Exxon Mobil is stepping into one of the biggest bottlenecks in artificial intelligence: electricity. The company has announced plans for a new power plant built to serve data centers, a move that shows how deeply AI demand is reshaping energy markets.
The project would run on natural gas, produce over 1.5 gigawatts, and be designed for customers outside Exxon’s own operations. That would make it a notable shift for an oil and gas company that already runs power plants internally, but has not used them this way for external buyers.
Why AI data centers are pulling energy companies closer
AI data centers require large, steady amounts of electricity. Tech companies are preparing for a decade in which their computing needs may rise faster than existing power systems can comfortably support.
The pressure is already visible. According to one estimate cited in the source article, nearly half of new AI data centers might not have enough power by 2027. That kind of forecast creates an opening for companies that can promise large blocks of reliable electricity.
Exxon’s proposed answer is a dedicated facility. The company described the project in an annual strategy document as “reliable, fully-islanded power with no reliance on grid infrastructure.” In plain terms, the plant is not planned as a normal grid-connected generator serving the broader power system. It would be built to supply data center demand directly.
That choice matters because many new power projects face delays when trying to connect to the grid. By avoiding that interconnection process, Exxon is positioning the plant as a faster route to power for customers that cannot wait for traditional grid upgrades.
What Exxon says it will build
The planned power plant would use natural gas and generate over 1.5 gigawatts. Exxon has not said where the facility would be located, and it did not reply to a request for comment before publication of the source article.
The company told The New York Times that the facility should be completed within the next five years. That timeline is important because energy-hungry technology companies have also shown interest in nuclear power, yet most of those nuclear projects are not scheduled to come online until the early 2030s.
Exxon’s proposal is built around three core features:
- Natural gas generation to produce a large supply of electricity for data centers.
- Carbon capture and storage, with Exxon saying it intends to capture and store over 90% of the carbon dioxide the plant produces.
- An islanded structure, meaning the project would not rely on grid infrastructure.
That combination is meant to answer two different concerns at once: the need for dependable power and the pressure to reduce carbon pollution from fossil fuel electricity. Whether the project can do both at commercial scale is the central question.
The competition is already moving fast
Exxon is not entering an empty market. Renewables remain a major competitor because they have become quick to deploy and continue to drop in price.
Google recently announced a renewable energy investment that, including partners, will total $20 billion. That investment is expected to start sending electrons to the grid in 2026. Microsoft is contributing to a $5 billion, 9-gigawatt renewable portfolio that has already made its first investment, with its inaugural solar project scheduled to come online six to nine months from now.
Those examples show why speed is not only a nuclear-versus-gas issue. Renewables can also move quickly, especially when projects are already financed and in development. For data center operators, the practical question is not just which source is cleaner or larger, but which can deliver usable electricity on a timeline that matches AI buildout plans.
Exxon’s islanded power model may appeal to customers worried about grid delays. But renewable portfolios backed by major technology companies are also being built with urgency, and their falling cost profile adds pressure on fossil fuel projects that require additional emissions-control systems.
Carbon capture is the hardest part of the pitch
The most ambitious part of Exxon’s plan is not simply that the plant would run on natural gas. It is the claim that the company intends to capture and store over 90% of the carbon dioxide produced by the facility.
Carbon capture and storage can add considerable cost to both construction and operation of a fossil fuel power plant. That makes the economics more complicated than building a gas plant alone. The technology also has a limited track record in power generation.
According to the Global CCS Institute, there are only a handful of power plants worldwide that capture some of their carbon pollution, and none of them run on natural gas. The source article notes that this may change because of tax credits available under the Inflation Reduction Act, which offer between $60 to $85 per metric ton of carbon captured and stored.
Even with incentives, commercial performance is still uneven. Some projects have hit their targets, while others have missed them by wide margins. One long-running CCS facility in Canada promised to capture 90% of the carbon dioxide from a small coal plant, yet after nearly a decade in operation, it managed to capture just under 60%, according to the Institute for Energy Economics and Financial Analysis.
What the plan says about the AI power race
Exxon’s data center power proposal shows how AI is pulling together industries that previously operated on separate tracks. Technology companies need electricity. Energy companies see new customers. Grid delays create demand for alternatives.
The result is a market where natural gas, carbon capture, nuclear power, solar projects, and large renewable portfolios are all being considered as answers to the same problem: how to power rapidly growing AI infrastructure.
For Exxon, the project is a chance to sell reliability at large scale while presenting carbon capture as part of the package. For data center operators, it could offer another route around the constraints of the grid. But the plan also depends on carbon capture working at a level that has not yet been widely proven in natural gas power plants.
The AI electricity boom is no longer just a technology story. It is becoming an energy infrastructure story, and Exxon’s move shows how high the stakes have become.