Vistra Stock Sits 37% Below Its High While Power Demand Keeps Climbing. Should You Buy It?

Vistra reported Q2 adjusted EBITDA from ongoing operations of $1.77B, up 31% YoY, supported by higher realized power/capacity prices and contributions from recently acquired plants. Management reaffirmed 2026 adjusted EBITDA guidance of $6.8B–$7.6B and expects to land at or above the midpoint, alongside $3.9B–$4.7B of adjusted free cash flow (≈9% FCF yield at a ~$47B market cap). The stock is down ~37% from its $219.82 52-week high, but ~100% of 2026 generation volumes are hedged and the company has added long-duration nuclear contracting with Amazon (1,200MW for 20 years starting late-2027) and Meta (2,609MW for 20 years, deliveries starting late this year), improving revenue visibility despite power-price and data-center demand risks.

Analysis

The market is treating this like a crowded AI-power factor trade rather than a cash-flow story. That creates a mispricing window: VST’s contracted load growth and buyback cadence should support a much higher floor multiple than a generic merchant generator, but the stock still behaves like a beta proxy for power prices and data-center sentiment. The key second-order effect is that long-dated PPAs with AMZN/META tighten the supply of credible nuclear baseload for everyone else, which should improve pricing power for owners of dispatchable clean generation with balance-sheet capacity.

The bigger winner is not just VST; it is the subset of nuclear/dispatchable power names that can credibly sign 10-20 year contracts without heavy newbuild risk. CEG should benefit from the same scarcity premium, though VST has more explicit buyback support and visible contract backlog, making it the cleaner near-term relative-value long. Downstream, hyperscalers that cannot lock power may face higher interconnect and capacity costs, which can compress margins on marginal AI data-center projects if load growth outruns grid buildout.

Risk is mostly 1-3 months around sentiment and 6-18 months around execution. Near term, a pullback in power prices, a delay in data-center build schedules, or an outage at a key nuclear asset can hit the multiple faster than the underlying contracts can help. Longer term, the thesis is falsified if 2026 EBITDA guidance gets cut, if hedging losses keep obscuring cash conversion, or if PPAs fail to translate into incremental free cash flow because of maintenance capex and regulatory costs.

Contrarian view: the selloff may be partially justified because the market is probably underestimating how much of VST’s reported strength is already hedged or contract-backed, leaving less upside from spot power than bulls assume. But the drawdown also looks excessive versus the visibility embedded in the backlog; this is a cash-yield re-rating story, not a high-growth software story. If the stock cannot reclaim the prior breakdown area while power fundamentals stay firm, the market is signaling that the AI-power premium is still being unwound.

Key Decisions for Investors

• Buy VST on weakness for a 3-6 month mean reversion trade; target a re-rating toward ~15-16x forward earnings if power sentiment stabilizes, with invalidation on a 2026 EBITDA guide cut or sustained loss of the $130s support.
• Pair trade: long VST / short XLU for 1-3 months to isolate merchant/contracted power upside versus regulated utility dullness; thesis works if scarcity value in dispatchable generation gets repriced.
• Watch CEG as the cleaner sector confirmation trade; if CEG outperforms on a new data-center contract cycle, add to the independent power basket rather than chasing the whole group after a factor squeeze.
• For options, consider a bullish call spread in VST rather than outright equity if entering ahead of the next guidance update; use defined risk because volatility is likely to stay elevated while the AI-power trade de-risks.
• Set a falsification alert on any slowdown in hyperscaler capex or power-price rollover; if load-growth rhetoric stops translating into signed PPAs, trim energy-duration exposure across VST/CEG.