Solving the US Data Center Power Crunch

The rise of AI is fueling a boom in data center investment and creating significant demand for the 24/7 reliable firm power needed to run them. However, with generating capacity unable to keep pace with this growing power demand, we expect a potential 80-gigawatt capacity shortfall to emerge in the US by 2030. (See Exhibit 1.)

Companies are busily searching for power solutions that can meet the substantial energy needs of their data centers. Any solution will need to fulfill several criteria. Most importantly, it must be available quickly to begin closing this looming shortfall. But an ideal power solution must also be cost-effective, scalable, proven, have regulatory support, and enable tech players to deliver on their low-carbon commitments.

While they currently rely on a variety of sources, including unabated gas-fired generation, data center developers are exploring multiple options for securing firm, low-carbon power, including geothermal, nuclear, and intermittent renewables paired with battery storage. Although these all have potential, each technology has drawbacks in one or more key areas. Based on our assessment, gas-fired generation fitted with carbon capture, utilization, and storage (CCUS) technology is currently one of the best-placed to meet these criteria at scale through 2030, primarily in the US.

If Big Tech were to embrace this solution for powering data centers, the impact on the CCUS industry could be huge. We estimate it could result in up to $80 billion of capex spending on retrofits and new gas plants with CCUS equipment, while creating cost benefits for other carbon capture applications. As gas and power with CCUS gain momentum in the US data center market, developers, operators, and owners may benefit from evaluating its potential role within their long-term power strategies. However, they will need to act quickly to qualify for the 45Q tax credit, which incentivizes US companies that permanently store carbon dioxide (CO₂) or use it in enhanced oil recovery.

How Gas with CCUS Stacks Up Against Other Solutions

We examined eight technologies to gauge their potential for meeting data center power requirements in the US market. (See Exhibit 2.) Of these, six have demonstrated in a real-world setting that they can provide round-the-clock power.

When examined more closely, many of these technologies have drawbacks that limit their ability to meet the 2030 capacity shortfall. For example, regulatory support for intermittent renewable generation with battery storage varies from one US state to another, creating headwinds in some regions that can delay projects and complicate their construction. In addition, the scalability of this technology depends heavily on whether regions have plentiful sun or wind.

While geothermal is a proven low-carbon technology with strong backing from regulators, its economics currently require wells to be located close to hot spots, making scalability difficult. Similarly, large-scale nuclear projects are beset by high costs, long construction times, and complex regulatory processes.

Across the six key criteria prioritized by data center developers in the US, gas with CCUS is a technology-ready solution that performs well in each area.

Speed

Behind-the-meter (BTM) gas plants, which are built to serve only a data center, circumvent the grid, and are fitted with CCUS, can deliver power in as little as three years after breaking ground; however, this comes at the price of making power supply dependent on a single asset. Securing abated gas delivered over the grid—typically via a power purchase agreement with a local utility, and where existing turbines are retrofitted with CCUS—takes longer because of grid connection queues that add a minimum of two to three years. Nevertheless, this option still scores ahead of competing technologies on other criteria.

Proof Points

There are multiple examples worldwide where CCUS has been applied to gas-fired power plants, such as the UK’s Net Zero Teesside project.

Regulatory Support

In the US, CCUS enjoys a degree of bipartisan backing among state-level policymakers and regulators that is higher than many other technologies. Such support can help ease and accelerate the regulatory approvals that often hold back large-scale developments. And while the political landscape can change, CCUS currently benefits from 45Q tax credits—which have continued to receive bilateral support, with a key timing stipulation that requires the assets to be under advanced development before January 1, 2033.

Cost

With electricity accounting for as much as 80% of data center operating expenses, identifying a cost-competitive energy solution is essential. While a new, unabated combined cycle gas turbine (CCGT) plant has a levelized cost of energy (LCOE) of $45 to $60 per megawatt-hour (MWh), a CCGT plant with CCUS compares favorably with this and other options, such as hydrogen and nuclear, due to its relatively low capex requirements and fuel costs. We estimate that, once tax credits are taken into account, the LCOE of a CCGT plant retrofitted with CCUS is $65 to $80 per MWh, while that of a new CCGT plant with CCUS can be less than $120. Furthermore, as greater capacity is deployed, experience curve effects will over time narrow the cost differential between unabated and abated gas. (See Exhibit 3).

Scalability

There are several reasons why gas with CCUS is more scalable than other solutions. While they are unabated, gas-fired plants are already the main form of power generation in the US. CCUS itself has specific scalability advantages. Abundant storage potential for captured CO₂ exists in key regions, such as Texas and the Midwest. Furthermore, a front-of-the-meter CCGT plant with CCUS (which relies on the grid to transport electricity) theoretically does not need to be near the data center it serves as long as it is located in the same wholesale electricity market.

Low-Carbon Credentials

Although the lifecycle carbon intensity of a CCGT plant with CCUS is higher than that of geothermal and legacy nuclear plants, it significantly outperforms an unabated CCGT plant. We estimate that CCUS technology can capture over 90% of a CCGT plant’s emissions and reduce its lifecycle carbon emissions by over 70% on a per-kilowatt-hour basis. This makes it a good choice for hyperscalers, such as Google and Amazon, which have committed to making 50% to 100% of their energy usage carbon-free by 2030.

A Supportive Landscape Emerges for CCUS Technology

Multiple factors are creating momentum behind the use of CCUS and gas in the US. The mechanics of CCUS technology require that CO₂ captured from stationary emitters is either transported to aquifers, where the gas can be sequestered, or to business customers where it can be used in industrial processes such as enhanced oil recovery.

In addition to having a network of viable CO₂ transportation pipelines, the US has substantial storage capacity. Its geological carbon storage potential is estimated to be about 6 gigagtons, which is equivalent to 20 times the current pipeline for CCUS projects. While storage does not exist throughout the US, there are significant volumes available in Texas, which is a leader in the technology, as well as other regions that are newer to CCUS, such as the Midwest, the West Coast, the Mountain West, and parts of the Northeast.

A growing number of players are eyeing the business opportunities presented by data centers and CCUS. Oil and gas companies have used CO₂ for decades to increase the amount of oil extracted from a reservoir. This makes them natural customers for CCUS. With data center demand for clean power growing, several oil majors have ventured into the gas with CCUS market. Both ExxonMobil and Chevron are developing plans to build BTM gas-fired power plants with CCUS that are dedicated to serving US data centers. Meanwhile, Meta has an agreement with Louisiana-based utility Entergy to power a $10 billion data center using gas plants with the potential to incorporate CCUS technology. Google is partnering with developer Low Carbon Infrastructure on power projects involving CCUS, including the Broadwing Energy plant in Decatur, Illinois.

CCUS has been steadily gaining momentum in the US, with the technology moving into new areas including steel, cement, and chemicals in recent years. But if demand from hyperscalers seeking a clean power solution for their data centers takes off, it would mean a huge boost for the technology, create valuable capacity-driven cost benefits, and likely fuel an expansion outside the US. (See “The International Potential of CCUS with Gas.”) For the four biggest US hyperscalers—Google, Meta, Amazon, and Microsoft—to achieve their 2030 clean power commitments using just CCGT plants with CCUS would require $70 to $80 billion of capex (in retrofits and new build costs) and involve over 105 Mtpa of CO₂ sequestration, according to our estimates.

Four Actions for Data Centers to Advance the CCUS Solution

Gas-fired generation with CCUS is on the cusp of being a valuable, low-carbon power solution for US data centers. As data center owners and developers start to realize the benefits of this approach and build it into their plans for the future, they should take the following steps:

Frame gas with CCUS as part of the journey.
Hyperscalers should start by determining how much firm power they will require to support their growth plans. They will then need to assess how trade-offs over costs, sustainability, and scalability will shape the role of gas with CCUS among other solutions. Data center developers should also establish a clear position on the role of gas with CCUS in their portfolios.

Create an asset-level plan
Develop an implementation strategy that balances the near-term need for energy against longer-term goals, such as the push towards decarbonization. Meanwhile, developers can unlock financing and enhance customer trust by building credibility and transparency around key proof points—including capture rates, lifecycle emissions, and storage integrity.

Proactively engage with stakeholders
Communicate a clear narrative to regulators, investors, and customers about how gas with CCUS can complement the role of renewables in achieving net zero. For their part, developers should position projects to capture incentives while building resilience if support schemes change.

Prioritize scalability
Choose partners and approaches that can replicate gas with CCUS quickly across multiple sites and regions rather than simply providing bespoke projects. Similarly, developers should design projects for repeatability, using modular builds, pre-negotiated CO₂ transportation and storage contracts, and replicable partnerships.

Data centers need firm, low-carbon power now—not in the next decade—if we are to meet the increasing demand for AI in an environmentally sustainable way and avoid a growing generating capacity shortfall. Gas with carbon capture, utilization, and storage is the leading scalable option that can deliver this goal. Developers and companies that want to make this transition should start by examining how the solution could work for them based on their speed, cost, and emissions priorities.