Pillar · ~3.4k words · Updated 2026-05-11
Louisiana's AI power grid problem — and how Entergy is solving it
Hyperscale AI is forcing Louisiana to build new firm generation, expand transmission, and resolve the cost-allocation fight at LPSC. Here's how the pieces fit, who pays, and what could still go wrong.
The load shock
In 2023, the largest single industrial customer on the Entergy Louisiana system drew on the order of 200 megawatts. By 2027, the system is forecast to serve at least one customer (Meta's Hyperion campus) at 2 gigawatts — ten times larger than any prior industrial load. Aggregating across all four announced AI campuses, the net new load forecast on the Entergy Louisiana system over 2025–2030 is roughly 4 GW, against a 2023 peak of about 23 GW.
A 17% peak-load increase over five years, concentrated geographically in two corridors (Richland Parish in the northeast and Caddo-Bossier in the northwest), is the kind of structural shift that historically takes a decade to plan and 15 years to execute. Entergy is trying to execute it in three.
The generation plan
Entergy Louisiana's 2024–2025 LPSC filings outline a three-plant natural-gas combined-cycle build-out totaling roughly 2,500 MW of new nameplate capacity, plus extensions of legacy units that had been scheduled for retirement.
Combined-cycle gas is the chosen technology because it is firm (24/7 dispatchable), capital-efficient compared to nuclear at this scale, and consistent with Entergy's existing gas-fired fleet. The trade-off is carbon: each new CCGT plant emits roughly 350–400 g CO2 per kWh, which is roughly half of a coal plant but well above renewables. Environmental advocates have pushed back on the IRP at LPSC, arguing for more aggressive renewables and storage build-out.
The transmission stack
Generation without transmission is wasted capacity. Entergy's plan includes transmission upgrades from the new generation locations to the load corridors: 500 kV upgrades in the north and additional 230 kV capacity along the Caddo-Bossier interconnect. Transmission build is permit-intensive, eminent-domain-adjacent, and historically the single biggest source of timeline slips in utility expansions.
The cost-allocation fight
Here is the political center of mass: who pays for the new generation and transmission? Two extreme positions and several middle paths:
- Pure 'customer-pays' — the data-center customer (Meta, Amazon) pays the full incremental cost of generation and transmission built to serve their load, through a special tariff. Consumer advocates favor this; data-center operators argue it makes Louisiana uncompetitive versus other states.
- Pure 'socialized' — costs roll into the system rate base and are recovered from all customers, residential through industrial. The data-center operators favor this; consumer advocates argue it forces existing ratepayers to subsidize new mega-customers.
- Hybrid — the data-center customer covers the incremental cost during initial term, with stranded-cost provisions that fall to the rate base only if the customer leaves earlier than planned. This is the most likely actual outcome.
What could go wrong
Three plausible failure modes:
- Generation slips — CCGT plants are 24–36 month builds with supply-chain dependencies on turbines (GE Vernova, Mitsubishi, Siemens). A 6-month slip on the turbine delivery cascades to the campus online-by date.
- Transmission opposition — Louisiana has historically permitted transmission with relatively low opposition compared to coastal states, but 500 kV lines through populated parishes can attract legal challenges.
- LPSC reversal on cost allocation — if the LPSC eventually orders pure customer-pays allocation, individual data-center economics worsen materially, and announced timelines might slip while operators renegotiate.
What to watch
Three milestones over the next 12 months:
- LPSC final orders on the Entergy CPCN filings — particularly the cost-allocation language.
- Turbine procurement disclosures from Entergy and from the GE Vernova / Mitsubishi / Siemens 10-Q earnings.
- Parish-level permit issuance for the new CCGT sites.
Frequently asked
How much new power capacity will Louisiana build for AI data centers?
Approximately 2,500 MW of new natural-gas combined-cycle generation is in Entergy Louisiana's filed plan with the Louisiana Public Service Commission, along with transmission upgrades. The total announced AI campus load across the state is approximately 4 GW; the gap is covered by retirement deferrals and capacity from neighboring Entergy utility subsidiaries.
Why not solar and wind for these data centers?
Three reasons: (1) Louisiana's solar resource is moderate rather than excellent, (2) AI training workloads require 24/7 firm power and operators are unwilling to sign PPAs that depend on intermittent supply without firmness contracts behind them, and (3) the transmission and permitting timelines for utility-scale renewables in Louisiana are slower than for natural-gas combined-cycle expansion at existing sites.
Will my Louisiana electric bill go up because of these data centers?
That is the central unresolved question at LPSC. The most likely outcome is a hybrid cost allocation where the data-center customer covers the incremental cost of generation built specifically for them, with stranded-cost protection for the system if the customer departs early. Under that outcome, residential rates do not rise materially. Under a pure 'socialized' allocation, residential rates would rise; under a pure 'customer-pays' allocation, they would not.
Primary sources
- Entergy Louisiana CPCN filings at LPSC — LPSC
- Entergy Louisiana Integrated Resource Plan — Entergy Louisiana
- EIA Louisiana power profile — U.S. Energy Information Administration
Cite this
LouisianAI. “Louisiana's AI power grid problem (and how Entergy is solving it).” Retrieved from https://louisianai.com/pillars/louisiana-ai-power-grid-problem.