How Chevron Became AI's Quiet Oil Trade

Microsoft's 20-year deal made the AI-power idea concrete for Chevron investors

This is where the AI-energy theme moves from narrative to execution. Chevron announced a 20-year power purchase agreement with Microsoft for Project Kilby, a co-located power facility in West Texas tied to a data center expected to draw nearly 2.7 gigawatts-about the power needed for 2 million homes. For an oil company, that scale is large enough to matter strategically, even if the revenue will not show up for years.

Why Project Kilby matters

The key word is co-located. The power infrastructure will be built and located at the West Texas site, which gives Chevron a direct way to turn its oil-and-gas execution skills into AI infrastructure optionality. Instead of only selling a commodity and waiting for price, it is securing a long-duration customer for steady electric output. A 20-year contract also suggests this is more than a pilot.

Why the timing matters now

Chevron's June announcement turned a broad theme into a live project with a visible decision window: the company expects to make its final investment decision later this year, with power delivery expected in 2028.

• Bulls see the first concrete bridge from Big Oil into AI power, with enough scale and contract length to change how investors think about Chevron.
• Bears see a cash register that still sits in the future, with no investment decision in place and no revenue yet to underwrite.

If Chevron pulls the trigger this year, investors get proof before the earnings show up. If not, the story remains interesting but less urgent.

Permian gas and behind-the-meter power are the practical setup

The real constraint in AI buildouts is increasingly power, not processing power alone. Microsoft has said it needs energy infrastructure that can scale quickly and reliably, and Project Kilby is designed to avoid grid bottlenecks by putting generation right next to the load.

Why behind-the-meter matters

Chevron has also laid out a 2.5-GW natural gas–fired facility in the Permian Basin designed for a co-located AI data center through a behind-the-meter, or off-grid, structure, with room to expand to 5 GW. In plain English, the power is generated next door instead of competing for limited grid capacity.

That helps explain why the model keeps coming back to the Permian. Chevron's approach leans on local gas supply and on placing generation close to the data center. The Project Kilby arrangement uses large gas turbines provided by GE Vernova, with Caterpillar also involved, and the power will be dedicated to the data center rather than tied to the grid.

Why Permian gas fits the economics

A nearby fuel supply matters when you are building at this scale. A Permian-based setup can, in theory, keep the supply chain shorter and the cost base simpler if local gas remains available and affordable.

This also looks less like a one-off headline and more like a direction Chevron has already been moving. The company has been in talks for more than a year about supplying natural gas-fired power to data centers, and Chevron and Williams have been framed as winners in the push to build gas-fired power plants and pipelines for AI.

The next catalyst is the investment decision

If that decision comes through, investors get evidence that oil companies can become part of AI's power toolkit, not just suppliers of crude. If it slips, the narrative keeps its long-term logic but loses near-term urgency.

Chevron still trades like an oil stock

Right now, Chevron still looks like a classic energy holding company on screen. The stock sits at $171.45, while the mid-range analyst target is only ~$175. The broader street target is ~$217, but that upside is measured in total return, not in an immediate rerating around an AI-power story.

That leaves the burden of proof on management. Bulls can point to a 20-year power purchase agreement with Microsoft and a facility sized at roughly 2.7 gigawatts. Bears can say the same contract may remain a future option rather than a current earnings driver for some time.

The valuation test

The cleanest way to frame the debate is simple: how much of Chevron's current price reflects this new power business, if any?

• Bull case: If the project clears Chevron's usual return hurdles, investors may eventually value a contracted revenue stream that is less tied to spot crude, can be developed project by project, and comes with a major counterparty.
• Bear case: The first power date is still expected in 2028, the final investment decision is still pending, and construction has yet to begin. Until those milestones advance, Chevron is still an oil stock by valuation, not by earnings mix.

What would confirm the thesis-and what would break it

One contract does not make a new business line. Project Kilby is best viewed as a prototype, not proof of a repeatable franchise.

The follow-on pipeline matters because it shows whether Chevron can do this more than once. That includes the 2.5-GW Permian facility that could expand to 5 GW and efforts to pursue additional gas-fired data-center plants.

What confirms it

• Chevron makes the final investment decision later this year for the Microsoft-linked project.
• Construction starts, showing the deal is moving beyond paperwork.
• Additional customers or sites follow, indicating repeatability.
• The planned Microsoft site still begins receiving power around the expected date.

What breaks it

• The investment decision slips or never arrives.
• Chevron announces more equipment or site work, but no second customer or site follows.
• Execution delays push the power date out.

Patience matters because this is still a build-first, cash-register-later business. Chevron's own framing is that this is its first large-scale facility designed to supply electricity to a data center. Watch for repeatability, speed, and follow-on deals. If those appear, the prototype starts to look like a platform. If not, it remains a promising experiment.