From Bottlenecks to Breakthroughs: Rewriting the Grid Planning Playbook in the Southwest Power Pool and Texas

Commentary by Arushi Sharma Frank

Published July 23, 2025

Origin Story: Planning as the Bottleneck

AI data campuses are not "loads" in the traditional sense of large industrial customers, but many regional grids continue to model these systems as they would inflexible steel mills. This results in long wait times for data centers to come online. One key difference is that they often invest in their own electricity generation, backup generators, and/or storage. By taking these paired resources into account, grid planning could be revolutionized with the benefit of bringing data centers online more quickly while simultaneously improving real-time grid operations.

In interconnection studies and grid planning cases, utilities and independent system operators/regional transmission organizations (ISOs/RTOs) should treat colocated baseload generators, backup generators, and long-duration batteries as zero‐peak import resources. That net-zero peak import would accelerate approvals, defer costly transmission upgrades, and spread reinforcement costs across a broader participant base. The principle is simple: If a campus invests in the means to manage its own peak demand, let it connect to the grid faster without stalling the whole project for firm transmission upgrades.

This innovation is easier to discuss than it is to implement. That is why we have more and more announcements, including support from the White House, the benefits of colocation behind-the-fence, advocacy for consumer-regulated power-sharing, and so forth.

Traditional planning practices lock in worst-case assumptions that presume these large loads would burden the grid with their full demand requirements. When developers in Federal Energy Regulatory Commission (FERC) regions do propose to bring their own generator or battery to serve the data center load, they must enroll as a capacity resource-invoking full resource adequacy obligations, lengthy interconnection studies, and firm transmission service requirements under the tariff. That registration immediately pulls the generator (and its associated load) into a network upgrade cost allocation process that socializes new transmission expenses across all customers, instead of directly offsetting the data campus's own peak import. Developers avoid this by simply fencing such assets, keeping them out of capacity constructs but also forfeiting their ability to show a planning benefit (deferred transmission upgrades) or an operations benefit (buffering grid needs in real time).

New grid policy and engineering assumptions are needed together. Some markets are proposing new measures to address these planning limitations and leverage the unique properties of data center loads to expedite interconnection.

The Southwest Power Pool's Breakthrough: Proposals Incentivizing Interruptible Loads

The Southwest Power Pool (SPP) is attempting something transformative to address this planning bottleneck. Its proposed tariff revisions for High-Impact Large Load Generation Interconnection Assessment (HILLGA) create a structured, 90-day system-impact study leading into the standard firm service queue that assesses the large load and its supporting generation in parallel-designed for customers who cannot tolerate any interruption. It also proposes a Conditional High Impact Large Load Service (CHILLS) framework in parallel that offers the same 90-day study but grants up to five years of interruptible (i.e., deferred upgrade) transmission service in exchange for agreed curtailment rights.

A quick terminology review:

• HILL (High-Impact Large Load): A load ≤ 69 kilovolts (kV) and ≥ 10 megawatts (MW), or ≥ 69 kV and ≥ 50 MW.
• CHILL (Conditional HILL): A HILL that cannot be reliably served on a firm basis but seeks non-firm service.
• CHILLS (Conditional High-Impact Large Load Service): A new non-firm point-to-point transmission product for CHILLS.
• HILLGA (HILL Generation Interconnection Assessment): The fast-track study process that pairs colocated generation with a HILL or CHILL.

HILL Generation Interconnections Assessment

The HILLGA path contains what some stakeholders describe as the "low-hanging fruit." By studying colocated or feeder-connected generation alongside the load, SPP's proposal could prevent the common "sequential drag" of waiting for a generator to clear the generator interconnection queue before the load can connect.

• Path 1 (Common Bus): Generation and HILL/CHILL share the same delivery point, studied in 90 days (or via an equivalent transmission owner stability analysis) with zero net injection and separate metering.
• Path 2 (Local Area): Generation may serve one or more HILLs so long as it lies within two substations (and no more than five total) and two transmission lines of each load, with output capped to the agreed HILL capacity and service valid for five years.
• Path 3 (Deliverability Area): Generation anywhere in the same SPP deliverability area (one of 23 deliverability areas) undergoes a 90-day Network Resource Interconnection Service study-requiring ≥ 20 MW/min ramp (≥ 50 percent of the HILL) and accreditation equal to the HILL-to secure permanent non-firm service.

By packaging colocated or feeder-connected generation and load into a single, rapid study process, this proposal prevents the usual interconnection bottleneck. It accelerates speed to power for large load campuses. (See Slides 29-31, Large Load Stakeholder Engagement Forum, July 1, 2025, Southwest Power Pool).

SPP Conditional High Impact Large Load Service

CHILLS gets the operations innovation right. This proposal seeks to balance grid reliability and load-onboarding speed: New loads get on in months, not years, but operators retain the right to shed them first in emergencies. In practice, a CHILLS customer could run their own contracted or colocated generation or storage during a curtailment event to avoid downtime, while preserving the SPP resource adequacy cost-causation principles and operational visibility.

The CHILLS proposal could go further, however. If customers could register a fixed MW of contracted generation or storage that SPP would reduce from the campuses' modeled peak import in its transmission planning cases, this would defer or downsize network upgrades for anything below that net threshold. This opens the door to a modular approach to large load interconnection that reflects how real infrastructure portfolios are designed today.

CHILLS would also introduce the concept of non-firm point-to-point service for high-impact loads, subject to refund when the load is curtailed. CHILLS customers thus would not be billed based on energy (megawatt-hours) consumed, but rather on individual non-coincident peak demand (MW). That means customers pay transmission charges based on the highest level of power they draw at any time in a billing period-regardless of when it occurs or how much total energy (megawatt-hours) they use across the month; a customer that flattens its peak or shapes its profile to avoid high MW draws pays less. Therefore, the proposal notably does not include rewards for customers to curtail during system-wide stress events.

Instead, CHILLS's mandatory "curtail or lose" obligation can be seen in two ways: as an implicit price signal-encouraging colocation and onsite peak shaving investments at large-load campuses-or as a premium, non-firm service option for customers that cannot wait for firm transmission upgrades. If the CHILLS framework is implemented with some flexibility for these campuses to maintain interruptible service beyond the five years, those upgrades could likely come with more utility to more grid users. Waiting for the firm loads to come is a strategy to avoid overbuilding the grid to accommodate peak demand from loads that can serve themselves.

Electric Reliability Council of Texas's Strength: Tools Without a Non-Firm Transmission Product

While SPP has introduced a new framework for non-firm transmission access, the Electric Reliability Council of Texas (ERCOT) has long held a competitive advantage in its market structure. Unlike other markets, ERCOT allows generators and batteries into grid interconnection without imposing resource adequacy requirements. This means that every new generator study does not immediately gum up transmission cost allocation and trigger long deliverability studies to develop firm transmission upgrades for the contributions of the generator or battery.

For generators, ERCOT supports detailed modeling, real-time locational marginal pricing, and granular asset visibility. These tools allow for fast simulation of non-coincident peaks, response behavior, and dispatchable load profiles compared to the often months- or year-long studies in other markets. However, these tools are not part of a grid policy (yet) for studying and operating large load campuses the same way-non-firm treatment is only available to generators and batteries in isolation. When batteries are added to a load campus, they are just treated as additional firm load instead of as a buffer-in-reserve. Further, Generators are stuck in site netting arrangements that do not encourage grid registration and participation to co-serve the load and enable grid support.

What ERCOT lacks in the present is what SPP is on a path to building: a defined pathway to credit colocated or contracted resources for reducing large load-induced grid upgrade costs and thus accelerating speed to power for large load campuses and offering an accelerated pathway for connection in exchange for interruptible service. While some loads can earn real-time incentives to curtail, Texas offers no framework to study load campuses in the planning phase for non-firm transmission access (see Arushi Sharma Frank's comments in PUCT Project No. 58317, SB 6 Implementation Workshop, at timestamp 3:37:11).

Nodal Protocols are ERCOT's rules of the grid, similar to tariffs-critical rule changes must move to implementation to unlock benefits for data center connection (see comments by Michelle Richmond of the Texas Competitive Power Advocates and Arushi Sharma Frank in PUCT Project No. 58317, SB 6 Implementation Workshop, at timestamp 3:12:49). Texas has approved Controllable Load Resources (CLRs) reform that would prevent opt-out of those resources from ERCOT's nodal dispatch instructions. When implemented, it will start to address the innovation challenge. Controllable load campuses could be credited for their ability to be a guaranteed path for grid curtailment (with a generator, a battery, a load that moves, or all three), because the grid operators can study with more certainty that site behavior in a model will be mimicked by how the campus behaves in real-time operations. Flexible load assumptions could be a step closer to getting baked into transmission planning models.

Texas Senate Bill 6 (SB 6) gave the state an opportunity to codify rules around real-time energy scarcity at data center campuses. This is not the only binding issue, however: Transmission constraints today, as identified by transmission and reliability studies, are arresting investment certainty for data centers and creating unnecessary complexity for other large loads in the heavy industrial sector, too. This must be solved by creative approaches that let some campuses energize with smart location choices to fill "valleys" of energy hours where the system has headroom and free up the network for everyone.

Follow-on SB 6 rulemaking efforts present a critical opportunity to integrate load-side benefits into how ERCOT plans the system with the right incentives. The key is not just to acknowledge curtailment capability-but to credit it in transmission studies the same way generation redispatch is already credited today. Regulators can prioritize reforms that:

• Allow registered loads and batteries to be studied together with colocated generators to offset their peak contribution impact in interconnection studies. Similar to SPP's proposed pathways, provide an option to temporarily defer network upgrades.
• Enable batteries to be studied as offsets to firm load in load campus interconnection studies-today, they receive no deferral or planning credit for their ability to shave peaks at these sites because they are modeled only for the worst-case scenario that the batteries and loads will draw max power together from the grid.
• Advance netting reforms so colocated generators are encouraged to voluntarily grid-register and participate in peak contribution management for the local system.

Fitting It All into National Trends

SPP and Texas ERCOT are not alone. Across the country, regulators and utilities are finding new ways to plan for various versions of de-risking the power opportunity for AI, derisking cost recovery risk, and/or addressing phantom load.

• American Electric Power in Ohio recently finalized a demand-side tariff requiring data centers over 25 MW to pay for 8 percent of their expected monthly load to guarantee the utility cost recovery and planning certainty.
• ComEd (Illinois) has filed a proposal to increase its interconnection deposit to help remove speculative projects from the large load queue.
• Dominion Energy has proposed a new rate class for high-energy users, including data centers, that requires a 14-year power purchase commitment. This proposal is designed to ensure cost recovery and rate stability for anticipated future load growth.
• Pacific Gas and Electric Company's FlexConnect program in California is pioneering customer-integrated curtailment strategies to defer transmission upgrades.
• In Virginia, Old Dominion Electric Cooperative is piloting a model that allows one service boundary to include multiple data centers under a shared demand profile-streamlining billing, planning, and flexibility management across tenants and separating the risk profile from these sites from that of the rest of its customers.

Conclusion: Planning as a Competitive Variable

SPP's HILLGA and CHILLS reforms could represent the clearest attempt yet to reengineer grid access around the realities of modern loads and infrastructure grid planning. By contrast, Texas holds the modeling tools and nodal market structure to lead-but has yet to build a regulatory pathway that credits what today's sophisticated load developers bring to the table to smooth out planning cases and grid upgrade timelines.

While ERCOT is largely exempt from federal oversight, SPP's proposed reforms must be approved by the FERC. The FERC's eventual review of the SPP tariff proposal could allow SPP to lay the groundwork now, even with an imperfect model, and call for revised iterations as usage grows.

FERC could also issue guidance encouraging all jurisdictional ISOs/RTOs to embed these strategies in their planning frameworks. This action would set an important precedent to accelerate the national trend: grid planning improvements can de-risk the power opportunity for AI, unlock economic opportunity, and plan a smarter grid for all.

Arushi Sharma Frank is a senior associate (non-resident) with the Energy Security and Climate Change Program at the Center for Strategic and International Studies in Washington, D.C.

Programs & Projects