$AIENERGY **September Rebalance Summary** **Removed:** - NEE 7% to 0. Intermittent power, wrong product for 24/7 load - EPD 6% to 0. Weakest link to the actual constraint **Increased:** - GEV 11% to 16%. Tightest bottleneck, backlog 83 to 116 GW - BE 4% to 8%. First $1B quarter, guidance to ~100% growth - ETN 10% to 12%, VRT 9% to 11%, WMB 6% to 8% **Trimmed:** VST 18% to 16%, hedged near 100% of 2026 output, which caps rent capture. CEG 19% and TLN 10% unchanged.
June Rebalance Summary The AI power trade is in a sharp pullback (GEV -5.3%, VST -5.2%, CEG -4.2% on June 10), but fundamentals are intact: record backlogs, reaffirmed guidance, PPAs holding. No roster changes. Weight shifts only, buying weakness in highest-conviction names: • CEG: 18% → 20% (near 52-wk low, Crane/TMI restart decision in Sept) • TLN: 9% → 10% (MS target $499, buybacks, new PJM assets cleared) • EPD: 5% → 6% (defensive yield) • GEV: 12% → 10% (Vineyard Wind dispute, insider selling, capex concerns) • BE: 7% → 5% (most extended, highest beta) • VST, VRT, ETN, NEE, WMB unchanged Watch for July: NRC comment window closes July 8, Crane decision in September, and rising state-level resistance to data center construction, the new regulatory risk.
Cutting ETR, NI, PPL, and IDA — regulated utilities with weak direct AI exposure that were crowding out higher-conviction names. Adding 5 high-conviction AI infrastructure plays: • VST (16%) AWS + Meta nuclear PPAs • GEV (12%) gas turbines, $2B+ data center orders • VRT (10%) liquid cooling, 30% Q1 growth • ETN (10%) switchgear, 11-yr backlog • BE (7%) Oracle’s 2.8 GW fuel cell deal Boosting CEG to 18% (anchor post-Calpine, 55 GW) and keeping TLN, NEE, WMB, EPD trimmed.
cuts:PNW: Arizona is seeing data-center and advanced-manufacturing growth, but the current public case is still broader “Arizona growth” rather than a standout, named AI/data-center exposure story versus other options. * ET: good midstream business, but the current materials I found show strong general volume growth rather than a sharp, differentiated AI/data-center thesis. * Trim IDA: Idaho Power does have exposure, including Meta in Kuna, but the company itself says residential demand is the largest share of near-term year-over-year load growth, so I would keep it smaller if the objective is specifically AI/data centers. * Trim ETR: not because the thesis is weak, but because 22.68% is too large for a single utility. Entergy is one of the strongest direct winners here, with multiple data-center agreements and Louisiana projects tied to Meta and Hut 8. Add * CEG: this is the cleanest direct “AI needs firm power” name. Constellation signed a 20-year Meta deal tied to the Clinton Clean Energy Center, and its 2025 sustainability report explicitly frames AI and hyperscale data centers as a major electricity-demand driver. * TLN: higher risk, higher upside. Talen expanded its relationship with Amazon in 2025 to supply electricity from Susquehanna for AWS operations supporting AI and cloud technologies. * NEE: I would add this as a cleaner, lower-beta growth utility/energy platform. NextEra announced a partnership with Google to scale multiple gigawatts of data-center capacity and energy infrastructure, and FPL has a large-load tariff designed to support data-center growth.