Not Oklo. Not NuScale. This Nuclear Stock Is Going to Be the Biggest Winner of the AI Power Boom.

Vistra is presented as a lower-risk beneficiary of AI data-center electricity demand versus pre-revenue nuclear developers Oklo and NuScale. Vistra has 44 GW of generation capacity, including 6.4 GW of nuclear capacity after acquiring Energy Harbor, while natural gas represents 62% of capacity; analysts project 2025-2028 revenue and EPS CAGRs of 13% and 75%, respectively. Despite regulatory limits on PJM capacity pricing through 2030 and Moss Landing battery-storage closures, Vistra shares are down more than 30% over 12 months and trade below 16x current-year earnings.

Analysis

The relevant valuation gap is not simply operating assets versus pre-revenue reactor concepts; it is contracted, dispatchable power in constrained regional grids versus long-duration technology optionality. VST's upside depends disproportionately on ERCOT and PJM forward capacity/energy curves, hedge-book roll-off, and data-center load arriving where its generation is physically deliverable. A generic AI-power narrative will not lift all generators equally: transmission congestion and interconnection delays should increase the value of incumbent, load-pocket capacity, favoring VST and other merchant exposure such as CEG, NRG, and TLN over developers without operating assets.

The PJM capacity-price limitation is the central earnings-risk variable, not a minor regulatory footnote. It can cap upside to merchant scarcity rents through 2030, while higher gas prices may be passed through only with a lag or absorbed under fixed-price retail contracts; this makes consensus EPS growth unusually sensitive to plant availability, hedging disclosures, and auction outcomes. Moss Landing also raises a broader battery-insurance and permitting risk that could constrain valuation credit for storage expansion for 1-3 quarters.

Near term, a 30% drawdown may create a tactical rebound if quarterly EBITDA, nuclear availability, and forward-power marks stabilize. Over 6-18 months, the better contrarian framing is that AI load forecasts are being capitalized before actual grid connections: hyperscaler announcements are not equivalent to energized demand. If data-center commissioning slips or ERCOT/PJM reserve margins improve through new supply, VST's multiple can compress despite positive EPS revisions; the same outcome is more damaging to OKLO and SMR, whose funding dilution risk rises as rates or equity volatility increase.

Key Decisions for Investors

• Initiate a 6-12 month long VST / short equal-dollar OKLO basket only after VST confirms stable generation EBITDA and no incremental Moss Landing reserve charge. The pair isolates cash-generating merchant-power exposure from long-dated reactor valuation; target 20-30% relative return, with exit if VST forward EBITDA falls more than 10% or OKLO secures fully financed, binding customer-backed construction commitments.
• For a lower-beta AI-power allocation, prefer CEG over VST on a 6-18 month horizon where nuclear baseload and contracted data-center demand are the objective; retain VST only for investors seeking higher merchant-power torque. Monitor regional basis and capacity-auction results rather than aggregate U.S. data-center forecasts.
• Avoid adding to OKLO or SMR on narrative-driven strength until there is independently verifiable project financing, regulatory milestones, and firm offtake with creditworthy counterparties. A reactor-development position should be treated as venture-style optionality and sized accordingly; dilution or a 12+ month licensing/construction delay is the key downside catalyst.
• Set a VST review trigger around PJM regulatory developments and next quarterly hedge/availability disclosures. Reduce exposure if management indicates materially lower unhedged merchant realizations, higher battery remediation costs, or retail-margin pressure from gas-price volatility.