Data centers are increasingly becoming an integral part of global economic growth and a fundamental societal need. However, growing demand for data centers is not without its drawbacks.

Fervent demand to advance technological innovation has enabled enormous opportunities for financing the development of the digital infrastructure that powers cloud computing and AI, among other applications. Their growth potential has positioned data centers as an apparent "must have" investment across many public and private equity portfolios.

We forecast global AI-related data center revenues will reach $650 billion by 2028, from less than $200 billion in 2023. Private credit is taking advantage of this opportunity, emerging as a prominent financing source for data centers by providing flexible capital solutions that support the acceleration and advancement in the asset class.

At the same time, data centers' constant energy needs pose significant environmental and energy transition risks. Despite often securing renewable energy supplies and seeking improved efficiencies, data centers could nearly double their emissions by 2030 due to their likely reliance on gas-fired power generation to meet growing baseload demand.

These projected increased carbon emissions are unlikely to pose near-term material credit risk to operators—but could prompt pressure from regulators, investors, and other stakeholders about data centers' environmental effects over the longer term. In our view, the undercurrent of sustainability-related risk and opportunity in data center development and operation is likely to continue to shape the industry's evolution.

As investors continue allocating capital toward financing data center projects, how increasing infrastructure development and energy demands are balanced today may shape tomorrow's credit and economic landscape. Below, S&P Global Ratings' private markets and sustainable finance experts answer questions on the risks and opportunities of data centers as an asset class, alongside assessments of the associated climate and transition risks.

Frequently Asked Questions

What are the requirements for data center development?

The development and financing of new and existing data centers require specific site selection, energy, and funding considerations per asset type. Data center properties differ in size and customer base, including expansive wholesale and hyperscale facilities leased to a few large tenants, as well as retail colocation properties leased to several small tenants.

Data centers can primarily be categorized into retail and wholesale primary asset types, with the potential for differentiating characteristics (including hyperscalers, edge data centers, and carrier hotels).

The increase in the number and capacity of data centers requires ample physical infrastructure, and the location of data centers (similar to other real estate assets) is critical for their valuations. These standard data center asset types generally require regional sites to ensure grid connectivity, transmission and fiber access, water (and climate) for cooling, workforce availability, and cheap energy. We view interconnection as a key competitive advantage for retail data center providers, with control over network dense assets. But after two decades of stagnating power demand, insufficient grid infrastructure—which results in long interconnection waits—will likely be the biggest hurdle to meeting data centers' energy needs. Credit risks differ between hyperscalers and retail/colocation leasing models.

Sustained growth in data center capacity (especially in remote locations) will require significant amounts of uninterrupted baseload power that alone may not be met solely by renewables. Many data center providers have transitioned to solar, wind, biodiesel, and fuel cells against a backdrop of elevated traditional energy prices and tenants' increasing preference for renewable energy sources when available. But the majority of energy utilized by these digital infrastructure assets is still from fossil fuel sources. We believe a combination of data center demand and ongoing energy security concerns will underpin hydrocarbon revenues (and natural gas demand in particular) until at least the end of the decade.

Data center development is contingent on construction, inflation, and financing expenses. Advances in AI, 5G adoption, and cloud services increase the demand for data centers, whose numbers and growth rates are currently highest in the U.S. This is notably due to the country's high concentration of major technology companies, which invest heavily in capacity expansion to support their global operations.

Globally, data centers may be privately or publicly funded—with advanced economies more experienced in building and running data centers but emerging economies becoming a target development area. Each region will have specific funding access across traditional banking lenders, sovereign capital, and private markets. In some markets, sustainability-linked or green bonds could play an important role, notably to fund green initiatives that aim to increase energy efficiency or provide renewable energy solutions and to improve engagement with stakeholders.

As projects multiply and average project sizes increase, constraints across power and water requirements, financing, tenant concentration, and cost inflation will emerge. Navigating these risks are key considerations when assessing data center owners' and developers' credit quality.

What are the implications of data centers' rapid global expansion for the energy transition, water resources, waste, and physical climate risk?

Increasing data center deployment likewise expands energy requirements. The surge in data center power consumption has resulted in a commensurate rise of carbon emissions that could double the technology sector's current annual carbon footprint by the end of this decade if it's not mitigated.

S&P Global Ratings research suggests that at the onset of the data center boom, data centers and their associated networks accounted for roughly 4% of total U.S. electricity demand (or 170 terawatt hours [TWh]), representing about 75 million tons in carbon emissions. Looking ahead, we expect data center power demand in the U.S. will increase even more, by 12% per year through 2030, even after accounting for improvements in efficiency.

According to our research, constraints on renewable generation growth (coupled with data centers' requirement for stable baseload power) could result in natural gas meeting approximately 60% of new energy demand in the U.S. Companies thus face considerable hurdles in lowering their data center emissions.

On-site, low-carbon power for data centers remains niche, even among the most advanced and deep-pocketed companies. At the same time, the use of carbon offsets to lower an operator's footprint faces similar limitations in scale and is not viewed favorably by many market participants. As a result, many of the largest hyperscaler firms have recently dialed back their carbon-reduction goals as computing demand outpaces attempts to curb associated emissions. With the sector's climate ambitions in flux, expected growth in energy and emissions in the medium term could lead to greater scrutiny from regulators, investors, and other stakeholders.

Data centers' carbon emissions are mainly generated by the power demands of their processors and storage systems. The use of this equipment can also have implications for local water resources (considering that a single large facility can consume upwards of 30 million gallons of water annually for cooling) and equipment disposal (since a data center server is typically only functional for five years or less).

Depending on ambient temperatures and technologies used, data centers' cooling systems can be its most energy-intensive component—accounting for 25%-40% of total energy use. In regions already experiencing high water stress, such as parts of the Western U.S. or Southeast Asia, data centers can exacerbate local resource scarcity and face increasing operational risks, including regulatory restrictions and community opposition.

With a typical shelf life of three to five years (due to high utilization rates, intense thermal load, and rapid hardware turnover), the short lifespan of data center servers generates substantial volumes of end-of-life equipment. While some jurisdictions such as the European Union and other regions have implemented regulation to promote electronic equipment recycling, global standards and practices remain inconsistent. Overall, the recyclability of advanced IT components is often limited.

Physical climate risk posed by extreme weather events, including flooding and hurricanes, is heightened for data centers because they are often concentrated in specific regions for efficiency and latency optimization. Such clustering increases exposure to climate-related service disruption and potential asset loss, underscoring the importance of structural resilience planning.

What mitigation strategies are data center operators implementing to address transition risks?

Companies are adopting strategic locations, nuclear energy technologies, and AI innovations, among other approaches, as mitigation strategies. These approaches can reduce climate and environmental-related risk, although the magnitude of the industry's growth will likely require an acceleration in sustainability efforts by operators.

Data centers' requirement for stable energy means that any deployment of traditional renewables needs to be supported by energy storage solutions. Strategically locating data centers near renewable energy sources (including wind, solar, and hydropower) reduces their carbon intensity but can also introduce latency challenges if these centers are located far from end-users or data hubs.

Climate-related physical risks are also shaping site selection strategies, with data centers increasingly located in geographies with lower exposure to extreme weather events, such as in the U.S. Midwest. To address the growing scrutiny over water withdrawals, companies are deploying closed-loop cooling systems that minimize waste and recycling water onsite. Selecting a site in water-abundant regions, such as the Nordics, further reduces exposure to water stress and enhances long-term sustainability.

Data centers can also provide a lifeline to nuclear assets—with new small modular reactors and existing nuclear generation serving as potential solutions for satisfying energy demand. Major tech players like Google, Microsoft, Amazon, and Equinix have announced plans to develop small nuclear reactor capacity, while Microsoft has announced plans to restart Pennsylvania's shuttered Three Mile Island nuclear generating plant.

On the operational side, AI-optimized climate control systems are becoming more prevalent, particularly in high-performance computing environments. There is a growing emphasis on distributed data center architectures, which not only enhance resiliency, but also reduce energy and water consumption by spreading computational loads more evenly and adapting to local resource constraints.

Innovations in computation could also help address the issue. For example, the January 2025 launch of the DeepSeek AI model, which purports to be substantially more energy efficient than American peers, could represent a path to material improvement for the AI ecosystem so long as rebound effects are sufficiently mitigated.

How does S&P Global Ratings analyze data center projects and assess associated environmental risks?

When rating transactions or companies backed by data center assets, S&P Global Ratings uses its criteria to determine which framework best captures associated credit risks. To date, we primarily use our analytical approaches for corporate finance, project finance, and structured finance (mainly methodologies for asset-backed securities [ABS] and commercial mortgage-backed securities [CMBS]). We published our first global data center ABS criteria in June 2024 and a request for comment for our global digital infrastructure corporate ratings methodology on March 10, 2025.

Our ratings approach is largely determined by the structural considerations of the underlying financing vehicle, rather than the asset itself. The transactions and companies we rate provide a holistic view into the array of energy sources that data centers have available to leverage, alongside their comparable costs and relative risks. As demand for investment in this sector expands, we believe that investors may use different structuring and advance new financial innovations. Under such evolution, we would likewise continue adapting to use the best analytical approach for these transactions.

Analyses of sustainability factors can help inform a data center operator's risk profile, whether for an existing facility or a new development.

S&P Global Ratings' Sustainable Finance Second Party Opinions (SPO) and Climate Transition Assessments (CTA) provide additional transparency to investors seeking to understand and act upon an entity's potential contribution to a sustainable future. Both methodologies leverage S&P Global Ratings' sector expertise; understanding of climate risk, environmental science, and social factors; and Shades of Green approach, which assigns qualitative scores to business activities, technologies, and investments according to their potential to deliver sustainable benefits.

An SPO is an independent, point-in-time analysis of a sustainable finance instrument, program, or framework, and the characteristics of the issuing entity that are relevant for their implementation. For data centers, SPOs could, as an example, assess how green debt financing may fund new and more efficient cooling systems that both conserve energy and water, lowering operating costs and reducing the risk of service interruption.

A CTA provides a qualitative opinion on where a company is on its current transition journey and how we expect its transition will evolve, based on an assessment of planned transition activities and implementation drivers. For a data center operator, our analysis would likely center on the likelihood of achieving emissions reduction goals in the face of rising computing demand and the expected impact of investments into emissions reduction efforts. This could include alternative energy technologies, carbon offsetting, and more efficient cooling systems. The additional analysis of key drivers such as regulation, organizational and financial alignment, and metrics and targets contributes to a holistic assessment of an entity's transition.

Writer: Molly Mintz

Related Criteria

• Request for Comment: Request for Comment: Sector-Specific Corporate Methodology: Global Digital Infrastructure, March 10, 2025
• Sector-Specific Corporate Methodology, March 10, 2025
• Rating Methodology And Assumptions For Global CMBS, July 26, 2024
• Data Center Securitizations: Global Methodology And Assumptions, June 13, 2024
• Corporate Methodology, Jan. 7, 2024
• Project Finance Rating Methodology, Dec. 14, 2022
• Sector-Specific Project Finance Rating Methodology, Dec. 14, 2022

Related Research

• Data Centers: Can Infrastructure Developments Keep Up With The Increasing Demand?, Dec. 4, 2024
• Data Centers: U.S. Not-For-Profit Electric Utilities Explore Ways To Mitigate Risks From Load Growth, Nov. 21, 2024
• Data Centers: Rapid Growth Will Test U.S. Tech Sector's Decarbonization Ambitions, Oct. 30, 2024
• Data Centers: Welcome Electricity Growth Will Fall Short Of U.S. Data Center Demand, Oct. 22, 2024
• Data Centers: Surging Demand Will Benefit And Test The U.S. Power Sector, Oct. 22, 2024
• Data Centers: More Gas Will Be Needed To Feed U.S. Growth, Oct. 22, 2024
• Credit FAQ: Can Operators Navigate Pitfalls In Asia-Pacific's Data Center Boom?, Sept. 17, 2024