AI supply-chain thesis — mapping bottlenecks, focus companies, and supply-chain exposure for investors.
**Bottleneck theme:** Power & Grid **Focus:** $VST — Vistra Corp. ## Why this matters Power, not chips, is now the binding constraint on AI infrastructure. Hyperscaler 2026 capex is tracking >$700B (~50% YoY), but interconnection queues for new generation are 4-7 years long across PJM and ERCOT. The companies that already own large, dispatchable, carbon-free baseload — and can wrap it in a 20-year PPA — are pricing as scarce assets. Vistra is the largest competitive nuclear/gas operator in the U.S. and has converted that legacy-utility coal-and-gas footprint into a 24/7-clean-energy product hyperscalers will pay a premium for. ## The contracted-cash-flow conversion is happening fast Vistra's nuclear fleet (Comanche Peak in Texas; Perry, Davis-Besse, Beaver Valley in PJM after the Energy Harbor acquisition) has been re-priced through a series of landmark 20-year PPAs: - **Comanche Peak — 1,200 MW PPA** (Sep 2025, later confirmed structure aligned with AWS), ramping deliveries from late 2027 to full volume by 2032, with a 20-year extension option. - **Meta — 2,609 MW for 20 years** (Jan 2026): 1,268 MW from Perry, 908 MW from Davis-Besse, plus 213+80+140 MW of uprate capacity at Perry/Davis-Besse/Beaver Valley. - **Nuclear uprate program**: ~433 MW of incremental carbon-free MW being added through technical upgrades — i.e. new MW without the 10-15 year build-out timeline of a greenfield reactor. - Management estimates **8-10% accretion to Adjusted FCF before Growth** from these contracted volumes. ## Beyond nuclear: the integrated platform - **Lotus Infrastructure acquisition** added 2.6 GW of natural-gas capacity, useful for matching hyperscaler campus loads while clean-energy projects come online and for ERCOT firm-capacity revenue. - **Moss Landing** battery storage in California remains a benchmark asset — and battery economics are entering an ROI sweet spot as solar + storage clears more PPAs. - **Permian Basin gas units** — incremental capacity to serve the electrification of oil & gas operations and West Texas data-center load. - **Retail (TXU)** is a structural advantage: Vistra is one of few generators that owns a sizable retail book, smoothing merchant exposure and giving it real-time visibility into industrial and C&I demand including direct-to-customer hyperscaler conversations. ## Why the contracted-revenue mix matters The market historically penalized Vistra (and Constellation) as "merchant" power players exposed to ERCOT or PJM volatility. Each long-term hyperscaler PPA reduces merchant exposure, improves credit quality, and unlocks rerating toward regulated-utility multiples for the contracted block while preserving merchant upside on the rest. Hyperscalers are buying the certainty; Vistra keeps the optionality. ## Position in the supply chain Vistra is a buyer of nuclear fuel, gas turbines (from $GEV / Mitsubishi / Siemens), power transformers ($HUBB / $ETN / $HPS.TO / $ABB / $SIEGY), and grid equipment, and a seller of firm 24/7 power into utilities and directly into hyperscaler campuses ($MSFT, $META, $AMZN, $GOOGL). ## Risks - **FERC co-location ambiguity**: FERC rejected Talen's revised AWS co-location ISA in 2024, signaling regulatory friction around behind-the-meter hyperscaler power. Co-location is a key Vistra commercial path; rule changes could constrain monetization. - **Operational risk**: A multi-year nuclear outage (refueling extension, NRC issue) would materially impair contracted delivery and trigger PPA penalty provisions. - **Valuation**: VST has been an ~8x multi-bagger from 2023 lows; the easy money came in the rerating from merchant-coal to AI-power-darling. Forward returns require continued contracting and upgrade execution. - **Hyperscaler pause**: A genuine slowdown in hyperscaler capex would slow the next wave of PPA contracting. Existing 20-year PPAs are insulated; *new* MW signing isn't. - **Gas exposure & emissions narrative**: Despite the nuclear-led story, gas remains a meaningful portion of EBITDA. Carbon policy or regional gas-permitting battles could compress merchant gas margins. ## What we're watching - AWS / Comanche Peak co-location structure finalization and FERC resolution - Nuclear uprate execution (433 MW fully delivered) - Additional PPAs at Beaver Valley (management cited up to 3.2 GW of incremental nuclear contracting potential including ~200 MW Comanche Peak upgrades) - Lotus integration synergies and ERCOT capacity auctions - Any SMR or deployment partnership announcements (the 8-K options on new nuclear)
Browse all AI supply-chain theses · Macro trends · Industries · Product categories