Chevron signs 20-year Microsoft power deal for 2.67 GW Project Kilby in Texas

Chevron Corporation (NYSE: CVX) has signed a 20-year agreement with Microsoft Corporation (NASDAQ: MSFT) to develop Project Kilby, a proposed 2.67 GW natural gas-fired power facility co-located with a Microsoft data-centre campus in West Texas. The agreement was announced on June 22, 2026, through Chevron subsidiary Energy Forge One LLC and gives the project a long-duration customer before final investment approval. Project Kilby is expected to begin delivering electricity in 2028 and expand through a phased, modular construction programme. GE Vernova Inc. (NYSE: GEV) will supply most of the generation equipment, while Caterpillar Inc. (NYSE: CAT) subsidiary Solar Turbines will provide additional capacity. The project could give Chevron a large, contract-backed power business linked to artificial intelligence demand, but permitting, capital cost, natural gas supply, turbine delivery and environmental scrutiny remain significant barriers before construction can begin.

Why does Chevron’s 20-year Microsoft agreement change the economics of Project Kilby?

The Microsoft agreement gives Project Kilby something many proposed data-centre power developments still lack: an identified customer willing to commit to electricity purchases for two decades. That long-term revenue visibility could support project financing, justify turbine reservations and reduce Chevron’s exposure to unpredictable wholesale electricity prices.

Without a power purchase agreement, Chevron and its partners would have to build the facility largely around forecasts of future demand and merchant power values. Microsoft’s participation instead connects generation capacity with a customer whose cloud and artificial intelligence operations require large volumes of electricity around the clock.

The agreement does not yet guarantee construction. Chevron still intends to reach a final investment decision by the end of 2026, indicating that several commercial, regulatory and technical conditions remain unresolved. The company must establish that the contract price, development cost and operating assumptions can deliver returns competitive with its upstream oil and gas portfolio.

Project Kilby’s economics may also differ from those of a conventional utility plant. Electricity will initially be supplied directly to a co-located customer rather than sold primarily through the Electric Reliability Council of Texas market. This can reduce transmission dependence and market-price exposure, but it concentrates revenue risk around one customer and one computing campus.

Microsoft’s financial strength makes counterparty default unlikely, but concentration still matters. Changes in data-centre design, technology efficiency or capital allocation could affect future load growth. Contract protections, minimum-payment obligations and expansion provisions will therefore be central to financing even though their details remain confidential.

How will Project Kilby convert Permian natural gas into dedicated AI data-centre power?

Project Kilby will be built in the Permian Basin region, close to one of the largest natural gas-producing areas in the United States. Chevron intends to use locally available gas to fuel generation rather than depend on electricity transported from distant power plants through an already constrained transmission system.

This location creates an unusual alignment between energy supply and electricity demand. West Texas periodically experiences weak or even negative natural gas pricing when production exceeds available pipeline capacity. A large local power facility could create an additional market for gas that might otherwise face transportation discounts.

The economic relationship is not automatic. Project Kilby will require dependable fuel-delivery infrastructure capable of supporting multi-gigawatt generation through extreme weather, maintenance periods and changing production conditions. Chevron and its partners must secure pipelines, compression, storage or alternative supply arrangements that prevent a temporary upstream disruption from interrupting Microsoft’s computing operations.

The project also creates an opportunity for Chevron to capture value beyond selling gas at the wellhead. Instead of receiving only a commodity-linked price, Chevron could convert gas into electricity and sell that power under a long-duration agreement. This shifts part of its earnings exposure from volatile natural gas prices toward contracted infrastructure cash flow.

However, fuel-price allocation will determine who carries commodity risk. The power agreement may include fixed prices, escalation mechanisms, gas-indexed components or cost pass-through clauses. A fully fixed electricity price would leave the project more exposed to fuel inflation, while a pass-through model would transfer more gas-price risk to Microsoft.

Why is the 2.67 GW phased design central to Chevron’s execution and capital strategy?

Project Kilby is expected to reach approximately 2.67 GW through a phased and modular construction programme rather than entering operation at full capacity on one date. This allows Chevron to align equipment installation with Microsoft’s data-centre expansion and avoid investing the entire amount of capital before computing demand is ready.

Phased development can reduce construction risk because each module can be tested and commissioned before the next phase is completed. Problems discovered during the first unit can be corrected in later stages, potentially improving schedule reliability and operating performance.

The structure also supports incremental financing. Chevron may be able to release capital in stages rather than funding the full project upfront. Microsoft’s data-centre campus is expected to expand over several years, giving both parties time to coordinate power capacity with building completion and server deployment.

A modular design does not remove coordination risk. The gas plant, electrical systems, data-centre buildings, fuel pipelines and water infrastructure must be delivered in a sequence that prevents completed assets from sitting idle.

The 2028 first-power target is especially demanding because large gas turbines, transformers and switchgear are subject to long manufacturing lead times. GE Vernova has experienced exceptionally strong demand from utilities and data-centre developers, meaning Chevron must secure production slots early enough to protect the schedule.

Project Kilby’s ultimate scale is comparable to several large conventional power stations. Building that capacity in stages may look more manageable, but the cumulative construction programme will still require significant engineering, labour and capital over several years.

Can co-located generation avoid ERCOT bottlenecks without creating a private-grid island?

The central attraction of Project Kilby is that generation will be built close to Microsoft’s electricity demand. This reduces the project’s dependence on waiting for new transmission lines or obtaining the full capacity required from the regional grid before the data centre begins operating.

Large industrial customers can wait years for utility interconnections because network studies, transmission construction and equipment procurement take considerable time. A co-located model offers a faster route by placing electricity generation and consumption on the same campus.

Joulent, the Engine No. 1-backed power company supporting the project, describes the model as an across-the-meter approach. Electricity is delivered directly to the customer while the facility may eventually gain the ability to connect with and export surplus capacity into the wider grid.

That future connection could create additional value. Project Kilby may provide backup or incremental electricity to ERCOT when Microsoft is not consuming the full available output, improving asset utilisation and potentially supporting regional reliability.

The challenge is ensuring that a private power campus does not become operationally isolated. Microsoft requires exceptionally high reliability, meaning the site may still need grid access, backup generation, battery storage or other redundancy when turbines undergo maintenance or unexpected outages.

A system with several turbine units can reduce single-equipment risk, but common infrastructure remains important. Failures involving the fuel pipeline, switchyard, cooling system or control network could affect several generators simultaneously.

Project Kilby must therefore combine the speed of private generation with the resilience normally associated with a diversified utility system. Avoiding grid queues is useful, but replacing one bottleneck with dependence on one campus would not represent much of an engineering victory.

What do GE Vernova and Caterpillar gain from Project Kilby’s turbine procurement plan?

GE Vernova is expected to supply most of Project Kilby’s generation capacity together with associated electrical infrastructure. Caterpillar subsidiary Solar Turbines will provide additional capacity, likely supporting modular generation, auxiliary services or parts of the campus requiring smaller turbine configurations.

The award strengthens GE Vernova’s position as one of the largest industrial beneficiaries of artificial intelligence electricity demand. Data centres require not only gas turbines but also generators, transformers, switchgear, control systems and long-term maintenance services.

Long-term service contracts could prove almost as valuable as the initial equipment order. Large turbines require inspections, component replacement, outage planning and performance optimisation over decades, creating recurring revenue after construction ends.

Project Kilby also provides GE Vernova with a reference for co-located hyperscale power infrastructure. A successful project could encourage other technology companies and energy developers to adopt similar configurations, increasing demand for repeat equipment packages.

Caterpillar gains exposure through Solar Turbines, which manufactures industrial gas turbines used in power generation and oil and gas applications. Smaller turbine units can provide flexibility, black-start capability and staged capacity where deploying another large turbine would be inefficient.

The supplier opportunity comes with execution pressure. GE Vernova and Caterpillar must deliver equipment during a period of strong global demand and stretched manufacturing capacity. Late turbines could delay both electricity generation and Microsoft’s associated computing infrastructure.

How could water use, emissions controls and future solar additions shape approval?

Large gas-fired power facilities can require substantial water for cooling and other operating processes, making water availability a politically sensitive issue in arid West Texas. Project Kilby plans to use non-potable brackish groundwater rather than rely principally on freshwater supplies used by households and agriculture.

Chevron is also exploring the reuse of produced water generated by oil and gas operations. This could create a second use for a waste stream that otherwise requires treatment, transportation, recycling or disposal.

Produced-water use is technically demanding because salinity, hydrocarbons, minerals and other contaminants can damage power-plant equipment if treatment is inadequate. The commercial case will depend on treatment cost, water quality, pipeline distance and regulatory approval.

Project Kilby’s design is expected to include selective catalytic reduction systems to reduce nitrogen oxide emissions, together with measures intended to limit noise and light affecting nearby communities. These controls can improve the project’s permitting position but add capital and maintenance requirements.

Carbon dioxide remains the largest environmental issue. A 2.67 GW natural gas plant operating at high utilisation would produce significant emissions even if it is more efficient and less carbon-intensive than coal-fired generation.

Chevron and Joulent have left open the possibility of adding renewable generation, beginning with solar power. Solar could reduce gas consumption during daylight hours, while battery storage could improve flexibility and provide short-duration backup.

The companies have also previously discussed designing data-centre power facilities with flexibility for lower-carbon technologies such as carbon capture and storage. No binding carbon-capture investment has been announced for Project Kilby, meaning it should not be treated as part of the initial project configuration.

Why is Project Kilby strategically important for Chevron’s move beyond commodity exposure?

Chevron’s traditional earnings depend heavily on oil and natural gas prices, production volumes, refining margins and chemicals cycles. Project Kilby creates the possibility of long-term electricity revenue linked to data-centre demand rather than directly to daily commodity movements.

A 20-year power agreement could produce more stable cash flow than selling uncontracted electricity into ERCOT or marketing natural gas at regional prices. This can improve portfolio diversification while still using Chevron’s existing expertise in gas supply, engineering and large-project execution.

The project also gives Chevron access to a fast-growing customer group. Technology companies are becoming some of the largest buyers of new electricity capacity as artificial intelligence models require increasingly powerful computing facilities.

Chevron is not alone in pursuing this opportunity. Exxon Mobil Corporation and other energy groups are developing concepts combining gas generation, carbon capture and dedicated data-centre supply. Utilities, independent power producers and infrastructure funds are also competing for hyperscaler contracts.

Chevron’s advantage is its access to Permian natural gas, balance-sheet capacity and experience delivering complex industrial facilities. Its disadvantage is limited operating history as a merchant or contracted electricity producer at this scale.

The company will need to prove that a major oil producer can manage power-market economics, turbine operations and data-centre reliability without allowing the project to become a costly strategic experiment.

Project Kilby becomes particularly attractive if Chevron can replicate the model. One project may produce useful returns, but a pipeline of standardised multi-gigawatt developments could create a new business platform with meaningful earnings.

What does the Microsoft agreement reveal about hyperscaler power procurement?

Microsoft’s participation shows that large technology companies are no longer willing to rely entirely on utilities and wholesale electricity markets to deliver the power needed for artificial intelligence growth. The pace of data-centre investment is moving faster than traditional grid expansion in several United States regions.

Dedicated generation gives Microsoft greater control over schedule and reliability. The company can coordinate server deployment, building construction and electricity capacity rather than waiting for uncertain grid-upgrade timelines.

The agreement also reveals a tension within corporate decarbonisation strategies. Microsoft has contracted large amounts of renewable and carbon-free electricity, but Project Kilby is based initially on natural gas because the company needs firm power at a speed and scale that intermittent resources alone may struggle to provide.

The company can still pursue renewable-energy matching, carbon removal and future solar additions, but the physical electricity serving the West Texas campus will include substantial gas-fired generation.

The commercial logic is understandable. Artificial intelligence infrastructure can lose enormous economic value when power is unavailable, making continuous supply more important than securing the lowest theoretical generation cost.

The reputational challenge will be demonstrating that the new gas capacity does not undermine Microsoft’s wider emissions commitments. The company may face pressure to disclose Project Kilby’s expected emissions, renewable additions and any future carbon-capture plan.

How should investors interpret CVX, MSFT, GEV and CAT share performance after the deal?

Chevron shares closed at $175.98 on June 23, 2026. The stock was down approximately 2.3% from its June 16 close and about 8.1% from May 22, while trading within a 52-week range of $142.40 to $214.71.

The decline indicates that oil prices and wider energy-sector sentiment remain more important to Chevron’s near-term valuation than Project Kilby. The power agreement is strategically notable, but final investment approval has not yet occurred and revenue is not expected before 2028.

Microsoft shares closed at $373.94. The stock was approximately 5.1% lower than its June 16 close and about 10.7% below its May 22 level, within a 52-week range of $356.28 to $555.45.

Microsoft’s weakness reflects broader scrutiny of capital spending, artificial intelligence returns and technology-sector valuation. Project Kilby helps address electricity availability but also illustrates how much physical infrastructure Microsoft must support to expand computing capacity.

GE Vernova shares closed at $1,034.98 after a sharp decline on June 23. The stock remained approximately 5.4% above its June 16 close but was broadly flat compared with May 22, within a 52-week range of $482.20 to $1,181.95.

GE Vernova’s medium-term performance reflects strong expectations around gas turbines, grid equipment and data-centre electricity demand. Project Kilby strengthens the order opportunity, but the company must convert exceptional demand into manufacturing output and profitable delivery.

Caterpillar shares closed at $984.24, approximately 4.1% above the June 16 close and 11.9% higher than May 22. The shares traded within a 52-week range of approximately $367.92 to $1,023.29.

Caterpillar’s proximity to its annual high indicates strong sentiment around infrastructure, power and industrial demand. Solar Turbines gives the company direct exposure to the expansion of distributed and co-located gas generation.

What risks could prevent Project Kilby from reaching final approval and delivering power in 2028?

The first risk is project economics. Chevron has not disclosed the total construction cost or electricity price under the Microsoft agreement. The final investment decision will depend on whether expected returns remain competitive after equipment, fuel, financing and environmental costs are included.

The second risk is permitting. A multi-gigawatt gas facility will require air-quality, water, construction and potentially grid-related approvals. Environmental groups may challenge the project because of its expected carbon emissions and the wider debate around fossil-fuel generation for data centres.

The third risk is equipment availability. GE Vernova’s large turbines and high-voltage electrical systems have long production lead times. Manufacturing or component delays could undermine the 2028 first-power target.

The fourth risk is natural gas infrastructure. Project Kilby needs dependable fuel volumes and adequate pipeline capacity even during extreme cold or disruptions elsewhere in the Permian Basin.

The fifth risk is water management. Brackish groundwater and produced water reduce pressure on freshwater resources but require treatment systems capable of protecting equipment and meeting environmental rules.

The sixth risk is integration. Generation modules, Microsoft data halls, substations, fuel systems and digital controls must be commissioned in a coordinated sequence.

The seventh risk is technology demand. Microsoft expects continued growth in artificial intelligence and cloud workloads, but faster improvements in computing efficiency or changes in platform strategy could alter long-term electricity requirements.

The eighth risk is concentration. Project Kilby depends heavily on one customer, while Microsoft depends on one large co-located generation platform. Contractual protections and physical redundancy must therefore be unusually strong.

Which milestones will show whether Project Kilby is becoming a bankable gas-to-power platform?

The first milestone will be Chevron’s final investment decision, currently expected by the end of 2026. That decision would confirm that capital costs, customer terms and projected returns meet the company’s investment threshold.

The second will be definitive turbine and electrical-equipment contracts with GE Vernova and Solar Turbines. Delivery schedules will reveal whether first power in 2028 remains realistic.

The third will be construction and environmental permits covering air emissions, water use, site development and fuel infrastructure.

The fourth will be disclosure of financing and ownership. Chevron, Engine No. 1 and Joulent must clarify who contributes equity, owns the completed facility and carries construction risk.

The fifth will be commencement of site preparation and gas-pipeline work. Physical construction would move Project Kilby beyond a customer-backed development proposal.

The sixth will be commissioning of the first generation module. This will determine whether Chevron can deliver usable electricity before the entire 2.67 GW campus is complete.

The seventh will be evidence that later phases remain supported by Microsoft’s computing expansion and contractual commitments.

Project Kilby is strategically stronger than a conventional gas-plant proposal because it combines fuel access, a long-term customer and a modular data-centre campus. The central uncertainty is whether Chevron can translate those advantages into a competitive capital cost and dependable 2028 delivery.

What are the key takeaways from Chevron’s Project Kilby agreement with Microsoft?

• Chevron has signed a 20-year agreement to supply a Microsoft data-centre campus through Project Kilby in West Texas.
• The proposed facility will provide approximately 2.67 GW through a phased natural gas generation programme.
• Project Kilby is expected to begin delivering electricity in 2028, subject to final investment approval and permitting.
• GE Vernova will supply most of the generation capacity, while Caterpillar subsidiary Solar Turbines will provide additional equipment.
• Co-located generation could avoid lengthy grid interconnection delays and reduce pressure on the existing ERCOT network.
• Permian natural gas provides a nearby fuel source and creates an alternative market for regionally constrained production.
• Brackish groundwater and potential produced-water reuse are intended to reduce dependence on freshwater resources.
• The undisclosed construction cost, power price and ownership structure prevent a complete assessment of project returns.
• CVX and MSFT shares weakened over the past month, while GEV and CAT retained stronger power-infrastructure momentum.
• Final investment approval, turbine contracts, permits and site construction will determine whether Project Kilby becomes a repeatable AI power model.