300 megawatts each, deployable in years rather than decades, at costs measured in hundreds of millions rather than tens of billions. Oklo's Aurora reactor takes this further with a fast reactor design that uses recycled nuclear fuel and runs for up to 10 years between refueling. The small footprint and long operating cycle make it suited for data center campuses that need reliable, on-site power generation.
The Aurora Reactor Design
Oklo's Aurora uses a liquid metal-cooled fast reactor design, distinct from the light water reactors that dominate the existing nuclear fleet. Fast reactors can use recycled nuclear fuel from existing reactors, addressing both the fuel supply question and the nuclear waste challenge. The design is intended for a 10-year refueling cycle, meaning once installed, the reactor operates continuously for a decade before requiring fuel services.
The first commercial Aurora Powerhouse is being built at the Department of Energy's Idaho National Laboratory, with targeted operation in 2027-2028. This site provides the regulatory and safety framework for demonstrating the technology before commercial deployment at customer sites. If the Idaho demonstration succeeds, Oklo would begin manufacturing reactors at scale for the Meta, Switch, and other data center agreements.
Financial Performance
- •Revenue Status: Pre-revenue; Oklo has no commercial reactors operating yet
- •Power Agreements: Meta (up to 2.8 GW), Switch (up to 12 GW); these are nonbinding agreements contingent on reactor availability
- •Stock Range: 52-week low $17.42 to high $193.84, reflecting extreme speculative volatility
- •Capital Structure: Went public via SPAC merger; ongoing capital needs for reactor development and NRC licensing
- •CEO Share Sales: Jacob DeWitte sold 840,000 shares (~$69M) at approximately $82/share
- •Timeline: First commercial reactor targeted 2027-2028 at Idaho National Lab; Meta reactors earliest 2032
Growth Catalysts
- •Data Center Power Demand: AI infrastructure requires hundreds of gigawatts of new baseload power; nuclear is the only zero-carbon technology that provides 24/7 reliability at scale
- •Meta and Switch Agreements: If executed, these agreements alone represent gigawatts of reactor orders; each additional hyperscaler deal expands the pipeline
- •NRC Licensing Progress: Successful resubmission and approval of the Aurora design would remove the single biggest execution risk
- •DOE Support: Idaho National Lab partnership and federal interest in advanced nuclear provide regulatory and technical support
- •Nuclear Fuel Recycling: Using recycled fuel addresses both supply concerns and waste disposal, creating a differentiated value proposition versus conventional reactors
Risks and Challenges
- •NRC Licensing Uncertainty: The NRC previously denied Oklo's license application; there is no guarantee the resubmission will succeed, and the regulatory process could take years
- •No Operating Reactor: Oklo has never built or operated a commercial reactor; the technology is untested at commercial scale
- •Timeline Risk: The first reactor targets 2027-2028; Meta reactors are not planned until 2032; significant delays are possible given nuclear construction history
- •Nonbinding Agreements: The Meta and Switch deals are nonbinding; customers can exit if Oklo cannot deliver reactors on time or at projected costs
- •Extreme Valuation: A pre-revenue company with no operating reactor trading at multi-billion dollar market cap; the stock is priced on hope rather than earnings
Competitive Landscape
The small modular reactor market is crowded with well-funded competitors. NuScale Power (SMR) received the first SMR design certification from the NRC but canceled its first project due to cost overruns. Bill Gates' TerraPower is building a sodium-cooled reactor in Wyoming with DOE support. X-energy is developing a high-temperature gas reactor. Kairos Power is building a molten salt test reactor. Each company takes a different technological approach, and none has delivered a commercial unit yet.
Oklo's differentiation is the fast reactor design with fuel recycling and the AI data center focus reinforced by Sam Altman's involvement. The Meta and Switch agreements are among the largest announced commitments to SMR power for data centers. However, competitors like TerraPower have more advanced regulatory status and larger engineering teams. The race to deliver the first commercial SMR will likely determine which company captures the majority of data center power contracts.
Who Is This Stock Suitable For?
Perfect For
- ✓Speculative investors with high risk tolerance who believe small modular reactors will power the AI data center buildout
- ✓Those seeking early-stage exposure to the nuclear energy renaissance driven by decarbonization and AI power demand
- ✓Investors with a 5-10 year time horizon who can wait for reactor commissioning and commercial revenue
- ✓Portfolio allocations toward transformative energy technology with potentially outsized returns
Less Suitable For
- ✗Risk-averse investors (pre-revenue, untested technology, NRC licensing uncertainty)
- ✗Income investors (no revenue, no dividend, no near-term path to profitability)
- ✗Value investors (multi-billion dollar market cap for a company with no operating reactors)
- ✗Short-term traders who cannot withstand 50%+ price swings driven by regulatory news and sentiment
Investment Thesis
Oklo sits at the intersection of two powerful trends: the nuclear energy revival and the AI data center power crisis. The Meta and Switch agreements demonstrate that hyperscalers are willing to commit to nuclear power for their future data center needs. Sam Altman's backing connects the company to the AI industry's demand side. The Aurora reactor's fuel recycling and 10-year refueling cycle are technically differentiated features.
But this is a pre-revenue company that has never built a reactor, was denied its first NRC license, and is competing against multiple well-funded startups and established nuclear companies. The 2027-2028 Idaho timeline is ambitious for a first-of-a-kind reactor, and the Meta deal's 2032 target is more than six years away. CEO DeWitte's significant share sales add a note of caution. Oklo is appropriate only as a speculative allocation for investors who understand that the downside could be near-total if licensing fails, while the upside could be extraordinary if the technology delivers.