Chevron Starts Selling Power, Not Barrels: Inside Project Kilby and the Microsoft Deal
On 22 June 2026 Chevron signed a 20 year power purchase agreement with Microsoft and committed to build a 2.67 gigawatt gas plant in West Texas that never touches the public grid. It is the clearest signal yet that a supermajor has found a new customer, a new contract shape and a new buying committee. This dossier unpacks the logic, the money, the risk, and what it changes for anyone selling into big energy.
- Chevron is monetising cheap Permian associated gas as long dated contracted electricity rather than as a commodity. The customer is a hyperscaler, the contract is 20 years, and the revenue is no longer exposed to the oil price.
- Power, not gas, is the binding constraint on AI growth. Chevron CEO Mike Wirth told CERAWeek in March 2026 that power is becoming the great limiting element for growth, and that you cannot take a big extension cord to the grid and plug in a data centre.
- Behind the meter is the point. Kilby bypasses the interconnection queue entirely, which is why speed to power, not price per megawatt hour, is the currency of these deals.
- The buying committee has widened. Turbine OEMs (GE Vernova, Solar Turbines), financial structuring partners (Engine No. 1), state and local incentive counsel, and emissions mitigation vendors all now sit inside a decision that used to belong to upstream procurement.
- The template will be copied, and so will the backlash. Chevron and GE Vernova are separately progressing 4 gigawatts of behind the meter power foundries, while Texas legislators have ordered a study into data centre tax incentives projected to reach 3 billion dollars by 2029.
A 20 year contract, a 2.67 gigawatt plant, and no grid
On 22 June 2026 Chevron announced that its subsidiary Energy Forge One LLC had signed a 20 year power purchase agreement with Microsoft for a co located power facility in West Texas (Chevron newsroom, 22 June 2026). The project, known as Project Kilby, spans more than 2,000 acres in Reeves County near Pecos, and is expected to deliver approximately 2.67 gigawatts of capacity in a phased, modular build.
Generation comes mainly from large GE Vernova 7HA gas turbines, with additional capacity from Solar Turbines, a Caterpillar subsidiary (TechCrunch, 22 June 2026). First power is targeted for late 2028, with the final investment decision expected by the end of 2026. Chevron says the project will generate more than 10 billion dollars in state and local tax revenue and support nearly 2,000 jobs. Reported project value is around 7 billion dollars, a figure carried by multiple outlets and treated here as an estimate rather than a Chevron confirmed number.
The single most important structural fact is the one that is easiest to skim past. The plant runs independently of the Texas grid. It does not join the ERCOT interconnection queue, it does not wait behind other projects, and it does not share its output. It is a private utility built for one customer.
Project 54Behind the meter: a 2.67 gigawatt gas plant built for one customer, bypassing the grid queue that everyone else is stuck in.The logic: stranded gas, a starving customer, and a contract that outlasts the oil cycle
Chevron produces very large volumes of associated gas in the Permian Basin, gas that comes up with the oil whether anyone wants it or not. In parts of West Texas that gas has traded at or below zero. Selling it as a commodity into a saturated regional market destroys value. Converting it into electricity and selling that electricity to a single creditworthy buyer on a 20 year contract does the opposite.
On the other side of the trade sits a customer with a problem that money alone cannot solve. Interconnection queues in the United States run for years. Utilities cannot add firm capacity at the speed that AI capital expenditure is being committed. Mike Wirth, Chevron's chairman and chief executive, put it plainly at CERAWeek in Houston in March 2026: "What you're seeing is these two worlds coming together, and power really is becoming the great limiting element for growth. What's really concerning people is access to power, so you see a lot of creative deals being done" (Fortune, 1 April 2026). He added that the technology sector has realised "you can't take a big extension cord to the grid and plug in a data center."
Read the deal as a swap. Chevron gives up commodity optionality and takes on construction and operating risk. In exchange it gets two decades of contracted, oil price independent cash flow from an investment grade counterparty. For a company whose 2026 organic capital programme is 18 to 19 billion dollars, that is a meaningful diversification of the revenue base, not a side project.
Three shifts that matter more than the megawatts
The megawatt number will be beaten by someone within a year. The structural changes will not.
From commodity to contract
Chevron is no longer selling a molecule at a spot price. It is selling a service, firm power, on a 20 year term. Revenue quality, counterparty credit and uptime now matter more than the strip.
From grid to behind the meter
By bypassing ERCOT, Chevron converts a regulatory queue into a private construction schedule. Speed to power becomes the product, and the interconnection bottleneck becomes the competitive moat.
From upstream buyer to project buyer
The people who spend the money are now power project teams, turbine OEM account managers, tax and incentive counsel, EPC firms and hyperscaler commercial negotiators. Traditional oilfield procurement is not the door any more.
Emissions, incentives, and a political clock
Two risks are already visible, and both land on Microsoft as much as on Chevron.
The first is carbon. The Environmental Integrity Project, cited by TechCrunch, modelled that Project Kilby could release more than 13 million tons of carbon dioxide annually, alongside 3,200 tons of criteria air pollutants and 278,000 pounds of hazardous air pollutants. A separate WIRED analysis, reproduced by Grist on 17 May 2026, put the Energy Forge plant's emissions at more than 11.5 million tons of carbon dioxide equivalent a year, more than Jamaica's 2024 national emissions. Both figures are third party model estimates, not Chevron confirmed disclosures, and they differ from each other, which is itself the point: a customer with a public 2030 carbon commitment is now underwriting decades of new combustion, and the numbers being argued over are not its own.
The second is fiscal politics. State comptroller documents reported by WIRED and Grist show the Energy Forge project could net more than 227 million dollars in tax savings over ten years under Texas's JETI incentive regime, with the Pecos Barstow Toyah school board approving the abatement application in February 2026. Greg LeRoy of Good Jobs First was blunt about the gap between a hyperscaler's public tax pledges and its suppliers' filings: "If they don't say, 'We will refuse tax abatements,' then they've got their fingers crossed behind their back." Texas lawmakers have already ordered a study into data centre incentives projected to cost the state 3 billion dollars by 2029. The commercial model is running ahead of the political consent for it, and that gap is where the next two years of friction will sit.
The numbers behind the shift
Kilby is the visible edge of something broader. Chevron, GE Vernova and Engine No. 1 are separately developing roughly 4 gigawatts of behind the meter gas power through so called power foundries serving co located data centres across the Southeast, Midwest and West of the United States (The Motley Fool, 12 July 2026). Goldman Sachs research cited in May 2026 projects that United States data centres will double their total electricity consumption between 2025 and 2027, and RAND projects that behind the meter capacity will roughly triple by 2030 to 49 gigawatts. Both are third party projections and should be read as estimates.
| Metric | Figure | Source | Date |
|---|---|---|---|
| PPA term | 20 years | Chevron newsroom | 22 Jun 2026 |
| Project Kilby capacity | 2.67 gigawatts | TechCrunch | 22 Jun 2026 |
| Reported project value (estimate) | approx. 7 billion dollars | Yahoo Finance / EnergyNow | 2026 |
| Projected state and local tax revenue | more than 10 billion dollars | Chevron newsroom | 22 Jun 2026 |
| Jobs supported | nearly 2,000 | Chevron newsroom | 22 Jun 2026 |
| JETI tax abatement over 10 years (estimate) | more than 227 million dollars | WIRED / Grist | 17 May 2026 |
| Chevron 2026 organic capital budget | 18 to 19 billion dollars | Oil and Gas Advancement | 2026 |
| Chevron / GE Vernova behind the meter portfolio | approx. 4 gigawatts | The Motley Fool | 12 Jul 2026 |
The commercial read for suppliers and marketers
If your account plan for Chevron still maps only to upstream engineering and oilfield procurement, it is now incomplete. The money in this deal moves through power project teams, turbine OEM relationships, financing partners, EPC firms, incentive and tax counsel, and emissions mitigation vendors. Every one of those is a buying centre that did not exist inside a supermajor's data centre budget line eighteen months ago.
Three positioning consequences follow. Speed is the pitch. Turbine backlogs are the binding constraint, so anything that shortens permitting, modularises construction or de risks balance of plant carries a premium that price alone will not beat. Incentive fluency is a differentiator, because the JETI filing, the school board vote and the emerging legislative scrutiny are now part of the deal, not administrative noise around it. And carbon mitigation is no longer optional, because the customer holds the net zero pledge, which is exactly the dynamic we mapped in the CBAM dossier and in our CSRD and CSDDD analysis: the buyer's disclosure obligation becomes the supplier's qualification gate.
The deeper lesson is one of positioning, not procurement. Chevron did not find a new market by drilling better. It found one by asking what its existing asset base was worth to a customer nobody in the industry was selling to. That is the same question behind every durable growth move we study, and it is the question most energy suppliers still have not asked of their own asset base.
Listen & take it with you
Prefer audio, or need the deck for an internal review? The full briefing is available as a podcast episode and a downloadable slide presentation.
If you sold into Chevron today, where would you put your next commercial hour?
Frequently asked
Project Kilby is a gas fired power plant that Chevron subsidiary Energy Forge One is building in Reeves County, near Pecos, Texas, designed to deliver about 2.67 gigawatts of electricity directly to a Microsoft AI data centre under a 20 year power purchase agreement announced on 22 June 2026. It runs behind the meter, independently of the ERCOT grid.
Chevron produces large volumes of low cost associated gas in the Permian Basin that is worth very little as a commodity in an oversupplied regional market. Converting that gas into contracted electricity for a single large customer on a 20 year term produces oil price independent cash flow from an investment grade counterparty, which is a materially better use of the same molecule.
A behind the meter plant supplies electricity directly to a customer's facility rather than routing it through the public grid. Data centre developers want it because grid interconnection queues can take years, and behind the meter generation lets them energise far sooner. Speed, not price, is the reason these deals get signed.
No. Chevron is not serving residential or general commercial customers and is not taking on regulated utility obligations. It is entering long term bilateral supply contracts with specific large industrial customers, which makes it closer to an independent industrial power producer than to a regulated utility.
It creates a new budget line inside a supermajor, tied to power generation for AI demand rather than to upstream capital expenditure, and it widens the buying committee to include turbine OEMs, project financiers, tax and incentive counsel and emissions mitigation vendors. Suppliers should expect speed to power, incentive fluency and carbon mitigation to become the differentiators, and should expect the model to be copied by other producers with cheap associated gas. See also our analysis of BP's strategic reset for how a different major answered the same capital allocation question.
Get the next intelligence drop
Join energy and industrial leaders getting our marketing, AI-growth and revenue-architecture intelligence, direct, no filler.
You're on the list
Welcome to The Energy Growth Brief, watch your inbox for the next dispatch.