Sources: Microsoft is in talks with Chevron and investment fund Engine No. 1 over a $7B Texas power plant that would initially generate 2.5 GW of electricity
Microsoft Corp. is in exclusive talks with Chevron Corp. and investment fund Engine No. 1 over a long-term deal that would underpin …
Context & Ripple Effects
Microsoft had just arranged to lease roughly 700 MW of Abilene data-center capacity from Crusoe after Oracle and OpenAI stepped away, making a separate, much larger power-backed project a meaningful expansion of its Texas infrastructure options. The prospective arrangement also pairs Microsoft with Chevron, a company it had previously approached about using AI in oil-and-gas development, extending that relationship from software customer to potential energy supplier.
The later 20-year natural-gas power agreement for a proposed West Texas data center indicates that the talks progressed into a long-duration supply relationship. Taken together, the coverage shows Microsoft seeking to lock in electricity alongside compute capacity rather than treating grid access as an external dependency.
First-order effects
- Microsoft gains a potential dedicated power pathway for a Texas data-center buildout, while Chevron and Engine No. 1 gain a prospective long-term anchor customer for a capital-intensive plant.
- The proposed plant’s initial 2.5 GW scale would materially exceed the Abilene leasing arrangement for roughly 700 MW, broadening Microsoft’s near-term options for placing power-hungry workloads in Texas.
Second-order effects
- A long-term corporate offtake can make financing and construction of new generation more feasible, shifting more project-development risk from a data-center operator onto an energy developer and its capital partners.
- Other large compute buyers competing for Texas capacity may face stronger incentives to secure generation, fuel supply, or similarly durable power contracts rather than rely solely on available data-center leases or grid connections.
Third-order effects
- If these structures proliferate, power procurement becomes a core component of AI infrastructure strategy: hyperscalers increasingly underwrite generation projects to obtain dependable powered capacity.
- The trade-off is greater execution and fuel-price exposure embedded in compute expansion; the durability of this model will depend on whether long-term contracts can reliably deliver power on the timetable data-center projects require.
The trend: AI infrastructure is moving toward long-duration, power-linked compute offtake, with major cloud buyers using contractual commitments to turn electricity availability into financeable capacity.