Texas is taking an unusual approach to grid planning: build the infrastructure first, then let the data centers follow. In April 2025, the Public Utility Commission of Texas (PUCT) unanimously approved a 765-kilovolt transmission strategy estimated at roughly $13.8 billion, marking the first extra-high-voltage backbone in ERCOT history.
From Oil Fields to AI Infrastructure
The roots of the plan stretch back well before the current AI boom. ERCOT’s 2019 Delaware Basin study and its 2021 Permian Basin planning work were originally focused on electrifying oil-field drilling operations. By 2024, however, transmission providers were forecasting an exponential increase in non-oil-and-gas demand, driven primarily by data centers, cryptocurrency mining, and hydrogen production.
Those non-traditional loads represented a projected 11.6 GW of demand by 2030, nearly matching the oil-and-gas load that originally justified the Permian reliability initiative. Combined demand was forecast to reach 23 GW by 2030 and 26.4 GW by 2038. At that scale, ERCOT concluded that simply adding more 345 kV lines was not the right answer.
Why 765 kV Won the Comparison
ERCOT and PUCT staff evaluated three voltage options before making a recommendation:
- 345 kV expansion: estimated at $12.95 billion
- 765 kV network: estimated at $13.8 billion
- 500 kV alternative: estimated at $15.32 billion
Staff recommended the 765 kV option, describing it as a forward-thinking policy that balances forecast uncertainty, cost, and reliability. Commissioner Kathleen Jackson noted the Permian Basin had “outgrown its infrastructure” and argued the higher initial cost delivers more transfer capability and longer-term benefits. ERCOT’s planning memo also warned that rapid interconnection timelines, as short as 6 to 12 months for large loads, could push actual demand beyond current 2038 forecasts.
Planning for Uncertainty, Not Just Capacity
The decision echoes an earlier Texas precedent. The Competitive Renewable Energy Zones (CREZ) buildout, originally designed to move West Texas wind generation, ended up shaping where hyperscale data center development clustered by creating high-capacity corridors and substations that later attracted major AI projects. The 765 kV plan is a more deliberate version of that same idea: build the wires, and the workloads will locate around them.
Neil Osnato, founder of Persistence Analytics Group, told Data Center Knowledge that traditional grid planning was designed around known generation additions and identifiable industrial customers. AI demand is different because it is larger, faster-moving, and less geographically predictable than most historical forms of economic development.
ERCOT’s planning documents acknowledge this directly, noting that the grid operator cannot know precisely where future load and generation will ultimately site in Texas. The 765 kV network is intended to provide the foundation for a looped, networked statewide plan that can absorb that uncertainty.
What This Means for Hosting and Data Center Operators
For operators evaluating Texas as a site for AI workloads or large-scale hosting infrastructure, the approved plan signals that the state is deliberately engineering long-term grid headroom rather than reacting to demand after the fact. Higher transmission capacity also reduces the risk of reliability constraints that have complicated power procurement for data centers in other markets. The commission’s unanimous vote and the multi-decade planning horizon suggest this is a durable policy direction, not a short-term measure.