Constellation Energy vs Oklo
ComparisonThe explosive growth of AI infrastructure has turned nuclear energy into one of the most strategically important sectors in the agentic economy. Two companies sit at opposite ends of the nuclear spectrum: Constellation Energy, the largest producer of clean electricity in the United States with 60 GW of capacity following its landmark 2026 Calpine acquisition, and Oklo, a venture-backed startup developing 75 MW small modular fast reactors designed from the ground up for AI datacenter deployments.
The comparison between these two companies is really a question about timing and risk. Constellation delivers power today — its fleet of 14 nuclear stations already supplies baseload energy to Microsoft, Meta, and CyrusOne datacenters under long-term agreements. Oklo is building for tomorrow, with its Aurora powerhouse targeting first commercial deployment at Idaho National Laboratory by late 2027 and a massive 1.2 GW campus for Meta planned in Ohio by the early 2030s. Both are essential to the energy layer underpinning AI scaling, but they represent fundamentally different bets on how nuclear power evolves.
In March 2026, both companies are at inflection points. Constellation just completed a $5 billion asset divestiture to sharpen its nuclear-AI focus after the Calpine merger, while Oklo secured DOE approval of its Nuclear Safety Design Agreement and is moving toward submitting its combined license application to the NRC. Understanding where each company excels — and where it faces headwinds — is critical for anyone mapping the energy foundations of the AI era.
Feature Comparison
| Dimension | Constellation Energy | Oklo |
|---|---|---|
| Company Stage | Mature utility; largest U.S. clean energy producer with 60 GW capacity post-Calpine merger | Pre-revenue startup; first reactor targeting late 2027 deployment at Idaho National Laboratory |
| Reactor Technology | Fleet of conventional large-scale light-water reactors (avg. 40+ years old) with NRC license renewals extending operations to 2050s | Next-gen 75 MWe pool-type sodium-cooled fast reactor (Aurora powerhouse) using recycled HALEU metallic fuel |
| Current Generating Capacity | ~60 GW across nuclear, natural gas, solar, and wind (post-Calpine) | 0 MW operational; targeting 75 MW per unit, with plans to scale modularly to multi-GW campuses |
| Revenue & Profitability | Profitable; 2025 EPS ~$9.41 with projected 26% growth to $11.86 in 2026; pays growing dividend (10% annual increases) | Pre-revenue; 2025 operating loss of $139.3M; no dividend; $1.4B cash reserves as of Dec 2025 |
| Key AI Datacenter Contracts | 20-year Microsoft PPA (Three Mile Island restart); Meta agreement; CyrusOne 380 MW deal via Calpine | Meta 1.2 GW campus in Pike County, Ohio (first phase ~2030); multiple LOIs with datacenter operators |
| Regulatory Status | Fully licensed and operating; secured 20-year NRC renewals for Clinton and Dresden stations | NRC Principal Design Criteria topical report under accelerated review; DOE NSDA approved March 2026; combined license application pending |
| Three Mile Island / Flagship Project | Restarting TMI Unit 1 with $1B DOE loan; 80% staffed; targeting 2027 restart to supply Microsoft | Aurora-INL first commercial unit at Idaho National Lab; DOE Other Transaction Agreement signed; targeting late 2027 criticality |
| Fuel Strategy | Conventional low-enriched uranium supply chain; well-established procurement | Requires HALEU (High-Assay Low-Enriched Uranium); can use recycled nuclear waste fuel; supply chain still scaling |
| Scalability Model | Acquires and operates large existing assets; scales through M&A (Calpine) and plant life extensions | Factory-built modular reactors designed for serial deployment; scales by stamping out identical 75 MW units |
| Investment Risk Profile | Lower risk; proven cash flows, growing dividends, strategic moat via existing fleet and long-term PPAs | Higher risk; pre-revenue, execution risk on first-of-a-kind reactor, but massive upside if SMR model proves out |
| Carbon-Free Credentials | Largest U.S. carbon-free generator; nuclear fleet produces ~90% carbon-free electricity | Designed as zero-carbon from inception; uses recycled fuel, reducing nuclear waste footprint |
Detailed Analysis
Operational Maturity vs. Technological Innovation
The most fundamental difference between Constellation Energy and Oklo is where they sit on the technology readiness curve. Constellation operates the largest nuclear fleet in the United States — 14 stations generating reliable baseload power around the clock. Following its January 2026 completion of the Calpine acquisition, the combined entity commands roughly 60 GW of generating capacity, making it the largest private-sector power producer in the world. This is proven, bankable infrastructure that AI hyperscalers can contract against today.
Oklo represents the opposite bet: that the future of nuclear belongs to small, factory-built reactors that can be deployed modularly wherever power is needed. Its Aurora powerhouse is a 75 MWe sodium-cooled fast reactor — a fundamentally different design from Constellation's conventional light-water fleet. The technology promises lower construction costs, inherent passive safety, and the ability to use recycled nuclear fuel. But no Aurora unit has yet generated a single watt of commercial electricity, and nuclear projects have a long history of delays and cost overruns.
The AI Datacenter Power Race
Both companies are squarely positioned to serve the AI datacenter market, but through different mechanisms. Constellation's approach is immediate: it signed a 20-year power purchase agreement with Microsoft to restart Three Mile Island Unit 1, secured a 380 MW deal with CyrusOne through its Calpine subsidiary, and has agreements with Meta. These contracts leverage existing or restartable assets to deliver power within established grid infrastructure.
Oklo's datacenter strategy is longer-term but potentially transformative. Its agreement with Meta to develop a 1.2 GW nuclear campus in Pike County, Ohio — with Meta prepaying for power and funding early fuel procurement — represents a new model where reactors are purpose-built adjacent to compute facilities. If Oklo can deliver on its modular deployment vision, it could offer datacenter operators something Constellation cannot: dedicated, on-site nuclear generation that bypasses grid constraints entirely. The first phase targets 2030, with full buildout by 2034.
Financial Strength and Risk
The financial contrast is stark. Constellation is a profitable, dividend-paying utility with EPS projected to grow 26% in 2026 to $11.86 per share. It has increased its dividend 10% annually and its long-term contracts provide revenue visibility for decades. The Calpine merger temporarily pressured free cash flow — roughly $2 billion negative on a trailing basis — but the strategic $5 billion divestiture of gas assets to LS Power in March 2026 is designed to deleverage the balance sheet and sharpen focus on nuclear-AI synergies.
Oklo posted a $139.3 million operating loss in 2025 with zero revenue. It holds $1.4 billion in cash and marketable securities, providing several years of runway, but the path to commercial revenue depends entirely on successfully building and licensing its first reactor. For investors in the nuclear energy space, Constellation offers utility-grade stability while Oklo offers venture-style asymmetric upside — with commensurate risk.
Regulatory Pathways and Timeline Risk
Constellation benefits from decades of regulatory history. Its fleet operates under existing NRC licenses, and it recently secured 20-year renewals for Clinton (through 2047) and Dresden (through 2049-2051). The Three Mile Island restart, while unprecedented, follows an established NRC framework and received a $1 billion DOE loan in late 2025, signaling strong government support.
Oklo faces a more uncertain regulatory path. Its initial NRC combined license application was denied in 2022, forcing the company to resubmit. In 2025, the NRC accepted Oklo's Principal Design Criteria topical report under an accelerated timeline, and in March 2026, the DOE approved the Nuclear Safety Design Agreement for Aurora-INL. These are meaningful milestones, but Oklo is navigating a licensing framework designed for large light-water reactors — adapting it for advanced fast reactors introduces uncertainty that Constellation simply does not face.
The Fuel Supply Question
An often-overlooked dimension is fuel. Constellation's conventional reactors use standard low-enriched uranium with well-established global supply chains. Oklo's Aurora requires HALEU — uranium enriched to between 5% and 20% — which currently has extremely limited commercial production capacity. The U.S. Department of Energy is working to build a domestic HALEU supply chain, but scaling it to support a fleet of advanced reactors remains a multi-year challenge. Oklo's ability to use recycled nuclear fuel from decommissioned fast reactors (like the Experimental Breeder Reactor-II) is an innovative workaround, but availability is finite. This fuel constraint could become a bottleneck if Oklo successfully scales its reactor deployments.
Long-Term Strategic Positioning
Looking at the next decade, these companies may ultimately be complementary rather than competitive. Constellation provides the baseload energy that keeps today's AI agents and large language models running. Oklo is building the distributed nuclear infrastructure that could power the next generation of edge AI deployments and dedicated compute campuses. The energy demands of the agentic economy are so vast — projected to grow from roughly 4% of U.S. electricity to potentially 12-15% by the early 2030s — that the market likely needs both approaches to succeed.
Best For
Powering AI Datacenters Today (2026)
Constellation EnergyConstellation has operational nuclear plants and signed PPAs with Microsoft, Meta, and CyrusOne delivering power right now. Oklo's first reactor is still under construction. For immediate energy needs, there is no contest.
Dedicated On-Site Datacenter Generation
OkloOklo's modular 75 MW Aurora reactors are designed to be deployed directly adjacent to compute facilities, bypassing grid interconnection bottlenecks. Constellation's large centralized plants cannot offer this co-location model.
Stable Long-Term Energy Investment
Constellation EnergyWith growing dividends, $11.86 projected EPS in 2026, and decades of contracted revenue, Constellation offers utility-grade returns. Oklo is a speculative bet with no revenue and significant execution risk.
Next-Generation Nuclear Technology Exposure
OkloOklo's sodium-cooled fast reactor with recycled fuel represents the cutting edge of nuclear innovation. Investors seeking exposure to advanced reactor technology and the SMR revolution should look to Oklo.
Grid-Scale Clean Baseload Power
Constellation EnergyWith 60 GW of capacity and the nation's largest nuclear fleet, Constellation is the definitive provider of grid-scale carbon-free baseload. Oklo's 75 MW units would need hundreds of deployments to approach this scale.
Remote or Distributed Energy Applications
OkloOklo's compact reactor design is suited for locations where large-scale grid infrastructure is impractical — military installations, remote industrial sites, or greenfield datacenter campuses far from existing transmission.
Reducing Nuclear Waste
OkloOklo's fast reactor can consume recycled nuclear fuel, actually reducing the stockpile of spent nuclear waste. Constellation's conventional reactors generate standard spent fuel requiring long-term storage.
Powering the Agentic Economy at Scale (2026-2030)
Constellation EnergyThe agentic economy needs massive, reliable power now and over the next five years. Constellation's existing fleet and Three Mile Island restart (targeting 2027) make it the pragmatic backbone. Oklo's first commercial unit may not deliver power until 2028 at the earliest.
The Bottom Line
For anyone building or investing in the energy foundations of AI, Constellation Energy is the clear near-term winner. It is the only company that can deliver nuclear-scale clean power to hyperscaler datacenters today, and its post-Calpine position as the world's largest private power producer — combined with long-term contracts with Microsoft, Meta, and CyrusOne — gives it an unassailable strategic moat through at least the end of this decade. The Three Mile Island restart, backed by a $1 billion DOE loan and targeting 2027 operations, further cements its role as the energy backbone of the AI era.
Oklo is the more exciting long-term story but demands patience and risk tolerance. If its Aurora powerhouse proves commercially viable and the NRC licensing pathway clears, the modular SMR model could fundamentally change how datacenter power is provisioned — purpose-built reactors deployed directly where compute lives, bypassing the grid constraints that limit even Constellation. The Meta partnership validating a 1.2 GW campus is a powerful signal. But Oklo remains pre-revenue with first-of-a-kind execution risk, an unproven regulatory pathway for advanced reactors, and HALEU fuel supply uncertainty.
The pragmatic recommendation: treat Constellation as the foundation and Oklo as the frontier. The energy demands of AI inference at scale are so enormous that both companies address different segments of a market that is growing faster than either can serve alone. Constellation powers the present; Oklo could reshape the future. Most portfolios and strategic plans benefit from exposure to both — but if forced to choose one for the next five years, Constellation Energy's proven fleet, growing cash flows, and locked-in datacenter contracts make it the stronger bet.