Chevron signs 20-year power supply partnership with Microsoft for AI data centers

Chevron has agreed to a 20-year partnership with Microsoft to supply natural-gas-fired power for AI data centers, according to a report published by Yahoo Finance. The arrangement is described as a Project Kilby partnership, a long-dated supply framework aimed at matching the growing electricity needs of companies building large-scale AI computing facilities.

Under the deal, Chevron would be the supplier of generation fuel-linked power for data center operations tied to Microsoft’s AI efforts, with the contract duration extending two decades. For Chevron, the structure points to a move beyond traditional oil and gas sales toward longer-term participation in electricity demand driven by cloud computing and machine-learning workloads.

The announcement also underscores how AI infrastructure is increasingly viewed as an energy procurement problem, not just a technology problem. Data centers require continuous power and often face grid constraints, prompting buyers to look for dedicated or firmer capacity arrangements that can be planned over many years.

Neither the Yahoo Finance report nor the information provided in this prompt specifies the exact geographic footprint for the power, the expected volumes, the pricing method, or whether Chevron will supply power directly through generation assets or through a contracting arrangement that relies on third-party power delivery. It also does not detail any timeline for when deliveries are expected to start.

Microsoft, as an AI hyperscaler, has strong incentives to secure power supply for data center build-outs that support training and inference at scale. For energy companies like Chevron, long-dated contracts can help shift exposure away from commodity price swings and toward contracted cash flows tied to electricity demand growth.

Still, several key elements remain unclear based on the material available here. The report does not describe the ownership or operation of any power generation facilities, any emissions-related requirements tied to the natural-gas-fired power, or whether the arrangement includes expansion options as data center capacity grows. It also does not provide contract sensitivity details, such as what happens if planned AI capacity is delayed or if grid conditions change.

For the market, the deal fits a broader pattern of technology companies and energy providers seeking durable relationships that can support infrastructure build-outs. With power increasingly central to AI economics, buyers may continue to pursue long-horizon agreements that can translate energy availability into better planning certainty for data center development.