Texas Is Winning the AI Data Centre War, One Gigawatt at a Time

Texas has become the most important American state for artificial intelligence infrastructure, and the reason is electricity. Where California designs models, Texas increasingly runs them.

The energy proposition:

Texas operates its own electricity grid, largely separate from the rest of the United States, managed by ERCOT. That separation, combined with a deregulated market structure, produces faster interconnection for large new loads than most states can offer, where queues to connect major projects can stretch for years.

Generation is abundant and diverse: substantial natural gas capacity, the largest wind fleet in the country, and rapidly growing solar and battery storage. Land is cheap and plentiful, permitting is comparatively fast, and local governments generally welcome industrial development.

For data centre developers, the combination of available power, available land and fast approvals is decisive, and capital has flowed accordingly. Large AI computing facilities have been announced and built across the state, some at scales measured in gigawatts.

The grid question:

The obvious concern is whether the grid can absorb this. Texas has experienced serious grid failures during extreme weather, and adding enormous new industrial loads to a system with limited connections to neighbouring grids raises legitimate reliability questions.

The response has been partly technical and partly contractual: large data centres increasingly agree to curtail consumption during peak stress, effectively acting as flexible load that can be reduced when the grid is strained. Some developers build dedicated generation alongside facilities.

Whether flexibility commitments hold under real emergency conditions, and whether residential consumers ultimately bear costs from grid expansion driven by industrial demand, are the central policy questions.

Beyond infrastructure:

Texas has substantial technology employment beyond data centres. Austin hosts significant operations for major technology companies, including chip design and engineering, and has developed a genuine startup ecosystem. Dallas and Houston host large corporate technology functions.

Semiconductor manufacturing is significant and expanding, with major fabrication investment supported by federal incentives, giving Texas a position in chip production as well as chip usage.

Houston's energy industry is a sophisticated AI user: seismic interpretation, reservoir modelling, predictive maintenance and trading analytics are established applications, and the same engineering culture is being applied to newer energy technologies.

Healthcare, with the Texas Medical Center among the world's largest medical complexes, provides research applications in diagnostics and clinical operations.

Constraints:

Water is a genuine issue. Data centre cooling requires water, parts of Texas face long-term water stress, and local opposition has emerged around specific projects.

Transmission capacity constrains where facilities can connect, and building new lines takes years.

The political environment cuts both ways: light regulation attracts development, and social policy has affected some companies' willingness to relocate staff.

The outlook:

Texas is likely to remain the leading American location for AI compute for the foreseeable future. Its advantages, power availability, land, speed of approval, are precisely what the industry needs most and what other states struggle to provide.

The risks are physical rather than economic: grid stability under extreme weather with much larger loads, and water availability in a warming climate. Both are manageable and neither is solved.