This is a continuously updated hub for Data Center Knowledge reporting on Texas and the Electricity Reliability Council of Texas (ERCOT), tracking how power-first strategies, regulatory changes, and transmission expansion are shaping AI-scale infrastructure.\u00a0<\/p>\n
Last updated: June 25, 2026.<\/p>\n
Texas has emerged as the testbed for AI-scale data centers \u2013 large campuses built to train and run AI systems that draw round-the-clock, high-magnitude electricity. At the center is ERCOT, which operates most of the state\u2019s grid and wholesale power market.\u00a0<\/p>\n
To keep pace, ERCOT is changing how projects are queued for grid connections. In the past, speculative interconnection queuing<\/a> allowed projects to hold spots with little evidence they would actually build, creating long backlogs. ERCOT is now moving to evidence-based queue management with readiness tests \u2013 such as site control, permits in progress<\/a>, financing plans, equipment orders, and realistic power-use forecasts \u2013 to separate committed loads (projects with a high likelihood of being built and energized) from aspirational requests. The goal is to focus limited grid-planning resources on buildable projects.<\/p>\n ERCOT warns that rapid AI demand will only land on schedule if supply expands in step. Both generation (power plants and storage that supply electricity \u2013 natural gas, wind, solar, batteries) and transmission<\/a> (the high-voltage lines that move power long distances) must grow. These dependencies will influence when sites receive power for the first time, what they pay for it, and whether projects ultimately come online.<\/p>\n In parallel, developers are adopting a power-first model. They are locking in long-term power contracts earlier and, in some cases, pairing data center sites with on-site or nearby generation to shorten time-to-power<\/a> and reduce risk. Others are exploring off-grid and behind-the-meter options<\/a> to gain more control over timelines and cost, though those approaches carry their own regulatory and financing hurdles.\u00a0<\/p>\n Across the stories below, readers will find how Texas policy is evolving, what tactics projects are using to get power sooner, how companies are buying electricity differently, and how Texas stacks up against other markets. The aim is to show how the Lone Star State is turning plans into delivered megawatts and where execution risks remain.<\/p>\n ERCOT Policy, Rules, and Market Signals<\/p>\n Texas Approves Batch Zero Study as Data Center Demand Soars<\/a> (June 2026) \u2013 The Public Utility Commission of Texas (PUCT) approved ERCOT\u2019s one-time \u201cBatch Zero,\u201d a statewide study to evaluate very large new power users, such as AI data centers, in a consistent way. Instead of each utility running its own review, ERCOT will run a centralized process that prioritizes projects with evidence they will actually be built over early-stage or speculative requests.<\/p>\n Gridlock or Growth? ERCOT Warns Texas AI Power Boom May Not Materialize<\/a> (May 2026) \u2013 ERCOT now projects that electricity demand could approach 368 GW by 2032 as AI and data centers scale. Some developers are pushing ahead by adding on-site generation to avoid delays, while ERCOT works to separate committed projects from those that are unlikely to move forward.<\/p>\n Texas Gets Tough on Data Center Power \u2013 Who\u2019s Next?<\/a> (January 2026) \u2013 Texas\u2019s Senate Bill 6, enacted in June 2025, shifts more grid-connection costs and reliability responsibilities onto very large ERCOT customers (roughly 75 MW and up). The policy aims to prevent household ratepayers from subsidizing large data center loads and to ensure these facilities can reduce demand or be curtailed in emergencies. It\u2019s part of a broader national trend toward \u201cpay your own way\u201d as AI growth strains grid planning and capacity.<\/p>\n Comment: Texas\u2019 New Large-Load Law Is an Opportunity for Data Centers<\/a> (June 2025) \u2013 In this opinion piece, CPower Energy Senior Regulatory Affairs Analyst Claire Swingle argues that higher costs and new curtailment rules under SB-6 can be turned to data centers’ advantage. By using demand response (being paid to lower usage during grid stress), adding on-site power, and choosing the right utility tariffs, data centers can manage costs while supporting grid reliability.<\/p>\n Building the Grid for AI: Technology and Transmission<\/p>\n DOE Is Quietly Building the AI Grid Operating System<\/a> (June 2026) \u2013 The US Department of Energy\u2019s \u201cAgora\u201d platform models how hyperscale AI data centers pull power, often in fast-changing bursts, and how they should behave during grid events. It helps validate scenarios like ride-throughs<\/a> (staying online during brief disturbances), reconnection (coming back in a controlled way after an outage), and controllable load (adjusting power use on command). The insights inform ERCOT\u2019s and Texas A&M\u2019s work on interconnection and operating requirements tailored to large electronic loads.\u00a0<\/p>\n Texas May Have Accidentally Built the Perfect Grid for AI<\/a> (May 2026) \u2013 Texas\u2019s 345-kilovolt (kV) CREZ transmission corridors, originally built to move West Texas wind power, are now becoming a critical enabler for AI data center development. However, while ERCOT\u2019s large-load queue remains crowded and key equipment is back-ordered, a substantial gap exists between announced gigawatt-scale projects and actual physical deployment, forcing regulators and utilities to distinguish viable developments from speculative proposals. \u00a0<\/p>\n Texas vs. the Competition: Market Position and Momentum<\/p>\n Texas Powers Past Virginia in Global Data Center Rankings<\/a> (May 2026) \u2013 Texas\u2019s edge \u2013 abundant power<\/a>, ample land, and fewer siting constraints \u2013 is enabling multi-gigawatt data center campuses to move faster than in congested legacy markets. This momentum is challenging Northern Virginia\u2019s long-standing lead.<\/p>\n Major Projects: Who\u2019s Building What and How They\u2019re Getting Power<\/p>\n Behind-the-Meter AI: Microsoft-Chevron\u2019s West Texas Bet<\/a> (June 2026) \u2013 Near Pecos, Project Kilby pairs a Microsoft-run AI campus with roughly 2.67 GW of co-located natural-gas generation developed by Chevron\u2019s Energy Forge One and Engine No. 1. The plan targets first power in 2028 and builds toward about 2 GW of data center capacity over five to seven years. In effect, Microsoft and its partners are bringing their own power to the site to speed up and de-risk buildout<\/a>.\u00a0<\/p>\n Google Bets on a Power-First Model for AI Data Centers<\/a> (June 2026) \u2013 Google and energy developer Intersect are pursuing a Texas campus with approximately 1 GW of dedicated power: mostly wind, solar, and batteries, supplemented by on-site gas to ensure an always-available, lower-carbon supply.<\/p>\n Unconventional Texas Data Center Explores Off-Grid Power<\/a>(May 2026) \u2013 After facing roughly $35 million in interconnection costs and potential delays to 2029, industrial developer BaRupOn is pivoting a 700-acre Liberty County site to an off-grid, behind-the-meter design. Options under review include gas-fired generation and modular microreactors<\/a> to shorten time-to-power and control costs.<\/p>\n Nvidia Places Massive AI Infrastructure Bet on IREN\u2019s 5 GW Pipeline<\/a> (May 2026) \u2013 Nvidia and IREN partnered to make IREN\u2019s 2 GW Sweetwater campus a flagship AI factory<\/a>, with a plan to deploy up to 5 GW of Nvidia-optimized infrastructure across IREN\u2019s global pipeline.<\/p>\n Exowatt Expands to Austin as Power Constraints Reshape AI Infrastructure<\/a> (March 2026) \u2013 Exowatt is establishing an Austin hub to deploy modular renewable power systems that deliver energy directly to data centers behind the meter. The model aims to bypass grid bottlenecks and support high-density AI computing.\u00a0<\/p>\n Crusoe and Microsoft Expand Abilene AI Campus with New 900 MW \u2018AI Factory\u2019<\/a> (March 2026) \u2013 Crusoe is adding a 900 MW AI data center to its Abilene campus for Microsoft, bringing the campus\u2019s total planned capacity to 2.1 GW. The project exemplifies a fundamental shift in hyperscale infrastructure toward energy-first site selection, with on-site power generation becoming essential to bypass grid constraints.<\/p>\n Logix Ramps Up Fiber Networks as Texas Data Centers Boom<\/a> (February 2026) \u2013 Logix Fiber Networks is building high-capacity routes in South Dallas and the Austin\u2013Bastrop corridor, linking new growth zones to Dallas-Fort Worth carrier hubs. More fiber means more bandwidth and path diversity, improving performance and resilience for large AI campuses.<\/p>\n NRC Intervention Tests the Data Center Case for SMRs in Texas<\/a> (February 2026) \u2013 Dow and X\u2011energy\u2019s proposal for four Xe\u2011100 small modular reactors at the Long Mott Generating Station in Calhoun County hit a regulatory speed bump. The Nuclear Regulatory Commission\u2019s Atomic Safety and Licensing Board accepted a challenge to the project\u2019s financial qualifications, highlighting financing and timeline risks that nuclear-backed strategies face.\u00a0<\/p>\n Soluna Expands Texas Campus With 100 MW AI-Ready Data Center<\/a> (January 2026) \u2013 Soluna and Metrobloks are developing Project Kati 2 in Willacy County, an AI\/high-performance computing data center next to Soluna\u2019s wind-powered campus. The plan starts just above 100 MW with a roadmap to exceed 300 MW as the buildout proceeds.<\/p>\n GridFree Unveils First \u2018Power Foundry\u2019 Site for AI Data Center Workloads<\/a> (December 2025) \u2013 GridFree AI launched South Dallas One, the first site in a planned 5 GW cluster of gas-powered, grid-independent campuses. The concept combines dedicated on-site power and industrial-scale cooling to support rapid deployment without drawing on public grid capacity.\u00a0<\/p>\n