# AI’s Duplicate Demand Problem Drives Grids to Commitment-First Planning — retained record

> Retained news record from news:datacenterknowledge.

Canonical: https://highsignal.app/data/records/5c3a0088abed1737

[← sources](https://highsignal.app/data)


source: [news](https://highsignal.app/data/news) (news:datacenterknowledge)  published 2026-06-29 09:00 ingested 2026-06-30 09:15

[original source ↗](https://www.datacenterknowledge.com/energy-power-supply/ai-s-duplicate-demand-problem-is-rewriting-grid-planning)

## retained text



```
AI’s Duplicate Demand Problem Drives Grids to Commitment-First Planning
Recent FERC filings show AI developers and grid operators converging on stricter readiness rules to separate real power demand from speculative projects.
AI and other high-density data center developers have been submitting large-load requests to multiple utilities while they finalize site selection – a practice that has inflated regional demand forecasts and skewed grid investment. Regulators and grid operators are now shifting toward “commitment-first” planning that prioritizes commercially executable projects, shielding transmission spend and, in capacity regions, market outcomes.
In a pending Federal Energy Regulatory Commission (FERC) rulemaking on large loads, in parallel with efforts at the Electric Reliability Council of Texas (ERCOT), PJM Interconnection (PJM), and the Southwest Power Pool (SPP), an emerging principle is taking hold: commitments – not every megawatt asked for – should shape long-term forecasts.
The Forecasting Distortion
Large-load forecasting has become more uncertain as utilities receive unprecedented numbers of data center requests, many of which may be delayed, canceled or ultimately require less electricity than initially proposed. The resulting uncertainty has prompted utilities, regional grid operators and industry groups to adopt new forecasting methods that weigh project maturity and financial commitments before treating proposed loads as likely to materialize.
Google argues that large-load interconnection requests remain “a blind spot for regional transmission planners” because transmission providers have detailed information about generation but far less certainty about which proposed large loads will ultimately energize.
Elizabeth K. Whitney, managing principal at Meguire Whitney, told Data Center Knowledge that the duplication distorted planning. “When each utility started planning for that load, it created duplicative projections on the overall system,” she said. “So the regional grid’s picture of the projected demand was way off.”
For utilities and grid operators, the stakes extend well beyond forecasting. Demand projections drive siting of transmission lines, substation expansion, and generation procurement, and in organized markets, they also influence capacity prices.
Commitment tests seek to align scarce engineering and transmission resources with projects likely to energize. (Image: Getty Images)
From Requests to Commitments
Google’s filing offers a clear expression of the emerging philosophy. The company proposes a Capacity Commitment Framework that would require meaningful commercial commitments before projects influence long-term transmission planning. Google points to instruments such as long-term service agreements, minimum demand charges, upfront collateral, and withdrawal penalties. Google also urges coordinated reviews of large loads alongside nearby generation, arguing that integrated studies can reduce the need for network upgrades, improve reliability, and shorten energization timelines.
Amazon, Microsoft, and OpenAI support variations on the same approach, including standardized readiness requirements, integrated generation-load studies, and expedited treatment for commercially mature projects. OpenAI favors expedited treatment for certain commercially ready projects designated by the US Department of Energy as nationally significant, while Amazon advocates that transmission providers should use artificial intelligence and advanced automation to accelerate engineering studies.
“You’ll continue to see these [commitment and maturity] requirements pop up that start to associate demand more with skin in the game than just announced plans,” Whitney said.
Not everyone agrees on the mechanism. NRG Energy proposes replacing traditional queue priority with competitive “open seasons” for transmission capacity – time-bound solicitations that would allocate capacity competitively – while requiring large loads to bear the costs of the network upgrades they trigger.
Those ideas closely resemble reforms already underway. ERCOT's “Batch Zero” process requires developers to demonstrate project maturity before entering regional transmission studies. SPP’s HILLGA framework – its approach to integrating high-impact, large-load requests – and PJM’s large-load proceedings are developing new pathways for evaluating electricity users. According to Whitney, those proceedings address the same underlying challenge but extend beyond verifying demand to include cost allocation, load flexibility, co-located generation, and the level of transmission service required.
"You’ll continue to see these [commitment and maturity] requirements pop up that start to associate demand more with skin in the game than just announced plans." – Elizabeth K. Whitney, managing principal at Meguire Whitney
The Queue Isn’t the Problem
Industry filings reflect a broader consensus: interconnection queues are symptoms, not the underlying constraint. The Data Center Coalition argues that today’s delays stem as much from inadequate transmission infrastructure and long-term planning as from interconnection procedures. The coalition calls for expanded transmission planning, wider use of grid-enhancing technologies, customer-funded network upgrades where appropriate, and broader deployment of automation and AI to accelerate engineering studies while preserving multiple pathways to serve large loads.
The WATT Coalition approaches the issue from another angle, arguing that grid-enhancing technologies – such as dynamic line ratings, advanced conductors, and advanced power-flow controls – could unlock more than 100 GW of additional transmission capacity before new transmission lines are built.
Consensus – And Friction
The industry is far from unified on details.
Meta urges FERC to issue flexible national guidance rather than prescriptive rules. Equinix warns against designing policy solely around hyperscale AI campuses. Oracle focuses on streamlining FERC’s regulatory process, while Tesla asserts that batteries and virtual power plants should play a larger role in supporting large-load growth.
State regulators are also drawing jurisdictional boundaries. For example, the North Carolina Utilities Commission asked FERC to preserve the Federal Power Act’s “bright jurisdictional line” between interstate transmission (federal) and retail electric service (state), noting that North Carolina already tracks large-load growth through integrated resource planning, semiannual reporting, and utility oversight. That tension is already spilling into related FERC proceedings: Maryland regulators have challenged PJM’s transmission cost-allocation rules, arguing that customers should not bear billions of dollars in transmission costs associated with AI-driven load growth elsewhere on the regional grid.
The Unfinished Business: Building Generation
A major challenge remains largely outside the current federal transmission debate. “Nothing that’s happening in the federal transmission space is really addressing the problem of how to build generation,” Whitney said.
Demand verification may improve transmission planning and reduce the risk of overbuilding around speculative requests. It does not answer the next question: Who will build the generation needed to power the next wave of AI infrastructure, and who will pay for it?
```



document fetched 2026-06-30 09:15 · canonical https://datacenterknowledge.com/energy-power-supply/ai-s-duplicate-demand-problem-is-rewriting-grid-planning