Lisa Thalheimer

Hyperscale data centers built to train and run artificial intelligence (AI) systems are driving one of the fastest sustained increases in US electricity demand in decades. The International Energy Agency (IEA) projects global data center electricity consumption to roughly double by 2030, with AI-optimized facilities driving most of that growth and data centers accounting for nearly half of US electricity demand growth over the same period (IEA, Energy and AI, 2025). Whether this demand growth is already raising the prices households and businesses pay is contested. Recent causal analyses reach opposite conclusions depending on geographic scope and time horizon, with one finding a modest local price increase concentrated in industrial rates (Rana et al., 2026) and another finding a modest decline in the national average retail rate (Watten et al., 2026). Beyond price, a parallel literature has begun documenting environmental and health concerns tied to facility siting, including air pollution from fossil-fuel generation and backup diesel generators, water use for cooling, and noise. Virginia's dense data center corridor has served as an early case study of these concerns (Gour, 2026), and facility-level modeling shows that how a campus sources its power materially shifts who bears the resulting air-quality burden (Gujral & Dominici, 2026). Here we ask four research questions. (1) Do hyperscale data centers raise the retail electricity prices paid by industrial and residential customers of the utilities that host them? (2) Does the size of a utility's data center exposure, beyond its presence, predict a larger price effect? (3) Does the type of developer, large hyperscalers building their own campuses versus other developers, change the size of any effect? (4) Does a utility's price respond to data center capacity sited in neighboring utility territories, not just its own? A companion analysis asks a fifth, related question at the county level. (5) Do host counties show any measurable local economic benefit, such as higher employment or income, that might offset these costs?

The devastating 2022 summer flood in Pakistan displaced about 7 million people in the Sindh province alone. Up to one‐third of the country's area, mostly the country's south, was flooded. Effective response to intensifying and compounding hazards requires a better understanding of these processes. We can gain insights if impact assessments include socio‐economic components and uncertainties arising from the interactions between impacts. However, the quantitative evidence from impact assessments remains limited and fragmented, due to methodological challenges and data limitations. Using the open‐source impact assessment platform CLIMADA, we study to what extent flood‐related hazards can be used to quantify displacement outcomes in a data‐limited region. Using flood depth, exposed population, and impact functions, we link flood vulnerability to displaced people. This allows us to estimate internal displacement resulting from the flood event, and to further assess how displacement varies across the region. We find that a flood depth threshold of 0.67 m, with a confidence interval (CI) from 0.35 to 1.10 m, provides a best fit to all data from Sindh province. We find a negative correlation between displacement and the degree of urbanization. By testing the performance of our model in explaining differing displacement estimates reported across Pakistan, we show the limitations of existing impact assessment frameworks. We emphasize the importance of estimating potential displacement alongside other impacts to better characterize, communicate, and ultimately mitigate the impacts of flooding hazards.

The devastating 2022 summer flood in Pakistan displaced about 7 million people in the Sindh province alone. Up to one-third of the country’s area, mostly the country’s south, was flooded. Effective response to intensifying and compounding hazards requires a better understanding of these processes. We can gain insights if impact assessments include socio-economic components and uncertainties arising from the interactions between impacts. However, the quantitative evidence from impact assessments remains limited and fragmented, due to methodological challenges and data limitations. Using the open-source impact assessment platform CLIMADA, we study to what extent flood-related hazards can be used to quantify displacement outcomes in a data-limited region. Using flood depth, exposed population, and impact functions, we link flood vulnerability to displaced people. This allows us to estimate internal displacement resulting from the flood event, and to further assess how displacement varies across the region. We find that a flood depth threshold of 0.67 m, with a confidence interval (CI) from 0.35 m to 1.10 m, provides a best fit to all data from Sindh province. We find a negative correlation between displacement and the degree of urbanisation. By testing the performance of our model in explaining differing displacement estimates reported across Pakistan, we show the limitations of existing impact assessment frameworks. We emphasise the importance of estimating potential displacement alongside other impacts to better characterise, communicate, and ultimately mitigate the impacts consequences of flooding hazards.

Globally, populations are increasingly located in areas at high risk of climate change impacts. Some populations lack the agency to move out of harm’s way, leading to involuntary immobility. The climate risks these populations face are insufficiently addressed in climate policy and disaster planning. While policy and planning should be data-informed, the lack of appropriate data should not limit governments and institutions from taking action to reduce the risk of involuntary immobility. Incorporating involuntary immobility within the broader sustainable development goals of climate action and safe, orderly, and regular migration may substantially reduce the risk of involuntary immobility.

The efficacy of early warning systems in saving lives and reducing other losses and damages is widely recognized. However, these systems often lack information about the potential impacts on people, assets, and systems. Impact-based early warnings that consider information on exposure and vulnerabilities could fill this gap, enabling a more effective public response and preparedness - notably for vulnerable groups often disproportionately affected by climate extremes. The increasing cost of climate extremes and the focus on non-economic losses and damages from climate change underpins the need to advance risk knowledge, notably integrating vulnerability and exposure information into existing EWS. The UNDRR-funded EarlyWarning4IGAD project addresses this gap by supporting countries in the Greater Horn of Africa to transition from existing hazard-based to impact-based early warning systems. Building on desk study, expert interviews, and stakeholder consultations, we present a novel approach for impact-based early warning for floods and droughts in the Greater Horn of Africa by integrating data and information on exposure and vulnerability into existing hazard-based systems. Thereby, one particular element is the co-creation of conceptual risk models with and for different vulnerable groups, such as (i) small-scale farmers for crop losses, (ii) vulnerable segments of society for harm to people due to floods, or, more specifically focusing on (iii) women and girls, (iv) people with disabilities, and (v) people in camp settings. In doing so, we show how such risk knowledge can be used to inform impact-based early warnings and ultimately integrated into existing operational flood and drought early warning systems at the regional level, as well as where remaining challenges for operationalization lie.

Drylands cover a significant portion of the Earth’s surface and support a large share of the global population. They are particularly vulnerable to climate change due to low and highly variable precipitation, changes in precipitation patterns, drought, and desertification. Historically, human mobility has been a critical adaptive response in drylands, supporting communities to cope with climatic and environmental shocks. We use a systematic review of the case study literature in the Web of Science (1976–2023) to assess the effects of climatic stressors on human (im)mobility in drylands. The keywords we use capture various forms of mobility and immobility (migration, displacement, pastoralism, relocation, and immobility) and climatic stressors (rapid-onset, slow-onset, and climate variability) in the world’s 107 countries that have at least 1% of territory classified as drylands. More than nine in ten (91%) of the papers identified climatic stressors as influencing human mobility and immobility, with 76% finding an increase in mobility and involuntary immobility, compared to 15% reporting a decrease in mobility. Migration is the predominant type of mobility assessed by the literature, with droughts and anomalies in precipitation and temperature being the primary climatic stressors. Rapid-onset stressors are more frequently found to increase displacement than slow-onset stressors or climate variability. Climatic stressors tend to have a stronger impact on human mobility in drylands compared to global analyses, which include non-dryland regions. We also show a deficit of empirical studies on pastoralism and immobility, limited research on the effects of compound events on human (im)mobility, and a disproportionate focus of studies on China, the USA, and Mexico, leaving African and South American dryland regions understudied.

Globally, populations are increasingly located in areas at high risk of frequent, extreme weather events. Some exposed populations have the ability to move to safer places; others are unable to get out of harm’s way. The climate risks facing these involuntary immobile populations are not often addressed by local and national authorities, despite increasing recognition by international development agencies and humanitarian actors. Here we discuss when and how climate and extreme weather events lead to involuntary immobility by considering the influence of political, socioeconomic, and environmental factors. Addressing barriers in policy and disaster planning, early warning systems and anticipatory action could be tailored to support involuntarily immobile communities. While policy and planning should be data-informed, lack of appropriate data quality should not limit governments and institutions from taking action. Immobility needs to be aligned with the broader sustainable development objectives which entail climate justice and orderly migration.

The science of event attribution is relatively nascent. While its body of knowledge is growing fast, numerous gaps remain, including on the appropriate statistical methods and the proper consideration of various degrees of vulnerabilities and exposure. We show that attributing climate change to internal displacement requires a focus on the interplay between climate and non-climate drivers of impacts via a feasibility study of the 2020 floods in Somalia. Using detailed internal displacement flow data, we describe the socio-spatial characteristics associated with the flooding event. We show that climate change has a limited role in revealing displacement impacts and magnitude following the extreme event.

The devastating 2022 summer flood in Pakistan displaced about 7 million people in the Sindh province alone. Up to one third of the country’s area, mostly the country’s south, was flooded. Effective response to intensifying and compounding climate change hazards requires impact assessments to include socio-economic components, as well as uncertainties arising from the dynamic interactions between impacts. Such quantitative evidence largely remains limited and fragmented, due to methodological challenges and data limitations. Using the open-source impact assessment platform CLIMADA, we study to what extent flood-related hazards can be used to quantify displacement outcomes in a data-limited region. Using flood depths, exposed population, and impact functions, we link flood vulnerability to displaced people. This allows us to estimate internal displacement resulting from the flood event, and to further assess how displacement varies across different areas. We find that a flood depth threshold of 0.67m (CI 0.35 - 1.10) provides a best fit to all data from Sindh province. We find a negative correlation between displacement and the degree of urbanisation. By testing the performance of our model in explaining differing displacement estimates reported across Pakistan, we show the limitations of existing impact assessment frameworks. We emphasise the importance of estimating potential displacement alongside other impacts to better characterise, communicate, and ultimately respond to the impacts of floods.

Non-technical summary: We identify a set of essential recent advances in climate change research with high policy relevance, across natural and social sciences: (1) looming inevitability and implications of overshooting the 1.5°C warming limit, (2) urgent need for a rapid and managed fossil fuel phase-out, (3) challenges for scaling carbon dioxide removal, (4) uncertainties regarding the future contribution of natural carbon sinks, (5) intertwinedness of the crises of biodiversity loss and climate change, (6) compound events, (7) mountain glacier loss, (8) human immobility in the face of climate risks, (9) adaptation justice, and (10) just transitions in food systems. Technical summary: The Intergovernmental Panel on Climate Change Assessment Reports provides the scientific foundation for international climate negotiations and constitutes an unmatched resource for researchers. However, the assessment cycles take multiple years. As a contribution to cross- and interdisciplinary understanding of climate change across diverse research communities, we have streamlined an annual process to identify and synthesize significant research advances. We collected input from experts on various fields using an online questionnaire and prioritized a set of 10 key research insights with high policy relevance. This year, we focus on: (1) the looming overshoot of the 1.5°C warming limit, (2) the urgency of fossil fuel phase-out, (3) challenges to scale-up carbon dioxide removal, (4) uncertainties regarding future natural carbon sinks, (5) the need for joint governance of biodiversity loss and climate change, (6) advances in understanding compound events, (7) accelerated mountain glacier loss, (8) human immobility amidst climate risks, (9) adaptation justice, and (10) just transitions in food systems. We present a succinct account of these insights, reflect on their policy implications, and offer an integrated set of policy-relevant messages. This science synthesis and science communication effort is also the basis for a policy report contributing to elevate climate science every year in time for the United Nations Climate Change Conference. Social media summary: We highlight recent and policy-relevant advances in climate change research – with input from more than 200 experts.