VST

Vistra is one of the cleanest ways to own the US power-demand supercycle: an integrated merchant generator with a ~44 GW fleet spanning Texas (ERCOT) and PJM, anchored by four nuclear stations. The forward story materially improved in 2025 — management locked in a 20-year, 1,200 MW carbon-free PPA with AWS from Comanche Peak and (Jan 2026) 20-year PPAs with Meta for 2,609 MW from its PJM nuclear plants, converting merchant nuclear volume into contracted, price-certain cash flow just as data-center, oilfield-electrification and EV load are driving 'fast paced load growth' in its regions. S&P's December 2025 upgrade to investment grade (BBB-) validates a de-risked balance sheet, and capital allocation is genuinely shareholder-friendly: $5.9B / 167M shares bought back to date with $1.8B remaining, plus a $306M dividend. Operating cash flow held up at $4.07B and ROE is a healthy 18.5%.

The catch is that the reported earnings look ugly and the valuation is full. FY2025 net income fell 64.5% to $944M and diluted EPS dropped 68.9% to $2.18 even as revenue rose 19.1% to $17.6B — a collapse driven far more by non-cash mark-to-market swings on the hedge book than by any operating deterioration (OCF of $4.07B versus $944M of net income tells that story). That distortion cuts both ways: the headline 70x P/E overstates how expensive the stock is, but even normalizing to cash the setup isn't cheap. With ~$18.8B of debt on top of $51.6B of equity value, EV/OCF sits in the mid-teens, and after $2.75B of rising capex free cash flow (~$1.3B) is a thin ~2.6% yield on the cap. P/S of 2.9 is rich for a capital-intensive utility-adjacent business.