Skip to Content Skip to Menu Skip to Footer

By Zoe Parker, Tanya Peevey, Morgan Smith, and Nelson Lleno


This is a thought leadership report issued by S&P Global. This report does not constitute a rating action, neither was it discussed by a rating committee.

Highlights

The world needs more energy to meet rising demand from AI-driven data centers. Global generation capacity is expected to grow 173% by 2050 under the S&P Global Energy Integrated House View base case scenario.

At the same time, power generation assets are increasingly exposed to climate physical hazards. By the 2050s, 87% of existing operating capacity is projected to be exposed to at least one such hazard, according to projections using the S&P Global Sustainable1 Climate Physical Risk dataset. (All projections in this analysis use the medium-high emissions scenario, SSP3-7.0, which corresponds to around 3.6 degrees C of warming by 2100.)

In the US, rapid data center load growth is placing increasing demands on power systems that are already vulnerable to climate physical risks, particularly in the Southwest. Roughly 60% of US power plants are currently exposed to at least one climate physical hazard. Drought exposure alone could impact power plants representing 36% of the nation’s generating capacity by the 2050s, an increase of 32 percentage points over four decades.

In Indonesia, as in many emerging markets, rapid economic growth requires significant capacity build-out that must also contend with high physical risk exposure. By the 2050s, all existing and planned power plant locations are projected to be exposed to at least one climate hazard.

While each country has unique power needs, many share exposure to drought risk that threatens hydroelectric generation and thermal power plants that rely on rivers for cooling and transportation of coal, potentially constraining economic growth and increasing the need for reliable alternative power sources.

Building climate resilience is often framed as a defensive challenge. Increasingly, it is also a competitive one. In an era of rising electricity demand, the power systems best able to withstand climate shocks will also be those best-positioned to support economic growth, attract new industries, and strengthen national energy security.

Power generation and grid reliability have entered everyday public discourse as consumers face higher electricity bills, and many countries grapple with a surge in power demand. In some markets, demand growth is driven by rapid AI data center development; in others, electrification and rising cooling needs are the primary drivers. This marks a departure from an era of steadier load growth, when supply-side factors primarily shaped energy markets.

Prolonged extreme weather events are adding a further layer of stress to power systems. The relationship between climate resilience and power generation is most evident during the summer months, when higher temperatures simultaneously increase cooling-related electricity consumption and reduce the performance of key generation assets. As summer peak demand continues to grow in both developed and emerging economies, maintaining reliable generation capacity during periods of extreme weather is becoming an increasingly important component of energy security.

These pressures come as policymakers and utilities balance competing priorities: ensuring reliable and affordable electricity supplies, advancing decarbonization goals and supporting new sources of demand. Because power generation assets operate for decades, investment decisions made today will shape system resilience far into the future, even as the pace and trajectory of future demand growth remain uncertain. As a result, governments are increasingly challenged to build power systems that remain reliable under worsening climate conditions and support strategic competitiveness in an era of significant geopolitical, geoeconomic and environmental change.

Power plant exposure to climate hazards is increasing in all regions

Power plants are long-lived assets that can operate for many decades. Increasingly, governments and utilities are extending asset lifetimes by deferring planned retirements or repowering plants to accommodate growing electricity demand. Over these extended horizons, climate hazards are projected to become more severe and frequent in many regions, increasing operational risk, particularly for facilities designed around historical climate conditions.

Climate physical risks can impact power plant performance. For example, rising water temperatures and low water levels due to drought can limit output at nuclear and thermal power plants that rely on freshwater resources for cooling. In severe cases, these conditions can force temporary shutdowns, as occurred at several plants in Hungary and Romania in August 2026 due to record-low water levels on the Danube River.

Physical risk exposure varies by plant location, technology, fuel type and design. Table 1 summarizes the material climate hazards examined in this report and their potential impacts on different generation types.

Table 1

This analysis covers global power plants that are planned or operating. We use the S&P Global Sustainable1 Climate Physical Risk dataset to assess generation capacity exposure to the hazards listed in Table 1. Because information on asset-specific adaptation measures is limited, this analysis excludes existing or planned interventions. This isolates changes in asset exposure over time and assesses potential impacts in the absence of adaptation.    

A plant location is considered materially exposed if its projected hazard values exceed the thresholds in Appendix Table 2 under a medium-high emissions scenario (SSP3-7.0, which corresponds to around 3.6 degrees C of warming by 2100). Because the climate hazard data does not account for future asset development, increases in exposure reflect changes in climate conditions rather than the addition of new power plants.

The share of capacity exposed to at least one climate hazard increases in all regions between the 2020s and 2050s (Figure 1). Globally, the proportion of capacity exposed increases from 73% in the 2020s to 87% in the 2050s.

Figure 1

More than 95% of operating capacity in the Middle East, Africa, Asia-Pacific and Latin America and the Caribbean is projected to be exposed to at least one hazard in the 2050s, up from roughly 83% in the 2020s.  Europe experiences a much larger increase from 44% in the 2020s to 72% in the 2050s.

Overall, the outlook varies across generation technologies. Nuclear generation sees the greatest increase in proportion of capacity exposed, increasing from 65% in the 2020s to 89% in the 2050s. The proportion of hydropower capacity exposed increases from 68% in the 2020s to 85% in the 2050s, due to factors such as increasing drought and water stress exposure. For both coal and natural gas, the proportion of capacity exposed is over 90% (96% and 91%, respectively) in the 2050s, up from 82% and 78%. Wind and solar generation have the lowest proportion of capacity exposed in the 2050s, with 62% and 79%, respectively, up from 48% and 71% in the 2020s.

Capacity build-out is increasing

The increasing exposure of existing power plant locations is only part of the picture. Generation capacity is projected to expand significantly over the coming decades, including in regions with elevated exposure to climate shocks, such as Asia-Pacific.

S&P Global Energy projects that global installed capacity across technologies will increase by 173%, from around 11 terawatts in 2025 to nearly 30 TW in 2050. [Installed Capacity projections – Base Case. Energy and Climate Scenarios, July 2026, S&P Global Energy.] Over the long term, nearly all capacity growth is projected to come from renewables, although short-term trends may differ. All regions globally are projected to see an increase in capacity, but drivers differ by region and level of economic development. While data center growth and energy security concerns are key drivers in many markets, emerging economies must also meet the needs of growing populations and advance universal energy access.

Figure 2

Case studies illustrate the importance of power generation resilience

We examine two case studies that show different relationships between power generation, climate exposure and economic resilience: the US and Indonesia. In the US, data center development is driving a 2.6% annual increase in electricity demand, a pace not seen in the past two decades. In Indonesia, significant GDP and population growth have fueled annual electricity demand growth of 4%-5%. The US has a mature electricity system supported by extensive grid infrastructure and a diversified generation mix. Indonesia, by contrast, remains more reliant on coal generation, and its aging grid infrastructure is more vulnerable to extreme weather events. But in both countries, worsening climate hazards threaten the reliable generation capacity needed to power future growth and security.

US: Data center race and power system risk

In the US, rapid data center development has both challenged and increased the importance of power system resilience. Data center growth is driving new generation investment and plant life extensions. Climate hazards are particularly relevant in these regions because rising electricity demand can amplify existing stresses on power and water infrastructure.

About 60% of US operating and planned generation capacity is exposed to at least one climate hazard. Hazard exposure is generally consistent across technologies but varies greatly by region. Power plants are more likely to be exposed to multiple hazards along the coasts and in the western half of the US, where wildfire and drought hazards are concentrated. 

Figure 3

Approximately 20% of total US generating capacity, or 307 GW, is currently exposed to water stress, wildfire, drought or a combination of these hazards. Drought exposure rises most sharply: The share of US generating capacity exposed to drought is projected to increase from 4% today to 36% by 2050. Nearly half of the exposed capacity consists of thermal and nuclear power plants that rely on water for cooling, making them particularly vulnerable to water-related constraints.

River flooding is a nationwide hazard, although exposure to a 1-in-100-year flood event is highest in the Northeast and Southeast. Approximately 198 GW — about 13% of US generating capacity — is exposed to flood-related risk.  Tropical cyclones pose the most widespread climate hazard to US power generation, with approximately 384 GW of capacity exposed to Category 3 or stronger storms each year. (Hurricane intensity is labeled on a range from Category 1 on the low end up to Category 5 on the Saffir-Simpson Hurricane Wind Scale, and major hurricanes are defined as Category 3 and above.) Exposure is concentrated along the Gulf Coast and Atlantic seaboard, declining farther inland and toward Canada. Among vulnerable assets are roughly 51 GW of wind generation capacity that could experience curtailments due to extreme wind speeds.

Extreme heat currently does not pose a significant risk, using a threshold equivalent to three months of severe heat exposure. By 2050, however, 74 power plants in Florida, representing nearly 15 GW of capacity, could be exposed to this hazard.

Regional focus

Two US power grids, the PJM Interconnection, covering the mid-Atlantic and portions of the US Midwest, and the Electric Reliability Council of Texas, covering most of Texas, account for more than half of the nation’s existing data center capacity and 60% of the planned data center growth over the next decade. However, their power plants face differing climate risk trajectories.

Southwest

US power plants face high climate physical risk in the Southwest, where  wildfire, drought and water stress exposure are among the highest in the country. At the same time, Arizona and Texas have emerged as two of the country's largest data center markets. Arizona's approximately 3.7 GW of data center demand is concentrated around Phoenix, while Texas supports roughly 6.4 GW across major markets including Dallas, Austin, Houston and western Texas. Secondary hubs across the Southwest, including Las Vegas, Albuquerque and southern California, are also expanding rapidly.

Figure 4

Data center demand is the primary driver of load growth across the Southwest. In Arizona, the projected 40 TWh of data center electricity consumption over the next decade is equivalent to roughly one-third of the state’s current power demand. Despite this rapid load growth, robust additions of renewables, battery storage and natural gas generation in the Southwest have helped maintain an adequate supply outlook. However, rising climate physical risks could challenge the resilience of the power systems supporting this data center growth.

Plants in locations where current drought risk exceeds the hazard threshold (three months of high drought likelihood days) are concentrated in southern Nevada, southern California and western Arizona. By 2050, however, drought exposure is projected to expand across Arizona, New Mexico and western Texas. This risk is further compounded by severe water stress in the Southwest, a metric that accounts for local competition for water resources. While power generation has historically received priority access to water supplies, worsening scarcity could force difficult trade-offs among the power, agricultural and municipal sectors.

Power plants in the Southwest are also highly exposed to wildfire risk, most evident in California, Arizona and western Texas. Wildfires can directly damage generation and transmission infrastructure, but more frequently disrupt power systems by necessitating preventive de-energization and power plant shutdowns. The 2024 Smokehouse Creek fire in Texas disrupted power for thousands of people due to both preemptive  power line shutoffs and direct grid damage. From 2000 to 2016, wildfire damage to California’s transmission and distribution systems cost more than $700 million. Additionally, smoke from wildfires can suppress solar output hundreds of miles beyond the fire itself. During California’s 2020 wildfire season, researchers found that smoke reduced solar power generation by approximately 10% to 30% during peak production hours.

Wildfires, drought and water stress share common drivers and reinforce one another’s impacts, creating a multidimensional challenge for power system resilience in the Southwest.

Mid-Atlantic

The mid-Atlantic is home to the world’s largest concentration of data center capacity. With 15.2 GW of existing data center power demand, northern Virginia remains the dominant market by a wide margin. Considerable data center development has also taken place in Ohio (4.6 GW) and Illinois (3.4 GW), with secondary markets in New Jersey and Pennsylvania adding to power demand across the region.

Figure 5

Compared with the Southwest, power plants in the mid-Atlantic are exposed to a narrower range of climate physical risks. The frequency and severity of hazards such as extreme heat, chronic drought and wildfire remain substantially lower than in the Southwest and parts of the Northwest and Southeast. However, flooding remains a significant vulnerability, especially for facilities situated near major rivers and along the Atlantic coastline. The region is also regularly affected by severe storms, which can directly damage generation assets and disrupt transmission networks. Hurricane Sandy in 2012 highlighted these risks when storm surge prompted an emergency alert at New Jersey's Oyster Creek Nuclear Generating Station as rising waters threatened key cooling-water systems.

Across the US, the Ohio Valley and Central Appalachia stand out as among the most climate-resilient regions. The region benefits from significant freshwater resources, low exposure to hurricanes and coastal flooding, and less water stress than many fast-growing power markets. While localized flooding along the Ohio River and other waterways remains a concern, this region is generally insulated from many of the climate hazards that threaten generation assets and electric infrastructure elsewhere in the country. These characteristics have made the region increasingly attractive to both power-intensive industries and large-scale data center developers. Ohio is currently home to one of the largest data center build-outs in the country and ranks among the top five states for data center power demand, while eastern Pennsylvania and Indiana also have significant data center projects planned for the coming decade.

Indonesia: Development-driven energy expansion

Like much of the Asia-Pacific region, Indonesia is projected to increase its generation capacity significantly over the coming decades. S&P Global projects installed capacity will more than triple, from less than 100 GW in 2025 to over 300 GW in 2050 [Installed Capacity projections – Base Case. Energy and Climate Scenarios, July 2026, S&P Global Energy]. Rapid economic expansion, urbanization and industrialization are driving demand for electricity. Electricity demand rose from about 90 TWh in 2000 to about 360 TWh in 2025 [Power Demand – Base Case. Energy and Climate Scenarios, July 2026, S&P Global Energy], nearly quadrupling to keep pace with Indonesia’s rapid economic growth. Data center development is a secondary driver, with current and planned sites primarily concentrated around the country’s largest city, Jakarta.

Indonesia is highly exposed to climate hazards, which are already affecting its energy system through weather-related blackouts, reduced hydropower output during droughts, and rising cooling demand. Aging grid infrastructure further increases vulnerability to these hazards, a challenge shared by many developing economies.

By the 2050s, all of Indonesia’s power plants are exposed to at least one hazard, and 95% are exposed to multiple hazards, increasing the risk of compounding and cascading impacts (Figure 6).

Figure 6

Extreme heat is by far the most common hazard, with 97% of power plants exposed to three months or more of extreme heat days by the 2050s. This could reduce the efficiency and output of coal plants, which provide most of Indonesia’s electricity. Further, hazard exposure varies by region: power plants in the south of the country, mainly on the island of Java, are more likely to experience tropical cyclones than in other regions, while those in the north are more exposed to pluvial flooding. Storms and flooding can cause disruption regardless of plant type. Drought exposure is not common, with only 4% of power plants exposed to three months or more of drought conditions by the 2050s. However, this figure includes hydropower plants, which have been impacted by drought historically, and are vulnerable in the context of a strengthening El Niño.  

More than half of Indonesia's population lives on Java, including over 40 million people in the Jakarta metropolitan area. Rapid urbanization has required substantial power network expansion, and growing data center development around Jakarta may place additional strain on the system. These factors, alongside rising sea levels and subsidence (the downward movement of ground, often related to water extraction), have driven Indonesia to move its capital to the less-exposed island of Borneo.

Power plants located near Jakarta face increasing exposure to extreme heat and flooding, as well as existing exposure to tropical storms. Jakarta is one of the fastest-sinking cities globally, and much of the city sits below sea level, which has severe implications during flooding and storm surges. Increased cooling demand due to extreme heat is already affecting power demand on the island — Indonesia’s National Adaptation Plan (NAP) states that heatwaves in 2023-2024 resulted in an increase in electricity consumption in Java of about 28%.

Indonesia includes the adaptation of its energy system and the linkages to energy security as a key component of its NAP. This includes measures to improve power plant and grid resilience, alongside investments that support emissions reduction goals, such as the build-out of renewable energy infrastructure including decentralized micro solar and hydro grids. Crucially, the Indonesian government considers energy system resilience to be a key component of broader systemic security within the country. Adaptation, including of the energy system, is viewed as a key enabler of Indonesia’s development goals.

Overview of other global regions

Many other developed and developing countries around the globe face similar demand drivers and the need for reliable power supply. Canada, like the US, has significant projected load growth from data center development in Alberta and Ontario, and benefits from abundant domestic energy resources that reduce its exposure to fuel supply disruptions. However, its heavy reliance on hydropower leaves portions of the power sector vulnerable to drought.

Major Latin American markets such as Brazil and Colombia also rely heavily on hydropower and face similar drought-related risks. Brazil’s average water inflows over the last decade fell 16.2% below the long-term historical average (1931–2025), indicating a persistent decline in water availability. [Source: S&P Global Energy; ONS.] The region has not yet experienced significant data center growth but is looking to attract hyperscalers as US data center markets saturate. Climate exposure could impede these aspirations and the region's economic development.

Figure 7

In many developing economies, including those in Latin America and Asia, electricity demand growth is closely tied to rising incomes and living standards. Sub-Saharan Africa faces a distinct challenge: fragmented and often unreliable grid infrastructure has increased reliance on behind-the-meter generation as transmission constraints limit access to utility-scale power.

China and the Gulf Cooperation Council countries are relatively self-sufficient in the materials and resources needed to expand power infrastructure. However, the Asia-Pacific region also has one of the highest proportions of power plants exposed to climate hazards, facing climate physical risks, including extreme heat and drought as demonstrated by the year-over-year drop in June 2026 of hydroelectric generation in Southeast Asia by an average of 13 GW. [Sources: S&P Global Energy; 10 Japanese transmission and distribution utilities; Korea Power Exchange; Taiwan Power Co.; Power Grid Bangladesh PLC; Grid Controller of India Ltd.; Malaysia Grid System Operator; Independent Electricity Market Operator of the Philippines; Vietnam Electricity.]

In Europe, data center growth is contributing to demand increases, but not at the same rate as in the US. After widespread heatwaves in 2026, focus has turned toward projected demand growth due to air conditioning, including in historically cooler locations like the UK. The EU has ambitious renewables targets, though the continent remains reliant on fossil fuel imports, and record low river levels due to droughts in 2026 have demonstrated the vulnerability of thermal generation that relies on river water for cooling.

Looking forward

Though the drivers may vary — from data center build-out and electrification in developed markets to urbanization, population growth and rising incomes in emerging economies — expanding reliable power generation capacity is a common priority worldwide. The increasing frequency and severity of climate hazards can disrupt power supply through reduced plant output and efficiency, as well as full shutdowns from acute events. Improving the resilience of power plant assets to climate physical risk is one pillar of a holistic approach to enhancing countries’ energy security and building a buffer against shocks. Countries that successfully expand generation capacity while strengthening resilience to climate hazards will be better-positioned to support economic growth, attract investment and enhance long-term energy security.

Contributor: Rick Lord

Appendix: Data and approach

The S&P Global Sustainable1 Physical Risk dataset contains climate hazard exposure data at the asset level for approximately 96,000 operating and planned power plant assets across different sectors. The dataset includes 10 climate hazards: extreme heat, extreme cold, coastal flooding, pluvial (rainfall-related) flooding, fluvial (river-related) flooding, drought, water stress, tropical cyclones, wildfire and landslides. The dataset includes decadal average projections (2020s-2090s) for these hazards using the four shared socioeconomic pathways (SSPs) included in the UN Intergovernmental Panel on Climate Change’s Sixth Assessment report. This research focuses on SSP3-7.0, which corresponds to a moderate-high emissions scenario with projected global temperature increase of 2.1 degrees C (1.7 C-2.6 C) by 2050 or 3.6 C (2.8 C-4.6 C) by 2100.

We use hazard data for coordinate locations of power generation assets and focus on eight hazards (excluding extreme cold and landslide) as most material to power plants. There have been instances of disruption to power generation due to extreme cold events, but these are not represented in the future projections due to the decadal average (rather than extreme event) approach — extreme cold days decrease globally over the decades.

The table below lists the thresholds applied in the analysis as a proxy for exposure to climate hazards. These thresholds are consistent with previous S&P Global research into subnational and national exposure (see “Sustainability Insights: Hot Spots: Subnational Regions Outside The U.S. Face Rising Physical Climate Risks”, Nov. 12, 2024) and represent a threshold of higher exposure for each hazard.

For each hazard, a different metric is used with a value between 0 and 1. The table describes what each of these metrics is for the different hazards, gives the numerical value, and explains what value of the metric this corresponds to. In our methodology, we then count the number of hazards to which a location is exposed using these thresholds in the 2020s and the 2050s. For example, if a location has an extreme heat value of 0.3, a water stress value of 0.5, and other hazard values below their respective thresholds, it is considered exposed to two hazards.

Table 2