This was initially published for the Center for Strategic and International Studies.
Texas has long stood apart from national energy policy—its own grid, its own rules. In 2026, that independence has made it the top destination for AI-driven electricity demand. While Washington grapples with the pressures of building data centers, ensuring reliability, and controlling costs, Texas is tackling everything at once. The Electric Reliability Council of Texas (ERCOT) market is solving for speed-to-power by surmounting transmission bottlenecks, phantom load growth, buying down residential rate increases, and more.
Last week, a single hearing of the Texas Senate Committee on Business and Commerce brought this into view. Public Utility Commission of Texas (PUCT) Chairman Thomas Gleeson and ERCOT CEO Pablo Vegas told the committee that they were managing 410 gigawatts (GW) of load applications, nearly five times the capacity of today’s ERCOT grid, and nearly 90 percent of them were data centers. In the face of potentially overwhelming demand, Texas will either become an example of success or failure in the United States’ approach to winning the AI race.
The Texas legislature and regulators have developed a suite of policies to manage this growth. Three stand out as particularly important:
- Market-Based Reliability: The move toward a Dispatchable Reliability Reserve Service (DRRS) (mandated in H.B. 1500, codified in PURA § 39.159(d),(e)), proves that even in liberalized markets, long-duration reliability (like gas and long-duration batteries) requires a specific price signal to survive alongside zero-marginal-cost renewables and short-duration batteries.
- The Denominator Effect: Texas is shifting the conversation from “How much will this cost ratepayers?” to “How much can we grow the load base to dilute system costs?” There is an emerging Texan answer to debate over who bears the fixed costs of grid upgrades, and how the grid is utilized (and utilized by what technologies) to spread out fixed costs.
- The End of the “Doom Loop” and Flexible Connection: By ditching the traditional load study process for a “batching” approach, Texas is in a first-of-kind exercise to solve load interconnection with strict project maturity criteria, bankable energization timelines, and as-available service options for loads to receive power faster.
Gas for Tail Risk and Solar for Affordability—Core Texas Themes Resonate Nationwide
Vegas reported that the scale of projected demand has fundamentally shifted Texas’s generation priorities. On April 1, the Texas grid was running entirely on wind and solar with 17 GW of operating reserves. Vegas noted that those operating reserves sit idle and unpaid until called. When asked how the state could justify adding more gas if investors only see a return on the handful of “risky days,” he noted that a new structural signal is required to price the attribute of availability for tail risk days.
>>>READ: How Renewables and Batteries Saved the Texas Grid in 2025
For the first time in years, the natural gas interconnection queue (currently 60 GW) has overtaken wind. However, a massive financing gap remains. While the legislature’s Texas Energy Fund (TEF) catalyzed an initial 9 GW through low-cost debt, roughly 5 GW of proposed gas capacity lacks a viable pathway in markets. Vegas was explicit that state-backed loans are a jump-start, not a permanent solution. He pointed to the DRRS proceedings, which are set to create a dedicated revenue stream to bring longer-duration reserves to the grid. There are now two dueling DRRS implementation proposals pending at ERCOT. NPRR 1309 excludes batteries. NPRR 1310 introduces a “release factor” mechanism that allows a battery to be paid for DRRS while also being available for energy or other ancillary services.
Who Pays for Transmission?
To protect residential ratepayers from grid costs incurred to meet speculative data center loads, the PUCT is developing a standardized large load interconnection process. The proposed rules aim to establish a rigorous financial gauntlet to ensure that only viable, “real” projects receive grid connection allocations. To achieve that, the commission is currently weighing how developers must prove land ownership, financial security, and procurement of long lead-time equipment before they can advance in the queue. As currently drafted in the proposal:
- Proposed Front-Loaded Entry Fees: Developers could be required to pay a $50,000-per-megawatt (MW) deposit when they sign their first paperwork (an “intermediate agreement”) to lock in their place in line.
- Nonrefundable Interconnection Fees: Once studies are finalized and move to a final interconnection agreement, an additional (currently proposed at) $50,000 per MW nonrefundable fee.
- The 80 Percent “Claw Back”: If a project is downsized or withdrawn, roughly 80 percent of the posted security (as currently proposed) could be forfeited and applied directly to the transmission provider’s rate base.
PUCT Chairman Gleeson emphasized that this mechanism intends to use forfeited funds from failed data center developments to directly “buy down” transmission charges for residential customers.
However this proceeding turns out, the national debate on AI and affordability might show that in the case of Texas, armed with better permitting regimes, abundant land, megawatt-hours powered by the sun, and the best free-market construct for batteries in the country, load growth in advance of new fixed costs can hedge rate hikes.
A National Blueprint for “First-Ready” Load Growth
The 410 GW of proposed interconnection in Texas reflects both the desirability of building there and the failure of the one-off bespoke methodology that dominates interconnection processes. A grid operator assesses each project as it arrives for firm power rights, but because dozens of other projects are advancing in parallel, the topology of the grid shifts constantly. This leads to the ultimate surprise for developers: a pre-energization study that announces massive, unforeseen system changes, effectively sending a project that has already invested billions back to the drawing board for a total restudy.
Transition to a proposed batching process for loads in Texas is a counterpoint to the failure of “cluster” studies for generators in other markets. The ERCOT process is being designed to (1) take a fixed snapshot of projects that meet strict maturity criteria, (2) study them as a single group, (3) identify required transmission upgrades, and (4) allocate firm power pro rata across the projects of the batch. ERCOT intends to approve batching rules for the first such group, “batch zero,” at the June 1, 2026, ERCOT board meeting and seek PUCT ratification in July 2026.
Even with these improvements, the latest from ERCOT indicates that batch zero projects will not see their first drop of “firm” power until 2028 at the earliest; many will not reach full requested capacity until 2033. A developer who makes it into a favorable batch might still only be offered a small fraction of the electricity they requested. This creates a high-stakes “misfit” risk between grid and commercial timelines: If that partial offer is less than the bare minimum needed to run at least one building on grid power, the project might not be able to begin operations—effectively leaving the investment sitting idle. This problem epitomizes the national conversation on flexible load connections (and is also addressed in a major rulemaking from the Federal Energy Regulatory Commission).
>>>READ: What CERAWeek 2026 Says About Energy’s Next Chapter
ERCOT is already a leader in connecting generators faster than other regional grids, because power plants accept curtailed use of the transmission system in congested hours and free use in the hours where grid headroom is available. Now, ERCOT is working on similar stakeholder-proposed solutions so that large loads in the batching process can connect earlier with flexible consumption tied to grid headroom. Doing so could allow the batch process itself to become more effective: More loads could “flex” between on-site sources, off-site workload shifting, and available grid headroom, derisking scenarios where they receive less firm power or delayed firm power from the grid.
A snapshot of Texas today captures the full debate over how to energize AI loads. Texas is unique, but policymakers around the country face the same questions: who pays for grid upgrades, who bears the cost if data centers do not materialize, how to energize new loads without losing reliability, and how to reward dispatchability effectively. Texas is confronting all of this at once, since it is a locus of growth and data center development. Major ERCOT and state decisions in the next 12 months will force the issue.
The views and opinions expressed are those of the author’s and do not necessarily reflect the official policy or position of C3.