Why Vistra's 28% Slide Persists Despite Meta's 6.6 GW Nuclear Deal

Vistra's capital expenditure hit $2.75 billion in FY2025 — up 32.4% year-over-year and 16% of revenue, the highest intensity among tracked power peers — yet VST shares have shed roughly 28% over the past twelve months, a gap that frames the central debate for analysts evaluating whether dedicated AI load premiums will materialize before the investment cycle peaks.

On September 5, 2026, Vistra Corp filed multiple 8-K reports and its latest 10-Q with the SEC — a dense cluster of disclosures arriving at an unusual inflection point for the Texas-based power company. Just four weeks earlier, on August 7, Meta announced a 6.6-gigawatt nuclear power agreement spanning Oklo, Vistra, and TerraPower, one of the largest dedicated AI load commitments the industry has yet produced. The deal appeared to confirm Vistra's position at the convergence of two of the most capital-intensive buildouts underway in the United States: nuclear power and artificial intelligence infrastructure. Yet even with that landmark contract in the background, VST shares have declined approximately 28% over the trailing twelve months, according to FinanceFeeds analysis published August 22. That divergence is the central question now in front of analysts.

The investment cycle Vistra is running is measurable and, by the data, still accelerating. Capital expenditure reached $2.75 billion in FY2025, up 32.4% from $2.08 billion the prior year, representing 16% of the company's $17.59 billion in FY2025 revenue — the highest capex intensity among the six power companies tracked by Slicast on this metric. In the first quarter of 2026 alone, Vistra deployed $883 million in property, plant, and equipment spending, a quarterly figure that, if sustained, implies an annualized run rate above $3.5 billion. By absolute capex, Vistra ranks second among seven tracked power companies, trailing only Constellation Energy (CEG), whose nuclear-heavy portfolio sets the sector benchmark. The pattern of SEC 8-K agreement and financing filings logged nearly every quarter going back to late 2023 underscores a company that has been systematically restructuring its balance sheet and deal architecture well ahead of what management appears to treat as a multi-year demand inflection.

That inflection thesis is not universally shared. Analysts at Quiver Quantitative observed in late August 2026 that utility valuations are beginning to decouple from traditional utility metrics, reflecting a premium pricing mechanism tied to exclusive AI load supply. The logic is straightforward in principle: a hyperscaler willing to sign a long-term power purchase agreement for dedicated capacity — as Meta has now done — removes volume and price risk from the generator's revenue model, a feature that historically commands a premium in contracted infrastructure. What complicates Vistra's case is its mixed footprint. The company operates across both nuclear baseload and renewables, meaning its economics straddle the stable, contracted world that AI buyers seek and the merchant power markets that remain subject to weather events, natural gas prices, and regional grid dynamics on ERCOT. Quiver Quantitative analysts were already noting strong demand signals from AI data centers as early as late June 2026, but the August commentary introduced the more nuanced observation that premium valuations require premium execution — and execution timelines for nuclear capacity remain uncertain.

The bull case, as structured by FinanceFeeds analysts reviewing the stock in late August, rests specifically on the nuclear assets. Nuclear generation is baseload, dispatchable, and carbon-free, making it the preferred power source for hyperscalers constrained by sustainability commitments and grid reliability requirements. The Meta deal — spanning three distinct nuclear developers — signals that the largest AI infrastructure buyers have moved well beyond spot market procurement and are willing to commit to decade-scale agreements. For Vistra, anchoring a portion of its nuclear fleet to that demand profile would convert what has historically been a volatile merchant revenue stream into something closer to a regulated utility's contracted cash flow. The bear case is equally structured: the 28% stock decline reflects genuine uncertainty about the pace at which nuclear capacity can realistically be contracted, extended, or brought to commercial operation, and whether the premium pricing that the decoupled-valuation framework promises will arrive before capital costs peak. Regulatory risk around nuclear permitting and the inherent complexity of multi-party deals of the scale Meta has announced add further variables that the market appears to be pricing, not dismissing.

Three signals will shape which scenario dominates over the next twelve to eighteen months. First, the substance of Vistra's September 5 SEC disclosures — particularly whether the cluster of 8-Ks reveals additional hyperscaler agreements, new financing structures, or material operational developments — will indicate whether the Meta deal has opened a repeatable pipeline or remains a landmark exception. Second, the trajectory of capital expenditure through the remainder of 2026: quarterly spending above $800 million signals continued infrastructure commitment but also sharpens the return-on-capital question; a material moderation would suggest either project-level delays or a deliberate shift toward capital discipline. Third, regulatory and permitting progress on nuclear capacity in Texas and Vistra's other operating markets will set the outer boundary on when contracted AI power revenue can actually begin to flow. Until those data points clarify, the gap between the AI infrastructure narrative and Vistra's stock performance will remain an open analytical question rather than a settled market verdict.