I think $Vistra(VST.US) looks compelling here.

At $138, Vistra feels like one of the cleaner ways to play the AI/data center power trade without paying a large multiple.

The numbers:

- Q2 adjusted EBITDA: $1.77B, +31% YoY

- 2026 EBITDA guide: $6.8B-$7.6B

- 2026 FCF before growth: $3.9B-$4.7B

- 2027 core EBITDA opportunity: $7.4B-$7.8B

- Shares outstanding down ~30% since 2021

The obvious bottleneck continues to be energy and it seems like $Vistra(VST.US) is attacking that head on:

- AWS signed a 20-year deal for up to 1.2 GW from Comanche Peak.

- Meta signed 20-year agreements covering 2.6+ GW across Vistra’s nuclear fleet.

Roughly 3.8 GW tied to AWS + Meta alone. Then there is also Helix, the AI infrastructure platform with KKR, NVIDIA and KIA. Vistra is the preferred power provider for this.

VST is also acquiring Cogentrix, adding ~5.5 GW of gas generation, with the deal expected to be accretive to FCF/share starting in 2027.

The valuation is compelling:

At $138:

~13-15x forward EPS

~9-10x 2026 EV/EBITDA

~8x 2027 EV/EBITDA

~10-11x 2026 FCF before growth

Meanwhile $Constellation Energy(CEG.US) trades closer to ~22x forward earnings and ~14-15x EV/EBITDA.

I don’t think VST deserves CEG’s full multiple, but the gap looks wide given how much exposure VST has to the same nuclear/power scarcity theme. At the same time, CEG has almost the same amount of debt as VST at $19B. CEG has much more hyperscaler exposure so it deserves a higher premium, but it seems likely for VST to be rerated on solid execution.

The risks:

- Merchant power prices can weaken

- Debt/leverage is meaningful

- Nuclear outages/regulatory issues can hurt earnings

- AI power demand is already a crowded theme

- Heavy hedging limits near-term upside from higher spot power prices

- Cogentrix/Helix still need to execute

- More generation or slower AI capex could reduce the scarcity premium

The risks are important here but they are mainly tied to regulation/execution, the underlying thesis is that energy continues to be a huge bottleneck.

What I also like is that you don’t need some crazy AI multiple for this to work and deliver returns from here as the stock is already down 15% YTD:

- Bear: $110

7x EBITDA / ~$7B EBITDA

- Base: $185

9x EBITDA / ~$8B+ EBITDA

- Bull: $240

10-11x EBITDA / ~$9B EBITDA

$Vistra(VST.US) has $1.22B remaining on its buyback authorization and 336M shares outstanding, after repurchasing $778M of stock in 2026 and $6.5B since November 2021, helping reduce its share count by roughly 30% since 2021. That’s actually interesting now because the stock is around $138 so if management keeps allocating FCF to repurchases at these lower prices, every $1B of buybacks would retire roughly 7.2M shares, or a little over 2% of the current share count.

Finally, Peter Thiel bought $59M of it in Q2 and David Tepper added to his position by 10% making it now worth 4% of his portfolio with a $350M position size.

I took a small starter position today to keep track of it as I'm still doing DD but it does feel more compelling than not.

Source: amit