Texas politics may delay Chevron data centre decision, Semafor reports

A final investment decision for Project Kilby, a proposed 2.67GW gas-powered data center near Pecos in West Texas, may be delayed from late 2026 to 2027 due to state political issues. The delay creates execution risk for a major AI/data-center power infrastructure project in the Permian Basin and could defer associated gas-turbine and electricity-demand investment.

Analysis

The relevant exposure is not the data-center developer but the deferred order stack: GE Vernova (GEV), Siemens Energy (ENR GY), Caterpillar (CAT), Eaton (ETN), Vertiv (VRT) and Quanta Services (PWR) could see a modest shift in the timing of turbine, switchgear, cooling and interconnection demand. A multi-gigawatt behind-the-meter design is unusually turbine-intensive, but any single project remains immaterial to GEV and ETN earnings; the market implication is a lower probability that hyperscale AI demand converts into 2026 equipment revenue rather than a change in the long-run demand thesis.

For Permian midstream, an eventual load of this size could absorb roughly 0.35-0.45 Bcf/d of gas depending on operating rate and heat efficiency, supportive to local basis and gathering utilization. The better second-order beneficiaries are Western Midstream (WES), Enterprise Products (EPD), Kinder Morgan (KMI), Energy Transfer (ET) and Targa (TRGP), although only where contracted transport and processing capacity links to the Pecos corridor. A political delay extends the period in which associated-gas growth depresses Waha pricing, favoring producers with firm takeaway or LNG-linked realizations over unhedged basin exposure.

Consensus is likely to treat any slippage as an AI-power demand warning. That is too broad: developers can substitute locations, grid interconnection structures, or smaller phased deployments, while Texas policy uncertainty may actually increase the value of existing permitted generation and transmission assets. The near-term risk is concentrated in premium-multiple power-infrastructure names if investors begin discounting a broader pipeline of self-powered campuses; the thesis is falsified if equipment backlog, bookings, and 2026 revenue guidance at GEV/ETN/VRT remain intact through the next two reporting cycles.

Key Decisions for Investors

• Do not initiate a directional trade solely on this development; set a 1-3 month alert for a second major Texas AI-campus delay or cancellation, which would make 2026 backlog risk for VRT, ETN and PWR more actionable.
• Maintain a 6-18 month preference for long GEV or ETN versus short VRT on a pair basis: turbine and electrical-distribution demand is less dependent on a single campus than VRT's high-expectation data-center cooling narrative. Reassess if GEV or ETN reduce 2026 order/backlog guidance by more than 5%.
• For Permian exposure, favor long WES or EPD over a basket of Permian gas-sensitive producers until a binding gas-supply/takeaway contract is disclosed. The upside requires a final investment decision and contracted volumes; exit if Waha basis tightens materially without corresponding pipeline-volume commitments.
• Use any broad 5-10% infrastructure selloff tied to this headline to add selectively to GEV/ETN only after confirming that the projects represent no meaningful portion of disclosed backlog; avoid treating a permitting timeline as a cancellation.