Chevron Taps Permian Gas to Power AI Data Center Pushes

Key Moments

AI’s Energy Demands Reshape Power Strategy

Generative artificial intelligence requires extremely high and consistent power consumption. Even as investors focus on semiconductor supply chains and software monetization, the infrastructure behind large-scale AI computing is driving a significant reconfiguration of energy sourcing and delivery.

Despite substantial buildout of solar and wind capacity, these renewable sources cannot reliably deliver the near-perfect baseload uptime needed for massive AI training clusters. Without cost-effective, grid-scale storage to smooth out intermittency, these sources are insufficient as standalone solutions for hyperscale data centers that must run around the clock.

This constraint is pushing major technology companies to secure dedicated power arrangements. The newly announced long-term natural gas power purchase agreement between Chevron Corporation (NYSE: CVX) and Microsoft Corp. (NASDAQ: MSFT) illustrates this emerging model. By pairing on-site gas-fired generation with a large AI campus, Chevron is positioning itself as a critical physical enabler of AI infrastructure.

Project Kilby: A New Off-Grid Blueprint

The agreement, called Project Kilby, is structured to significantly alter the conventional role of public utilities in serving large customers. Chevron’s wholly owned subsidiary, Energy Forge One LLC, is set to develop a natural-gas-fired power plant in Reeves County, Texas. The facility will serve exclusively as the power source for a proposed Microsoft AI data center campus.

Initial plans call for capacity of up to 2.67 gigawatts with the ability to scale modularly toward as much as 5 gigawatts into the 2030s. The project carries a projected capital cost of about $7 billion.

A key design element is that the plant will operate entirely behind the meter, avoiding reliance on the Electric Reliability Council of Texas transmission network. By staying off the ERCOT system, Project Kilby is structured to sidestep local congestion, exposure to sharp fluctuations in commercial power prices, and lengthy interconnection queues that have slowed other data center developments.

The Texas grid already faces stress from rapid population gains, severe weather patterns, and a surge in industrial buildout. In this context, a fully dedicated, off-grid supply gives Microsoft a measure of insulation from broader system pressures.

Partnership and Industrial Ecosystem

Chevron is not solely financing the initiative. Engine No. 1’s energy platform, Joulent LLC, holds a 50% equity option to participate in project funding, aligning capital from an energy transition-focused investor with a major fossil fuel producer.

To deliver the physical components, Chevron has engaged GE Vernova (NYSE: GEV) and Solar Turbines, a subsidiary of Caterpillar Inc. (NYSE: CAT), for heavy equipment and turbine technology. These partners are working with Chevron to build an integrated system intended to operate independently of the public transmission grid.

Turning Permian Gas Constraints Into Long-Term Revenue

For Chevron, Project Kilby addresses a persistent challenge in the Permian Basin. The company’s upstream operations in the region generate substantial associated natural gas alongside oil production. Limited pipeline takeaway capacity in West Texas has frequently pushed Waha Hub pricing into negative territory, effectively forcing producers to pay to move gas.

Installing a 2.67 gigawatt power station within the basin directly alters this dynamic. Chevron can route its stranded, low-cost natural gas into its own generation assets instead of relying on constrained midstream infrastructure. The project transforms a transportation bottleneck into a 20-year fixed revenue stream linked to power sales, rather than spot gas pricing.

This structure shifts a portion of Chevron’s future cash flow away from the swings of global commodity markets and aligns it with the sustained expansion of AI-driven computing. With a targeted internal rate of return in the mid-teens, Project Kilby is effectively structured as a high-margin, long-duration utility-style contract that provides added earnings visibility for the integrated energy major.

Funding Capacity Backed by Hess Integration

Investors evaluating Chevron’s capacity to pursue large-scale AI-linked energy investments must consider its current financial position. After closing its $53 billion acquisition of Hess Corporation in July 2025, Chevron added material upstream assets in the Bakken shale and offshore Guyana to its portfolio.

These additions have already bolstered profitability. In the first quarter of 2026, Chevron reported adjusted earnings per share of $1.41, exceeding consensus expectations by roughly 42%. Total global production increased 15% year over year to a record 3.86 million barrels of oil-equivalent per day.

Even with a period of significant downstream margin pressure, which resulted in an $817 million quarterly loss in refining and marketing, the strength of the upstream segment supported what management characterizes as a fortress balance sheet. Chevron maintained a debt-to-equity ratio of about 0.25.

Capital Returns and Valuation Metrics

Institutional support remains substantial. Large investors continue to hold meaningful positions, and short interest stands at just 1.16% of free float. This limited bearish activity aligns with Chevron’s assertive capital return profile.

In the first quarter of 2026, Chevron distributed $6 billion to shareholders, including $3.5 billion in dividends and $2.5 billion in share repurchases. The shares trade at a forward price-to-earnings multiple of roughly 11 and currently offer a dividend yield of 4.04%.

Chevron has also recorded 38 consecutive years of annual dividend increases. These consistent payouts provide income support as investors await Project Kilby’s planned initial power output in 2028.

Natural Gas as a Bridge Fuel for AI Expansion

The significance of Project Kilby extends beyond a single data center complex. The agreement implicitly acknowledges that ambitious carbon-neutral objectives in the technology sector are being tested by the energy intensity of AI. On their own, variable renewables are not sufficient to supply the continuous load required by large-scale machine learning operations.

Although other zero-carbon technologies such as next-generation nuclear may play a role over time, localized natural gas-fired generation currently represents a readily deployable and scalable option to meet escalating data center demand over an extended period.