Texas Market

Market Deep Dive

The Largest — and Most Speculative — Data Center Pipeline in North America

Texas, served primarily by the ERCOT grid, has become the epicenter of U.S. data center development — and of the debate over how much of the pipeline is real. ERCOT's April 2026 Large Load update reported the operator was tracking roughly 410 GW of large loads seeking interconnection, of which ~87% are data centers. Subsequent figures put the pending large-load queue near 474 GW (~90% data centers) — more than five times ERCOT's all-time record peak demand of 91.3 GW.

The vast majority of that queue is speculative. BloombergNEF's tracked Texas project pipeline totals about 50 GW, of which 36 GW is early-stage, 9 GW committed, and only 5 GW physically under construction. BNEF forecasts roughly 8.25 GW of additional ERCOT data center capacity coming online through 2030, bringing the installed total to about 17.2 GW — a small fraction of the headline queue number.

Sources: ERCOT Large Load Update (Apr 2026); BloombergNEF (Aug 2026).

Project Pipeline: From Stargate to Speculative Filings

Beneath the headline queue, a real build-out is underway. The flagship is Stargate's Abilene campus — the OpenAI/Oracle joint project announced in January 2025 as part of a $100B initiative, with the first Abilene phase above 1.2 GW of planned capacity Confirmed. Lancium's neighboring Abilene campus adds another gigawatt-scale commitment to the same West Texas corridor Probable. In the metros, Dallas–Fort Worth remains the anchor market: QTS, Digital Realty, and Compass operate large hyperscale campuses, Google continues to expand at Midlothian, and Microsoft around San Antonio Confirmed.

Our own tracker carries the Texas AI Corridor project — an 800 MW AI campus program in the Fort Worth area — at Speculative stage pending interconnection progress. That label matters: in Texas more than anywhere, announced capacity and energized capacity are different numbers, and the state's own audit process is now the sorting mechanism.

Sources: Company announcements; White House Stargate event (Jan 2025); CBRE DFW market reports; DataPowerDemand project tracker (Aug 2026).

Batch Zero and the Push to Rationalize the Queue

To separate real projects from speculative filings, ERCOT approved a new Batch Zero large-load interconnection process in June 2026, assessing projects in batches rather than individually. BNEF estimates 204 GW of load was eligible for Batch Zero study, with another ~294 GW applying but not eligible. The process is designed to allocate scarce transmission and capacity more efficiently and could become a national model.

Under Texas Senate Bill 6, large-load developers must already provide site-control documentation, $50,000 per MW in financial security, disclosure of duplicate interconnection requests, and on-site backup-generation reporting. The PUCT is expected to finalize the implementing rules (16 TAC § 25.194) around September 2026.

Sources: ERCOT Planning Guide; Texas SB 6; BNEF.

ERCOT's Market Structure and the Large-Load Problem

ERCOT is the only major U.S. grid that is both energy-only and largely outside FERC's interstate jurisdiction, and that combination made it the fastest place in America to connect a data center — historically months, not years, from application to energization. The trade-off is that ERCOT never had a cost-screen for speculative requests: filings were cheap, duplicate requests across multiple sites were common, and by 2025–26 the queue had swollen past 400 GW of nominal large load against a system whose all-time peak is 91.3 GW.

The 2025 legislature responded with Senate Bill 6 (signed June 2025), which for the first time gave ERCOT and the PUCT statutory tools to manage large loads: mandatory disclosure of duplicate requests, $50,000 per MW in financial security, on-site backup-generation reporting, and — most consequentially — provisions allowing curtailment of loads 75 MW and larger during firm-load-shed events. The PUCT's implementing rulemaking (16 TAC § 25.194) is due to finalize around September 2026 Estimated.

The curtailment provisions deserve emphasis because they invert the industry's reliability bargain. Historically, data centers paid for firm, uninterruptible service; SB 6 instead contemplates large loads as a demand-side resource that ERCOT can call down during scarcity, in exchange for speed of interconnection. Sophisticated developers are already designing for it — on-site backup sized not just for outages but for curtailment events, which makes generator and storage procurement a grid-compliance decision as much as a resilience one Estimated. See our queue reform explainer for how Texas compares with PJM and the Southeast.

• Batch Zero: ERCOT's June 2026 move from serial studies to batch assessment — ~204 GW eligible, ~294 GW applying but ineligible (BNEF).
• SB 6: financial security, duplicate-request disclosure, backup-generation reporting, 75 MW+ curtailment provisions.
• August 2026 audit pause: all new connections frozen pending a project-by-project queue audit.
• Next catalyst: PUCT final rules (~Sep 2026) and the January 2027 legislative session.

Sources: Texas SB 6 (2025); ERCOT Planning Guide & Market Notice M-A080326-01; PUCT rulemaking docket; BNEF (Aug 2026).

Generation Response: Gas, Solar, Storage — and a Nuclear Option

Texas is attacking the supply side from every angle at once. The Texas Energy Fund — the $5B low-interest loan program voters approved as Proposition 7 in November 2023 — is subsidizing new dispatchable gas capacity, with the first awards flowing to combined-heat-and-power and simple-cycle projects Estimated. Meanwhile the market keeps doing what it does without subsidy: ERCOT leads the nation in utility-scale solar additions and battery storage deployment, and batteries increasingly set the evening peak price.

The strategic wildcard is firm power. Constellation's Calpine acquisition created the country's largest gas fleet with a Texas-heavy footprint, and SMR developers court West Texas campuses for behind-the-meter nuclear Possible. For hyperscalers, the practical 2026–28 answer is hybrid: grid interconnection where ERCOT allows it, plus on-site gas or storage to bridge the gap — the same playbook we detail in our backup power analysis.

The demand math explains the urgency on the supply side. ERCOT's own long-term scenarios sketch data centers as the single largest load-growth driver through 2030, and every gigawatt of campus load that energizes is a gigawatt that must clear the summer evening peak — after solar ramps down. That is why the Texas Energy Fund's dispatchable focus matters, and why ERCOT's reserve-margin math is now a data center story as much as a weather story. Winter Storm Uri (February 2021) remains the systemic reference point: the market will pay for firmness, and regulators will not again be caught short on firm capacity Estimated.

Sources: Texas Energy Fund / PUCT; ERCOT capacity, demand & energy reports; Constellation–Calpine transaction filings; company disclosures.

Transmission: CREZ's Legacy and the 765 kV Question

Texas has done a grid-scale build-out before: the ~$7B Competitive Renewable Energy Zones (CREZ) program strung 3,600 miles of 345 kV line to move West Texas wind to the metros, finishing in 2013. That network is why West Texas solar and storage interconnect so quickly today — and why data center developers followed the power to Abilene. The new question is whether ERCOT needs a 765 kV overlay: AEP Texas and other transmission owners have floated extra-high-voltage proposals, and the Permian Basin reliability plan is already driving multi-billion-dollar 345 kV additions Estimated.

The practical constraint for data centers is less the bulk lines than the local stations: 345/138 kV substations, breakers, and transformers with 18–36-month lead times Estimated. ERCOT's batch process is, in effect, a rationing system for that equipment as much as for study capacity.

The capital numbers are already large. ERCOT's transmission owners have queued billions of dollars in 345 kV projects across the Permian, the Panhandle, and the I-35 corridor, and the PUCT has signaled openness to a 765 kV backbone if load commitments firm up after the audit Estimated. A 765 kV decision would be a decade-defining signal: it is the voltage class AEP built its eastern backbone on, and it would mark Texas's first adoption at scale. For data center siting, the practical takeaway is shorter-term — sites adjacent to CREZ-era 345 kV stations with existing breaker positions are trading at a premium because they shave 12–24 months off energization Estimated.

Sources: ERCOT transmission planning; PUCT CREZ history; AEP Texas filings; BNEF.

Water and Siting: The Quiet Constraint

Texas campuses increasingly specify dry or hybrid cooling, and West Texas projects lean on air-cooled designs almost by necessity — summer wet-bulb economics and aquifer limits make evaporative cooling at gigawatt scale a hard sell. Our water analysis shows why: cooling choice now drives both permitting risk and operating cost in hot, dry markets.

Land, by contrast, is Texas's structural advantage. Multi-thousand-acre assembled sites with fiber and high-voltage proximity exist around Abilene, the Panhandle, and the DFW exurbs at prices coastal markets cannot match. The new friction is community: water-use disclosure fights and noise complaints around West Texas campuses are now routine, and SB 6-era politics have made large loads a visible target in Austin.

There is also a circular opportunity unique to the oil patch: several West Texas developers are evaluating power-and-cooling designs co-located with produced-water treatment and oilfield gas, turning flare gas and brackish water into behind-the-meter supply Possible. None of these concepts is yet at gigawatt scale, but they illustrate where Texas development goes if the ERCOT queue stays slow: off-grid first, grid-connected later.

Sources: Project permitting filings; Texas Tribune; DataPowerDemand analysis.

August 2026: The Abbott Audit Pause

On August 3, 2026, Governor Greg Abbott ordered the PUCT and ERCOT to pause all new data center grid connections pending a comprehensive audit of every project in the queue. ERCOT responded by suspending the first deliverable of Batch Zero. BNEF estimates the pause could delay up to 49.8 GW of data center load — nearly 20% of the entire U.S. development pipeline — and cost projects $8–15 billion depending on the AI-compute mix, if additions slated for 3Q 2026–1Q 2027 slip to 2Q 2027.

The directive is framed as a reliability and compliance action, not a moratorium, but BNEF notes it is politically charged ahead of the November 2026 election, and warns that a prolonged audit could collide with the January 2027 legislative session and invite further statutory restrictions. Community opposition and water use are becoming material constraints even in a state that has actively courted the industry.

Sources: Gov. Abbott directive (Aug 3, 2026); ERCOT Market Notice M-A080326-01; BNEF (Aug 5, 2026).

Competitive Positioning and the 12–24 Month Outlook

Texas's offer is unique: fast land, cheap power, and no capacity charge — but, as of August 2026, a queue frozen by political audit. Versus Northern Virginia it offers power availability without PJM's capacity-auction exposure; versus Phoenix it offers scale without Colorado River water politics; versus Atlanta it offers a deregulated market but none of Georgia Power's all-in-one utility execution. The audit pause is the swing factor: a credible, fast audit that clears speculative filings would accelerate real projects by freeing transmission studies; a prolonged one pushes development to behind-the-meter West Texas and out-of-state peers.

Watch items: (1) duration and findings of the Abbott audit and the resumption of Batch Zero deliverables; (2) PUCT's final SB 6 rules and the first curtailment provisions tested in a summer peak; (3) 765 kV overlay decisions; (4) whether ERCOT's 2030 load forecast is revised materially upward once the queue is cleaned. Follow the data on our markets dashboard and project tracker.

The second-order effect to watch is capital rotation. Developers with Texas projects stuck in the audit are already re-weighting toward Georgia, Ohio, and the Carolinas, where regulated utilities offer slower but clearer paths Estimated. Texas can reverse that quickly — the underlying fundamentals (land, gas, renewables, speed) are intact — but the state is now competing on process credibility, a metric it invented the market to avoid.

Sources: DataPowerDemand analysis; ERCOT; PUCT; BNEF (Aug 2026).

Outlook

Texas will remain a top-tier market on land, power availability, and speed-to-market, but the binding constraint has shifted from demand to queue credibility and transmission. Watch three things: the duration and rigor of the Abbott audit; PUCT finalization of SB 6 rules; and whether Batch Zero materially shortens interconnection timelines. A credible clearing of the speculative queue could push ERCOT's 2030 demand forecast significantly higher; a drawn-out political fight could push more developers toward behind-the-meter generation in West Texas and the Panhandle.

Our base case: the audit clears committed projects by early 2027, Batch Zero resumes, and Texas re-emerges with the cleanest — if most regulated — large-load pipeline in the country. The bear case is a politicized audit stretching into the legislative session, in which case expect a visible shift toward self-supplied West Texas campuses and a measurable share shift to the Southeast. Either way, Texas remains the market where the industry's defining question — how much of this pipeline is real? — gets answered first.