Hyperscalers and energy developers are pivoting to private "shadow grids" and co-located power-and-compute infrastructure to bypass severe…

Read-only snapshot of The AI Power Bill

Jun 22, 2026 · 3 findings · ran 4m 49s

TL;DR

Hyperscalers and energy developers are pivoting to private "shadow grids" and co-located power-and-compute infrastructure to bypass severe utility interconnection queues. Simultaneously, state regulatory commissions are intensifying their scrutiny of utility deals, attempting to isolate residential ratepayers from the long-term risk of stranded fossil-fuel assets built for tech companies. This dual movement is carving out a highly fragmented, regionalized map of who profits from AI's power demand and who is left holding the bill.

Private Shadow Grids and Co-location

To bypass grid interconnection queues, hyperscalers and energy developers are increasingly constructing private power networks that directly integrate merchant generation with computing facilities.

"Power generation and grid interconnections are critical gating factors for AI infrastructure deployments. Helix brings together data center development, infrastructure and power capabilities under a single umbrella, providing a one-stop shop for large load customers." — KKR-Vistra Helix Launch via Datacenter Dynamics

This model is mirrored in Southeast Louisiana, where Hut 8 is building a massive AI campus co-located directly adjacent to a nuclear generating station to power Anthropic's models under a lease backed by Google Louisiana AI Power Boom +2 via Opportunity Louisiana. By shifting from public utility contracts to direct private power alliances, hyperscalers can secure massive energy capacity without waiting in public queues. This trend is further fueled by private funding, such as a $35 billion capital solution led by Apollo and Blackstone to finance AI compute capacity KKR-Vistra Helix Launch .

What to watch: Watch whether Hut 8 successfully begins operating its 1,000 MW nuclear co-located River Bend AI campus in Southeast Louisiana by its target date Louisiana AI Power Boom +2.

The Battle Over Stranded Assets and the 15-Year Trap

State regulators are forcing a reckoning over who bears the long-term cost of specialized grid buildouts when hyperscalers sign power contracts that expire decades before the useful life of the new infrastructure.

"PGE proposed a 29% rate increase specifically for large-load data centers... while simultaneously providing a 1.3% rate decrease for residential and small-business customers." — Ratepayer Protections Backlash via Utility Dive

In Louisiana, Entergy is seeking to build multiple new gas-fired power plants to serve Meta's Hyperion campus Louisiana AI Power Boom +2. While Meta is directly financing the construction, consumer advocates warn that the company's contract lasts only 15 years, leaving captive ratepayers exposed to the cost of these long-lived assets if the tech giant exits early Ratepayer Protections Backlash . This mismatch between short-term tech commitments and the multi-decade lifespan of heavy power infrastructure is the primary catalyst for the regulatory backlash. This extends the regulatory firewall trend seen previously with Oregon's large-load tariff proposal, where Portland General Electric attempted to shift costs away from residential ratepayers Ratepayer Protections Backlash .

What to watch: Watch for the Louisiana Public Service Commission's decision on Entergy Louisiana's fast-tracked application to build new gas plants for Meta Ratepayer Protections Backlash .

What surprised us

The scale of Meta's self-funded gas grid in Louisiana: Meta is directly financing multiple gas plants for $11 billion to power its Hyperion campus Louisiana AI Power Boom +2 via Bloomberg. It is remarkable that a consumer tech company is essentially funding a private utility generation fleet equivalent to a major portion of Louisiana's entire existing grid capacity to power a single campus.
Former AWS CEO Adam Selipsky's rapid jump to the merchant power side: Shortly after leading AWS, Selipsky is heading Helix Digital Infrastructure, a $10 billion joint venture with Vistra and NVIDIA KKR-Vistra Helix Launch via Reuters. This move underscores that the primary bottleneck for cloud growth is no longer software, sales, or silicon, but securing raw, physical megawatts.
Google backstopping Anthropic's power bill: In Southeast Louisiana, Google is providing the financial backstop for Fluidstack's lease of the River Bend AI campus, where Anthropic's models will be deployed Louisiana AI Power Boom +2 via Opportunity Louisiana. This highlights a complex web of hyperscaler alliances where tech giants are actively underwriting competitors' physical power footprints to secure scarce nuclear capacity.

Open threads worth a vote

Louisiana PSC Decision on Entergy's Seven New Gas Plants for Meta

Findings from this cycle

Current topic brief

Shown for context; the brief may have changed since this cycle ran.

Track how the AI data-center buildout flows through to electricity markets, utilities, and ordinary ratepayers — and who profits or pays. That it's happening is well covered; the gap is the regional, investable exposure map: which utilities, grids, and regions absorb the load and where consumer bills rise as a result. Core entities: regulated and merchant utilities signing data-center load (Constellation, Vistra, Talen, NRG, Southern, Dominion, AEP); grid operators and interconnection queues (PJM, ERCOT, MISO); independent power and nuclear/gas supply; and the hyperscaler buyers (Microsoft, Amazon, Google, Meta) signing PPAs. I want to track new load commitments and PPAs, rate-case filings and commentary about cost allocation between data centers and residential ratepayers, interconnection and capacity-auction outcomes, and EIA/FRED data on electricity prices and demand by region. Follow utility earnings calls for load-growth guidance and the capex to serve it. Flag where a region's residential rates are rising to fund AI load, and any divergence between utility load forecasts and what's actually contracted. The thesis: AI's power demand is quietly repricing electricity by region — surface where the cost lands and who captures the upside.