Are AI Datacenters Increasing Electric Bills for American Households?

Aishwarya Mahesh · SemiAnalysis · March 03, 2026 at 14:27 · ⏱ 33 min read  | Read on Substack ↗
Summary
AI datacenters are not the root cause of rising household electric bills in PJM; instead, PJM's flawed capacity market design — specifically a simulated auction called the BRA — is responsible for a ~15% bill increase. ERCOT in Texas has absorbed similar datacenter load growth with stable prices due to its energy-only market and regulatory agility. This means market design, not AI demand, determines consumer impact, and the structural asymmetry between PJM and ERCOT creates investable implications for power generators and equipment suppliers.
  • •PJM capacity prices (BRA) spiked 9.3x from $29/MW-day to $270/MW-day in 2025/26, driven by a simulated VRR curve based on PJM's own load forecast.
  • •PJM's internal market monitor estimated that removing all datacenters from the forecast would reduce capacity payments by 64% ($9.33B), yet the forecast itself is flawed—PJM repeatedly cuts datacenter load projections by 800MW–1.1GW year-over-year.
  • •Forward energy prices in PJM (Western Hub) rose only 12-20% over the same period, contradicting the panic reflected in the capacity auction—real market participants do not confirm the simulated spike.
  • •ERCOT's projected datacenter load more than doubled to 77.9 GW by 2030, but the market ignored it; ERCOT applies aggressive haircuts (49.8-55.4%) to developer requests and uses real-time scarcity pricing, not a forward auction.
  • •Winter Storm Fern (Jan 2026) tested both grids: PJM lost 21 GW of generation (15% of capacity) despite record-high capacity payments, needing DOE emergency orders; ERCOT held without crisis, with real-time prices peaking at ~$300/MWh.
  • •ERCOT's SB 6 curtailment authority and single-state regulation allow faster market adaptation, while PJM's 13-state + FERC oversight makes reform slow and politically contentious.
  • •The capacity auction in PJM costs households an estimated $25-30/month more than two years ago, totaling ~$16 billion annually across the region.
  • •The article identifies IPPs (Vistra, Constellation, Talen), equipment suppliers (GEV, CAT, Bloom Energy, Vertiv), and datacenter developers/cryptominers as likely winners from shifting market bottlenecks.
Read time 33 min
Length 33,324 chars
Category finance
Trade Ideas
Aishwarya Mahesh Substack author, SemiAnalysis
The article states that shifting market bottlenecks 'impacts major AI winners such as IPPs (Vistra, Constellation, Talen..)' – implying that these independent power producers benefit from PJM's capaci
The article states that shifting market bottlenecks 'impacts major AI winners such as IPPs (Vistra, Constellation, Talen..)' – implying that these independent power producers benefit from PJM's capacity price surge and market constraints. Risk: Regulatory price caps or FERC interventions could cap capacity revenues; also Vistra has significant PJM exposure which may face political backlash.
VST
Aishwarya Mahesh Substack author, SemiAnalysis
Vertiv is cited as an 'equipment supplier' that benefits from the datacenter buildout and the shifting market bottlenecks, as datacenters require cooling, power distribution, and infrastructure upgrad
Vertiv is cited as an 'equipment supplier' that benefits from the datacenter buildout and the shifting market bottlenecks, as datacenters require cooling, power distribution, and infrastructure upgrades. Risk: Customer concentration among hyperscalers; supply chain constraints for components.
VRT
Aishwarya Mahesh Substack author, SemiAnalysis
Talen Energy is explicitly listed as an IPP winner in the article's paywall tease, indicating it is positioned to gain from the capacity auction dynamics.
Talen Energy is explicitly listed as an IPP winner in the article's paywall tease, indicating it is positioned to gain from the capacity auction dynamics. Risk: Talen's reliance on PJM markets; any reform to capacity market rules could negatively impact revenues.
TLN
Aishwarya Mahesh Substack author, SemiAnalysis
The article identifies 'equipment vendors of onsite gas solutions like GEV (GE Vernova)' as beneficiaries of the market shift, likely due to demand for backup generation and peaker plants in PJM.
The article identifies 'equipment vendors of onsite gas solutions like GEV (GE Vernova)' as beneficiaries of the market shift, likely due to demand for backup generation and peaker plants in PJM. Risk: Execution risk on gas turbine orders; competition from battery storage and renewable alternatives.
GEV
Aishwarya Mahesh Substack author, SemiAnalysis
Bloom Energy is named alongside GEV and CAT as an equipment vendor for onsite gas solutions, suggesting its fuel cell technology may see increased adoption for datacenter backup and peak capacity.
Bloom Energy is named alongside GEV and CAT as an equipment vendor for onsite gas solutions, suggesting its fuel cell technology may see increased adoption for datacenter backup and peak capacity. Risk: High reliance on government subsidies; fuel cell economics still challenged by cheap natural gas.
BE
Aishwarya Mahesh Substack author, SemiAnalysis
Same rationale as VST – Constellation Energy is named in the same list of IPP winners that benefit from the PJM capacity market design and datacenter load growth.
Same rationale as VST – Constellation Energy is named in the same list of IPP winners that benefit from the PJM capacity market design and datacenter load growth. Risk: Exposure to PJM regulatory risk; potential nuclear plant retirements or operational issues.
CEG

This newsletter, published March 03, 2026, features Aishwarya Mahesh discussing VST, VRT, TLN, GEV, BE, CEG. 6 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Aishwarya Mahesh  · Tickers: VST, VRT, TLN, GEV, BE, CEG