Chevron's off-grid bet: serving AI data centers without spiking consumer electricity bills
AI demand for compute is surging, and with it, a new strain on electricity. Chevron CEO Mike Wirth outlined a strategy aimed at meeting that demand while keeping residential and small business bills in check: build "energy parks" that sit off the grid and feed hyperscale data centers directly.
"As data centers need more and more electricity, and as we're seeing pressure on the grid and electricity prices, we're working to build an energy park that's not connected to the grid, so the costs don't flow to all consumers," Wirth said. He framed it as an advantage that lets the company "convert energy into intelligence."
How the model works
Chevron plans to tap abundant U.S. natural gas to generate electricity on site for large data centers, bypassing the traditional grid and its rate structures. That means fewer transmission constraints, faster timelines, and a tighter link between fuel cost and delivered compute.
The approach is already moving from concept to build. In 2025, Chevron revealed plans for an off-grid natural gas facility in West Texas designed to power hyperscale sites. The company also announced a partnership with Engine No. 1 and GE Vernova to explore gas-fired solutions that keep up with AI's growth in compute.
Why executives should care
- Cost containment: Behind-the-fence generation can insulate data center operators from grid congestion charges and volatile retail tariffs.
- Speed to capacity: Avoiding multi-year interconnection queues can compress project schedules and reduce stranded capex risk.
- Control and reliability: Purpose-built microgrids can be engineered for high uptime, with modular expansion as compute scales.
- Public optics: Off-grid supply helps limit pass-through costs to the general public, easing community and regulatory friction.
Context: Chevron's balance sheet and momentum
Wirth noted a record year in oil production helped offset lower commodity prices. Free cash flow rose 35% even as oil prices fell roughly 15%. That financial position supports multi-year energy infrastructure bets aligned with AI demand.
Key questions for your strategy
- Emissions profile: What methane intensity, NOx controls, and potential carbon capture options are included? How will you handle REC/offset strategy for corporate targets?
- Fuel price risk: What hedging or indexation keeps delivered $/MWh predictable over 7-10 years?
- Resilience: What redundancy (N+1/2N), on-site storage, and black start capabilities are available?
- Permitting and siting: How will water use, noise, and local air permits be addressed? What's the timeline advantage vs. grid interconnection?
- Scalability: Can capacity be added in modular blocks aligned to your rack delivery and AI cluster growth?
Action items for CIOs, CFOs, and Heads of Infra
- Run a side-by-side model: grid PPA vs. behind-the-fence gas with a renewable/offset overlay. Compare TCO, timeline, and risk.
- Issue RFPs that include off-grid options with firm fuel supply, emissions controls, and clear SLAs on uptime and expansion.
- Prioritize sites near gas supply and fiber backbones. Validate logistics for construction, workforce, and long-lead equipment.
- Map corporate climate commitments to an off-grid stack: high-efficiency turbines/engines, waste-heat use, and verified offsets where needed.
- Build optionality: design for future hybridization with solar, storage, or grid intertie if economics or policy shift.
The bigger picture
AI's energy appetite is becoming a board-level issue. Off-grid models like Chevron's won't fit every site, but they provide a lever: secure compute capacity without pushing costs onto the broader grid. For leaders, the play is simple-treat energy procurement like core infrastructure, not a commodity.
For market context on data center energy use, see the International Energy Agency's view on data centers and data transmission networks and the U.S. Department of Energy's guidance on data centers and servers.
Next step for leadership teams
If you're aligning AI ambitions with infrastructure readiness, develop the talent bench in parallel with your siting and energy strategy. Explore role-based learning paths here: AI courses by job function.
Quote to remember: "The abundance of energy that this country has can translate into energy dominance and AI dominance," Wirth said. The companies that plan for both will set the pace.
Your membership also unlocks:
