What you need to know
- Vistra's business had a strong quarter: cash earnings (Ongoing Operations Adjusted EBITDA) up 31% year over year, full-year guidance reaffirmed, and roughly 30% of its shares retired through buybacks.
- The stock is down about 38% from its high and sitting on its 52-week low, in a clear downtrend below both major moving averages.
- The gap is explained by an unresolved regulatory question in the PJM power market about plugging data centres straight into power plants, plus the unwinding of the 2024–25 "AI needs power" trade.
- The stance: a name to watch, not chase. The setup only becomes actionable if the $132.66 low holds and the regulatory picture clears; a decisive close below $132.66 invalidates it.
The Business
Vistra (NYSE: VST) is an integrated power company — it both generates electricity and sells it to retail customers, which smooths out the wild swings a pure merchant generator would face. Two pieces:
- Generation fleet. A big, dispatchable mix — natural gas, nuclear, coal, plus growing solar and battery storage — spread across Texas (ERCOT) and the East (PJM). The nuclear is the crown jewel: Comanche Peak in Texas, plus the PJM plants it picked up buying Energy Harbor. Nuclear is exactly the 24/7 carbon-free baseload hyperscalers want to power data centers.
- Retail. One of the largest competitive retail electricity businesses in the country, from California to Maine. This is the hedge — when it owns both the power plant and the customer, it's less exposed to raw commodity swings than a merchant generator.
The thesis in one line: US electricity demand is inflecting for the first time in decades — data centers, electrification, and reindustrialization all pulling at once — and the companies that own dispatchable generation (especially nuclear) are the landlords of that scarcity. Vistra is one of the biggest.
Where the Price Is
As of the August 28 session:
- Price: $136.70 (−2.2% on the day)
- 52-week range: $132.66–$219.82 — the stock is only ~3% above its 52-week low and 37.8% below its high
- 50-day MA: ~$152 — price is ~10% below it
- 200-day MA: ~$159 — price is ~14% below it
- RSI(14): 39.3 — approaching oversold, not yet turned
This is the opposite technical picture from the AI-chip names. VST isn't a high-flyer digesting a beat — it's a broken-momentum name grinding along its lows, below both major moving averages. The chart is in a downtrend, full stop. Any bullish case here is a value/contrarian case, not a momentum one, and the tape is not yet confirming it.
What Q2 Actually Said
Reported August 7 (quarter ended June 30). The disconnect between the business and the stock is the whole story:
- Ongoing Operations Adjusted EBITDA: up 31% YoY — this is the real cash-generating metric, and it surged
- By region: Texas EBITDA up 119% to $311M; East up ~53% — broad-based strength, and notably strongest in ERCOT where power demand is booming
- GAAP net income: took a hit from a ~$488M year-over-year swing in unrealized mark-to-market derivative (hedging) losses. That's non-cash accounting noise, not a deterioration in the business — but it's why the GAAP headline looked ugly
- Capital return: aggressive — Vistra has retired roughly 30% of its shares through buybacks, with more authorization to deploy through 2027, plus a (modestly raised) $0.23 quarterly dividend. Fewer shares means every dollar of that growing EBITDA is spread over a smaller base
- Guidance: reaffirmed full-year 2026 (~$6B+ Ongoing EBITDA). Analysts model roughly $9 in EPS this year and ~$11 next
So: cash earnings up double digits, guidance intact, shares shrinking fast. On the fundamentals, this was a strong quarter. The stock did not care.
Why It Got Cut By a Third
Three forces, in order of how much they matter.
The PJM regulatory overhang. This is the specific thing spooking the whole IPP group. Much of the AI-power thesis rested on "co-location" and behind-the-meter deals — plugging a data center straight into a power plant. Regulators in PJM (via the amended interconnection-service-agreement fight at FERC) have raised hard questions about whether and how those deals are allowed. That uncertainty froze a chunk of the deal flow the market had already priced in. Notably, management pointed out that ERCOT — Texas, where Comanche Peak sits — doesn't fall under that PJM jurisdiction, which is a real advantage. But the cloud is over the sector.
The AI-power trade unwound. VST, Constellation, Talen, GE Vernova and the rest went vertical in 2024–25 on the nuclear/AI narrative. When momentum money rotated out, the names that ran the most gave back the most — and VST round-tripped a huge move regardless of the quarterly numbers.
The "is this multiple even real" question. Under the AI story, the market paid a growth multiple for what is, structurally, a capital-intensive, commodity-exposed power business. The de-rate is partly the market re-asking whether an IPP deserves a premium at all. Tellingly, management itself said the business's fundamentals "don't depend on doing a lot of data center deals" — reassuring if you believe Texas demand carries it, but also a quiet admission that the data-center deal flow is slower than the hype implied.
The Thesis
Bull case: you're buying a structural winner in dispatchable power — gas plus scarce nuclear — at ~15x this year's earnings and ~12x next year's, down 38% from the high, while the company grows cash EBITDA 31% and retires 30% of its own shares. The load-growth story (data centers + electrification + reindustrialization) is real and multi-year, ERCOT sidesteps the PJM mess, and the buyback is a powerful per-share tailwind that works even if the stock stays cheap. If the power-demand thesis is even half right, this is a beaten-down compounder.
Bear case: it's still a capital-intensive merchant-power business whose earnings swing with power and gas prices, dressed in an AI costume the market is now removing. The PJM regulatory question is unresolved and could keep a lid on the co-location deals the premium assumed. GAAP earnings are noisy, the chart is in a clear downtrend at the lows, and "cheap" can stay cheap — or get cheaper — if power prices soften or the data-center buildout digests.
What would change my mind: a clean resolution of the PJM/FERC co-location rules (either direction — clarity itself would help), a signed marquee data-center power deal (especially at Comanche Peak), or the stock printing a higher low above its 52-week low with RSI turning up. On the downside, a decisive break of the $132.66 low, or a cut to that reaffirmed guidance, would say the bear case is winning.
Trade Plan
Setup: deep-value/contrarian in a broken chart. This is a "wait for the turn," not a "it's cheap so buy it" — cheap names in downtrends are how value investors get run over. Momentum is down; the plan is to buy confirmation, not to guess the bottom.
Levels that matter:
- $132.66 — the 52-week low and the line in the sand. Holding it (a higher low, a reversal candle with follow-through) is the earliest credible base signal. Losing it decisively opens fresh lows and voids the setup.
- ~$152 (50-day) — first overhead resistance. A daily close back above it on volume is the "trend is repairing" trigger to size up, not to start.
- ~$159 (200-day) — the bigger hurdle; reclaiming it would mark a genuine trend change.
How I'd treat it: starter only on evidence — an RSI curl back above ~45 while holding the 52-week low, or a higher low off ~$133. Add on a 50-day reclaim. Do not commit full size into a name making 52-week lows with an unresolved regulatory catalyst hanging over it. The buyback provides a fundamental floor over time, but buybacks don't stop a downtrend in the short run.
Invalidation: a decisive close below $132.66 on elevated volume. Below that there's no visible support until lower, and the "value" thesis needs to wait for a new base.
Horizon: months to quarters — this is a thesis trade on power-demand growth and a regulatory de-cloud, not a swing. The PJM/FERC resolution and any Comanche Peak deal news are the catalysts to watch.
Bottom Line
Vistra is a genuinely strong business — 31% EBITDA growth, reaffirmed guidance, and a buyback shrinking the share count by nearly a third — trading like a broken one, because the market yanked the AI premium and a real PJM regulatory cloud froze the deal narrative it had leaned on. That's a much more interesting setup than a stock priced for perfection: the pessimism is doing the work the froth used to. But "cheap and strong" isn't a buy signal while the chart is making new lows and the regulatory question is open. This is a name to stalk, not chase — let the 52-week low hold and the PJM fog lift, then act. The power-demand wave is real; the debate is whether Vistra's stock has finished pricing out the hype before it starts pricing in the cash. I'd want the tape, not just the valuation, to agree before stepping in.
Terms used here
- EBITDA: earnings before interest, taxes, depreciation and amortisation; a rough measure of the cash a business generates from operations. "Ongoing Operations Adjusted" is Vistra's version that strips out one-off and non-cash items.
- IPP: independent power producer, a company that generates and sells electricity rather than a regulated utility.
- PJM and ERCOT: the two electricity grids that matter here. PJM runs the Mid-Atlantic and Midwest; ERCOT runs most of Texas and is not under the same federal jurisdiction.
- 50-day and 200-day moving averages: the average closing price over the last 50 or 200 trading days; a price below both usually means the trend is down.
- RSI: relative strength index, a momentum gauge from 0 to 100. Below about 30 is called oversold; the 39 here means weak but not yet extreme.
- Buyback: a company buying its own shares, which spreads future earnings over fewer shares.
Sources and data notes
- Price and technical data: Yahoo Finance, August 28, 2026 session. Levels are point-in-time and will move.
- Fundamentals: Vistra's second-quarter 2026 results, reported August 7, 2026, and earnings coverage. EPS and EBITDA figures are approximate and partly analyst-estimated.