Main

Mitigating climate change will require substantial decarbonization of transportation sectors globally1. A key tool for decarbonizing transport in low-, middle- and high-income countries is electric vehicles (EVs)2. Driven by battery technology improvements, declining costs and policy incentives, EV sales have surged in recent years, increasing from 3% to 18% of all new car sales globally from 2019 to 20233.

Increasing adoption of EVs will coincide with worsening climate change, some of which is already committed due to historical emissions4,5. A growing body of literature examines the consequences of climate change for various energy technologies, including growing decarbonization technologies such as wind6,7, solar8,9, hydro power10,11 and nuclear power12. The performance of each of these technologies is dependent on local weather conditions and, in turn, climate change.

EV battery performance and longevity are highly sensitive to local weather conditions. Studies using semi-empirical EV battery degradation models find that battery lifetimes could vary by more than twofold across climates within the United States13 and by more than fourfold globally14, with mild climates having the best longevity. Real-world data confirm these insights; hot climates have threefold greater EV battery degradation rates than temperate climates in the United States. Battery degradation models primarily examine battery degradation as two processes: calendar aging and cycling aging. Calendar aging encompasses all degradation mechanisms that occur after the battery is manufactured. Cycling aging refers to the deterioration that occurs as a direct consequence of charging and discharging the battery. Calendar and cycling aging are primarily governed by battery temperature and the battery’s electrical behaviour, including state of charge (SOC), discharge depth and charge rate. Heat stress over 40 °C destabilizes electrodes and gives rise to parasitic reactions, exacerbates electrolyte decay through more active decomposition and gas evolution and undermines overall mechanical strength, which leads to faster calendar aging and cycling aging. Under cold environments (0 °C and lower), side reactions contributing to calendar aging are suppressed due to low-temperature kinetics, thus slowing calendar aging. However, cycling aging is accelerated due to increased lithium plating risks15,16,17.

Despite the large sensitivity of EV battery performance and lifetime to weather conditions, existing research has not quantified the effect of future climate change on battery performance and lifetime. As climate change increases temperatures and the frequency and magnitude of extreme heat, calendar and cycling aging could accelerate with greater heat exposure but also decelerate with less exposure to extreme cold. The net effect of these factors could shorten EV battery lifetimes, increasing EV costs and lifetime carbon footprint18,19. Given that most EVs will be deployed in the following decades and are expected to operate for 15–20 years and that EV battery longevity is a core concern for both manufacturers and consumers, understanding the impact of climate change on EV battery longevity and performance is crucial20.

The relationship between climate change and EV battery longevity and performance, though, is not static. Rather, it evolves with advances in battery technology, such as through reduced temperature sensitivity and longer lifetime21. Commercial EV battery lifetimes and durabilities have improved substantially in recent years22,23 by refining pack design, materials and manufacturing processes24. Greater lifetimes and durabilities are a major reason for increasing EV resale values3. As a result, battery technology advancement driven by decarbonization can effectively double as an important climate adaptation strategy for EVs. Capturing interactions between the dynamics of climate change and technology development is rare. Typically, studies use a static representation of technology when quantifying the consequences of climate change25,26,27,28,29,30. In contrast, ref. 31 shows that improvements in onshore wind technology more than counteract the negative effects of climate change on wind generation globally.

In this study, we quantify the consequences of climate change on the performance and longevity of EV batteries and how these consequences are moderated by recent battery technology advances. We make two contributions. First, we assess EV battery performance and longevity under climate change: that is, in weather conditions under which most EVs will actually operate. Second, we add to a nascent body of literature that captures interactions between technology development and climate change, providing a more nuanced and complete picture of climate change impacts and adaptation needs.

We combine EV simulation models, semi-empirical physical degradation models and high-resolution climate data to estimate how climate change accelerates EV battery degradation while accounting for battery technology improvements (Supplementary Fig. 1). Climate change will primarily affect EV batteries’ lifetimes in two ways: first, by changing vehicle energy efficiency, which in turn influences charging cycles and therefore cycling aging; and second, by changing battery temperatures due to shifts in ambient environmental conditions, which therefore influences calendar and cycling aging. To capture these dynamics, we simulate battery hourly operating temperatures, SOC and charging rates from an EV driving and thermal management model. We use these variables as inputs to degradation models to analyse battery lifetimes. By using degradation models published in varying years, we capture the effect of historical technological evolution while acknowledging that such models may lag behind the state of the art and that longevity is only one of several competing priorities in battery design. To study the effects of climate change, we simulate EV operations and degradation under warming levels of 1–4 °C relative to the historical era by statistically downscaling and bias-correcting output from an ensemble of eight global climate models (GCMs) from the Coupled Model Intercomparison Project 6 (CMIP6) archive32. Our downscaling yields hourly weather data, capturing mean, extreme and variability changes in weather. We analyse 300 major global cities and consider the implications of regional changes in battery lifetimes on globally equitable transportation electrification.

Battery lifetime reductions driven by climate change

Commercial EV battery durability has markedly improved through advancements in battery component design, electrode materials, electrolyte formulae and additives, coating and overall manufacturing processes21,22,23,24 (see Supplementary Section 2.7 for a detailed discussion of advancements). These developments are reflected in enhanced cycle and calendar lifetimes, reduced degradation rates and reduced temperature sensitivity. Literature reporting these durability gains began emerging in 2019, around the end of the innovator-adopter phase33 (Supplementary Fig. 2).

To capture this technology improvement, we classify battery models into two time periods: those from 2010 to 2018, or ‘old’ batteries, and those from 2019 to 2023, or ‘new’ batteries. In our analysis, we use performance ensembles of these two types of battery degradation models to represent technological improvement (Methods, ‘Battery degradation model’).

Throughout our results, we report degradation in years for interpretability, assuming constant battery capacities across different climate scenarios. In practice, variations in battery size, annual mileage and vehicle efficiency can influence calendar lifetimes. However, our analysis accounts for these factors by sampling uncertainties in common EV configurations and operations, as detailed in ‘Sampling uncertainties for robust results’.

Battery technology advances offset expected impacts of future warming on battery lifetime (Fig. 1). As a result, battery lifetimes with technology improvement, even at 4 °C warming, exceed battery lifetimes without technology improvement in the historical climate (Fig. 1a). This offset arises from two dynamics: (1) increasing overall lifetimes of new batteries, and (2) greater resilience of new batteries against a warming climate. Under a historical climate, median battery lifetimes from old to new batteries increase from roughly 15 to 17 years. Without technology improvements, old battery lifetimes decrease substantially as warming levels increase, with median lifetimes across cities declining from 15 to 12 years (or by 20%) at 4 °C warming. With technology improvements, battery lifetimes exhibit a smaller decline with increasing warming levels, with median lifetimes remaining at 17 years at 4 °C warming. The smaller distributions of lifetimes, which capture variability across cities, for a given warming level show that new batteries have greater resilience to climate warming than old batteries. At 3 °C warming, for instance, old batteries experience lifetime declines of up to 30% in some cities, whereas new batteries experience declines of up to 8% (Fig. 1b).

Fig. 1: Impact of climate change and technological progress on battery lifetimes.
Fig. 1: Impact of climate change and technological progress on battery lifetimes.The alternative text for this image may have been generated using AI.
Full size image

a, Lifetimes of old batteries (brown and red bars) and new batteries (blue bars) under historical and future climates (the latter given by warming level). Old and new batteries correspond to battery degradation models published in 2014–2018 and 2019–2023, respectively. Box plots show the distribution of battery lifetimes across global cities, where battery lifetimes equal the mean lifetime of the entire ensemble of old or new batteries. For individual battery model outcomes, see Supplementary Section 2.8. Dashed lines indicate a baseline of old batteries in historical climate (brown dashed) and trendlines in old (red dashed) and new (blue dashed) battery lifetimes across future warming levels. Results in the boxed section are further detailed in Fig. 2b,c. b, Relative changes in battery lifetime (%) under future climate warming levels compared with lifetime under historical climate for old and new batteries (red and blue bars, respectively). Brown and blue dashed lines indicate the lifetimes of old and new batteries, respectively, under historical climate. Results in the boxed section are further detailed in Fig. 3a,b. In a and b, n = 300 global cities (unit of study: city). Box plots show the median (centre line), 25th and 75th percentiles (box limits), whiskers extending to the minimum and maximum values within 1.5 × interquartile range (IQR) and points beyond whiskers as outliers.

With old technology, battery lifetimes vary from 7 to 18 years across cities under 2 °C warming (Fig. 2b). These lifetimes exhibit large heterogeneity across space, as cities with high mean air temperatures have lower lifetimes than cities with low mean air temperature (Fig. 2a). Warming of 2 °C drives major reductions in lifetimes, with the maximum reduction reaching 30% and 46% of cities experiencing lifetime reductions of 10% or greater (Fig. 3a). Five percent of cities see an increase in lifetime with future warming, primarily in China and South America, where moderate temperature increases and reductions in solar irradiance occur.

Fig. 2: Old and new battery lifetimes.
Fig. 2: Old and new battery lifetimes.The alternative text for this image may have been generated using AI.
Full size image

a, Density plot showing the distribution of battery lifetimes across varying mean temperatures for three scenarios: baseline (old battery, historical climate), scenario 1 (S1) (old battery, future climate at 2 °C warming) and scenario 2 (S2) (new battery, future climate at 2 °C warming). Each dot represents a city. b,c, Global distributions of battery lifetimes under 2 °C warming for old (b) and new (c) batteries (corresponding to S1 and S2 in a, respectively). Lifetimes of old and new batteries under historical climate are available in Supplementary Fig. 6. Each dot represents a city (n = 300). Basemaps in b and c from Natural Earth (https://www.naturalearthdata.com).

Fig. 3: Battery lifetime changes due to climate change.
Fig. 3: Battery lifetime changes due to climate change.The alternative text for this image may have been generated using AI.
Full size image

a,b, Battery lifetime changes for old (a) and new (b) batteries from historical to 2 °C warming. Lifetime reductions are portrayed spatially (top) and as histograms (bottom). Colour bars apply to histograms and maps and are held constant between a and b to facilitate comparison. Basemaps from Natural Earth (https://www.naturalearthdata.com).

Relative to old battery technology, new batteries have longer and less spatially variable lifetimes and experience much smaller reductions in lifetimes due to future warming (Fig. 2). With new technology, battery lifetimes vary from 15 to 19 years across cities under 2 °C warming (Fig. 2c). For roughly half of our studied cities, future warming decreases battery lifetimes, by an average and maximum of 3% and 10% (Fig. 3b). These lifetime declines occur in cities with high mean temperatures, but these cities have pronounced climate resilience benefits from new versus old technologies (Figs. 2a and 3). In the other half of our studied cities, future warming actually increases battery lifetimes, by an average and maximum of 5% and 12%, respectively (Figs. 2a and 3b). Battery lifetimes exhibit the largest increase in northern cities, where reduced cycling aging due to moderate climate outweighs increased calendar aging (see ‘Mechanisms explaining lifetime changes under climate change’ for an illustration). Nearly a third of cities see less than 1% change in battery lifetimes due to future warming.

Regional inequities in lifetimes driven by climate change

Large regional disparities in battery lifetimes under future warming given old and new battery technology exist (Fig. 4). With old battery technology, lifetime reductions driven by climate warming are greater in countries with low gross domestic product (GDP) per capita than those with high GDP per capita. Average lifetime reductions in the countries with the lowest GDP per capita range from 5% to 25% from 1 °C to 4 °C warming (Fig. 4b). These large reductions are concentrated in Africa, Southeast Asia and India (Fig. 4a). In contrast, average lifetime reductions in the countries with the highest GDP per capita, such as those in the European Union and North America, range from 5% to 15% from 1 °C to 4 °C warming.

Fig. 4: Battery lifetime changes by region.
Fig. 4: Battery lifetime changes by region.The alternative text for this image may have been generated using AI.
Full size image

a, Relative change compared with baseline in battery lifetime (%) for old and new batteries (red and blue boxes, respectively). The baseline (brown dashed line) is defined as old batteries in the historical climate. Each subplot shows results for cities in one of six regions. In a, the bar consists of cities in each region out of 300 total cities. Box plots show the median (centre line), 25th and 75th percentiles (box limits), whiskers extending to the minimum and maximum values within 1.5 × IQR and points beyond whiskers as outliers. b,c, Regional equity analysis shows the relationship between GDP (per capita, PPP-adjusted) and battery lifetime reduction under different warming scenarios (+1 °C to +4 °C), shown in separate graphs for old (b) and new (c) batteries. Steeper slopes indicate larger lifetime reductions with warming in low compared to high GDP countries. Scatter plots aggregate 300 cities into 10 bins using quantile-based binning, with each bin representing the average change in battery lifetime for cities. The x axis indicates the GDP basis for bin aggregation. Solid lines are linear fits; shaded bands indicate 0.75 confidence intervals. PPP, purchasing power parity; SEA, Southeast Asia.

New battery technologies, because they are more resilient to climate change, dampen the inequitable inverse relationship between GDP per-capita and lifetime reductions. Average lifetime reductions in the countries with the lowest GDP per capita range from 1% to 4% from 1 °C to 4 °C warming (Fig. 4c), much less than with old battery technologies. The countries with the highest GDP per capita tend to see decreases of less than 1% or even increases of up to 2% from 1 °C to 4 °C warming. Thus, although lifetime reductions are still inversely related to GDP per capita, new battery technologies eliminate most of the differences between these groups of countries and yield major overall lifetime benefits (of 25–40%) to the lowest-income countries, particularly in Africa, Southeast Asia and India (Fig. 4a).

Mechanisms explaining lifetime changes under climate change

Two mechanisms primarily drive battery lifetime changes in a warming climate: full equivalent cycles (FECs) and cell temperatures (Supplementary Fig. 3). Our EV model tracks FECs and cell temperatures as a function of drive cycle, ambient air temperature, radiation intensity and battery thermal management. To explore the role of these mechanisms, we divide our cities between those with cold, medium and hot climates34 (detailed in Supplementary Table 1). Cell temperature and FEC changes are driven by climate change in our analysis and are the same across old and new battery technologies, whereas their impact on lifetime varies based on the different technologies’ durabilities.

Across all cities, climate change elevates battery operational temperatures (Fig. 5a). Median cell temperatures across cities in each climate zone increase by 1–6 °C from the historical climate to 4 °C warming. Climate change also affects FECs (Fig. 5b), although these effects are mixed across climate zones and more modest than cell temperature changes. In cold climates, climate change decreases FECs by up to 5% from the historical to 4 °C warming, whereas in hot climates, it increases FECs by up to 6%. Marginal changes in FECs are observed in medium-climate cities.

Fig. 5: Decomposition of factors affecting battery lifetime.
Fig. 5: Decomposition of factors affecting battery lifetime.The alternative text for this image may have been generated using AI.
Full size image

a, Old and new battery cell temperatures across cold, medium and hot cities under historical and future warming levels. Box plots show distributions across global cities. Dashed brown lines indicate median temperature under historical climate for comparison with future warming levels. Classification of cities by current climate is given in ref. 34 (detailed in Supplementary Table 1). b, Old and new battery cycling, quantified as FECs, across cold, medium and hot cities under historical and future warming levels. Box plots show distributions across cities. Dashed brown lines indicate median FECs under historical climate for comparison with future warming levels. In a and b, the bar consists of cities in each climate zone out of 300 total cities. Box plots show the median (centre line), 25th and 75th percentiles (box limits), whiskers extending to the minimum and maximum values within 1.5 × IQR and points beyond whiskers as outliers. c,d, Waterfall charts comparing the effects of cycling and cell temperature (T_Cell) on battery life with 2 °C warming for old (c) and new (d) batteries. Purple bars show changes in lifetime due to changes in cycling (FECs) under future climates relative to historic. Blue bars show changes in lifetime due to combined changes in cycling and cell temperatures. Green bars (second bar in each subplot) show changes in lifetime due to changes in cell temperatures from historical to future climate, calculated as the difference between the blue and purple bars. In c and d, values of bars represent the mean across cities in each climate zone out of 300 total cities.

We find that cell temperature, not FEC, changes are the primary drivers of lifetime changes under future warming across old and new battery types (Fig. 5c,d). This temperature impact is from both mean warming and changes in weather variability. For old batteries, mean warming is the dominant driver (∼80% of lifetime loss), and increased variability is also responsible for a large portion (∼20%) (Supplementary Fig. 4). Nonetheless, FECs and cell temperatures interact to drive battery lifetime changes under climate change (Supplementary Fig. 5a,b). In hot-climate cities, increased cell temperatures under climate change drive a stronger decline in battery lifetimes than in mild- and cold-climate cities, whereas new batteries moderate this reduction compared to old batteries. Increasing FECs also generally reduces battery lifetimes, and new batteries also moderate this reduction. Conversely, decreasing FECs, which occurs in cold climates, generally increases battery lifetimes for new but not old batteries; for old batteries, the benefit from decreased FECs at high warming levels is more than offset by the correlated increase in cell temperature.

Discussion

Our study demonstrates that technological advancements in battery technology have largely offset future adverse effects of climate warming on battery lifespans. These technological advancements also reduce geographic variability in the impacts of climate change (Figs. 1 and 3), suppressing regional inequities (Fig. 4). Our analysis is grounded in substantial real-world evidence that demonstrates (1) a clear and rapid trend of improving battery durability and (2) the relationship between battery longevity and its usage and temperature (Supplementary Section 2.7).

Our research focuses on durability improvements under climate change in commercially available battery technologies. These durability improvements are a result of degradation-mitigating strategies, including improvements in battery component design, electrode materials, electrolyte formulae and additives and coating35. Emerging battery technologies, such as solid-state batteries36 and silicon anode batteries37, have shown promising enhancements in energy density, safety and material cost reduction. These new technologies, which are still under development to meet durability goals for commercial application38, could be sensitive to elevated temperatures caused by climate warming, affecting their durability37.

Our findings underscore the need to enhance durability during development and diffusion of advanced batteries. Future research and development efforts for EV batteries, for instance, should consider larger cycle times and higher operational temperatures expected in a warming climate. In particular, considering the highly uneven effects of climate warming on battery lifetime across different regions of the world, it could be valuable to focus more on localization to account for local climate and projected climate change during the EV lifetime. Our findings also provide a foundation for future research to add even greater realism to battery durability modelling. A key avenue is to build upon our analysis by moving from cell-level degradation models to more complex pack-level analyses, which would better capture the influence of specific battery pack designs and thermal management strategies. Our results also suggest that for battery ‘passports’ to accurately assess a battery’s state of health, it is crucial to know not just number of cycles but also the temperatures at which those cycles occurred; and these analyses should account for changes in weather variability and extremes, not just means, to avoid underestimating climate change impacts and adaptation needs (Fig. 5 and Supplementary Fig. 4).

Our analysis has several limitations. We do not assess the direct impacts of climate change on other critical parts of the EV ecosystem, such as drivetrain efficiency, the resilience of charging infrastructure or overall system-level vehicle reliability. Climate warming may also cause EV integration challenges and consumer behavioural changes, both of which affect EV longevity39. In the absence of rich driving behaviour datasets for many global regions, we adopt diversified mobility patterns from Germany40. More location-specific mobility data could yield more spatially nuanced insights. Given limits on publicly available and validated data, we use EV battery degradation models trained on battery cells rather than battery packs, which may make them optimistic in absolute terms. However, the potential optimistic estimation applies to packs of all vintages. Therefore, our estimate of progress is likely more robust than our estimate of absolute degradation (Supplementary Section 2.7).

Real-world vehicle fleets could be more diverse and could include EV models with inferior battery thermal management systems (BTMSs). For such EV models with inferior thermal management, our results could place a lower bound on the impacts of climate change, which future region-specific research could analyse in detail.

Our findings highlight a broader need to consider the role of technological progress in mediating the impact of climate change on energy technologies. Limited research on energy technologies successfully captures the dynamic relationship between technology and climate evolution31. Capturing these intertwined dynamics can shift our understanding of future climate adaptation needs, which in turn could affect investment and policy decisions. Capturing these intertwined dynamics of climate and technological evolution could also be invaluable for other sectors, including land management41, residential energy demand27 and transportation infrastructure29.

Methods

Our analytical framework simulates the operation and degradation of common commercial EV battery modules under historical and future climate scenarios (Supplementary Fig. 1). We employ semi-empirical physical degradation models to estimate how climate change accelerates EV battery degradation and account for different technological levels by incorporating battery degradation models published in various years. Battery degradation depends on the EV battery’s hourly operating temperature, SOC and charging rate (‘Battery degradation model’ and Supplementary Section 2.7), which we simulate with EV driving and thermal management models under different environmental conditions (‘Dynamic EV operation model’). Finally, we determine changes in EV battery lifespan from accelerated aging. To project EV driving and battery degradation under future climates, we downscale and bias-correct outputs from eight CMIP6 GCM ensembles using the high-resolution ERA5 reanalysis dataset (‘Climate data and bias correction’). Our analysis includes 300 globally distributed cities and incorporates socio-economic indicators to capture regional variations in future EV battery lifespan (‘Global cities and regional equality’).

Dynamic EV operation model

We modify the open source tool emobpy40 to simulate vehicle operations and to specifically estimate a given EV battery’s hourly operating temperature, SOC and charging rate. Hourly operating temperature, SOC and charging rate are input into our degradation model (‘Battery degradation model’ and Supplementary Section 2.7). We simulate vehicle operations in each of 300 global cities (‘Global cities and regional equality’) under historical and future climates (‘Climate data and bias correction’) by modifying input meteorological variables, specifically temperature and shortwave radiation, and by sampling uncertainties across different vehicle types, travel profiles, charging strategies and parking shading conditions. For more information and visualization of vehicle simulation, see Supplementary Sections 2.3 and 2.4.

In all cities, we model operations of a Tesla Model 3 and a Volkswagen ID.3 vehicle, widely available and popular EV designs, similar to prior research using small or middle-size EVs14,42,43. We use the ID.3 and Model 3 as two representative vehicles capturing a large range of the EV market. These two vehicles have consistent, well-characterized vehicle platforms, allowing us to separate the effects of climate change and battery technology on battery longevity. However, real-world vehicle fleets, particularly in the Global South, could include EV models with worse BTMSs than our two models. For such EV models with inferior thermal management, our results could place a lower bound on the impacts of climate change. To expand the diversity of our analysed EVs, we conduct a comprehensive sensitivity analysis on the Nissan Leaf S. The Leaf S has a smaller battery than the ID.3 and Model 3, representing another important segment of the EV market. See Supplementary Section 2.9 for more information.

A key output of our EV operation simulations is timeseries of battery temperatures. Primary drivers of battery temperatures are operational mode, drive cycle, and BTMS operations. We differentiate three operational modes: driving, parked while charging and parked while not charging. We assume the BTMS operates while driving or charging. We construct timeseries of operational modes using empirical travel profile and charger availability datasets. Because such empirical datasets are rarely available, we use original emobpy datasets for Germany40 and scale travel profiles based on national vehicle kilometres travelled (Supplementary Section 2.3), similar to prior research44. Although local empirical data would be preferred, our empirical profiles from Germany capture heterogeneous and globally relevant travel behaviour, including workday commuting, greater travel during days than nights, overnight parking at home and irregular trips: for example, long-distance holiday travel. To capture uncertainty in travel profiles, we sample three travel profiles per city. Travel profiles indicate whether a car is driving or parked. To differentiate between parked while charging or not charging, we use emobpy to probabilistically model charger availability at public and home locations. Given the availability of a charger, charging may or may not occur, which we account for using four charging scenarios that vary decision rules for charging in public and at home at night40,42 (Supplementary Section 2.6). We therefore run 12 operational simulations per city: four charging strategies for each of our three sampled travel profiles.

We also simulate battery temperatures and cycling (including SOC) during driving. Emobpy uses a two-stage simulation for driving, an approach used in other research14 (Supplementary Section 2.4). First, travel profiles define whether a vehicle is driving and at what average speed at 15-minute resolution. Second, when driving, higher-resolution drive cycles aligned with the reported average speed are used to simulate battery temperatures and cycling. These drive cycles are from Worldwide Harmonized Light Vehicles and the US Environmental Protection Agency standard drive cycles. Battery temperatures and cycling during driving are affected by air temperatures and solar radiation vis-a-vis effects on heating, ventilation, and air conditioning and BTMS energy consumption and by powertrain energy consumption.

Given the importance of BTMS in moderating battery temperatures, we add a BTMS model and improve EV thermal modelling within emobpy. Our EV thermal model is a lumped capacitance thermal network approach using a thermal equation of state that captures interactions between battery temperatures, cabin temperatures and ambient conditions. Ambient temperatures and solar radiation serve as boundary conditions for the lumped battery–cabin model. Our thermal model captures battery self-heating effects during charging and discharging operations. It also captures thermal energy exchanges between the battery and cabin and between each and ambient conditions (including solar radiation for the cabin).

We explicitly model a BTMS that actively heats or cools battery cells using a coupled electrothermal model at 10-s intervals45. BTMS operations follow temperature thresholds, activating cooling when temperatures approach upper battery operational limits (50 °C) and heating when temperatures fall below optimal ranges (15 °C) (see Supplementary Section 2.5 for more details). From our BTMS model, we also obtain thermal management energy consumption (Supplementary Section 2.5).

Battery degradation model

To model degradation of EV batteries, we use the BLAST package46 from the National Renewable Energy Laboratory. BLAST includes 11 battery degradation models published between 2014 and 2023, as summarized in Supplementary Table 3. All of these models are semi-empirical physical–chemical degradation models or data-driven degradation models that have been validated and fitted on experimental data. For these models, battery cells were calendar-aged at varying temperatures and SOCs and cycled at varying voltage windows, rates and duty cycles47 and widely used in prior research for battery lifetime prediction14,18,43. Further, the functional form of degradation in BLAST has been validated against EV battery degradation under real-world conditions48. By dividing these degradation models into two groups spanning 2014–2018 and 2019–2023, we capture technological progress in battery degradation. Degradation model inputs are hourly battery temperatures and SOCs from ‘Dynamic EV operation model’. From BLAST, we obtain calendar and cycling aging. We then translate aging to battery lifetime given an end-of-life threshold for 75% remaining capacity compared with the initial capacity. The threshold is aligned with warranties and industrial practice20. Our model applies a 25-year cap on all battery lifetimes to reflect a realistic operational lifespan of a vehicle. For more details and a discussion of limitations related to degradation modelling, see Supplementary Section 2.7. For validation of our degradation modelling against real-world fleet-level data, including from hot climates, and further details on the validation underlying BLAST, see Supplementary Section 2.7.

Our main results report the lifetime of our two battery groups (those from 2014–2018 and 2019–2023) by averaging the battery models within each group. However, our analysis quantifies degradation separately for each battery type, capturing variability in durability between battery models, as shown in Supplementary Section 2.8.

Climate data and bias correction

Accelerated degradation of EV batteries is caused by the accumulation of stresses from vehicle energy consumption and battery thermal dynamics in extreme hot or cold operating conditions, which can be captured only at high spatial and temporal resolution. We create a highly temporally and spatially resolved, bias-corrected weather dataset for historical and future climates, within which to simulate EV operations and battery degradation. We use historical and future periods spanning 2000–2100 from eight CMIP6 GCMs (Supplementary Table 2). Equilibrium climate sensitivities (ECSs) of our eight GCMs are shown in Supplementary Fig. 7. These models were chosen to represent a range of ECS values to capture the uncertainty in future climate projections. Variables are retrieved at 3-h resolution and include 2-m air temperature and specific humidity, surface downward solar radiation. We linearly interpolate variables from 3-h to 1-h resolution and bias-correct and downscale these variables using the reanalysis dataset (described below). These meteorological variables are used as boundary climate conditions to calculate EV driving energy consumption and EV battery thermal dynamics using the models as described in Supplementary Sections 2.4 and 2.5, including air temperature and solar shortwave radiation.

In line with the growing consensus to quantify the impact of climate change on discrete levels of global warming49, we present our results for warming levels of 1 °C to 4 °C relative to the historical climate (2000–2010) in 0.5 °C intervals. This approach reduces uncertainty caused by varying ECS between models50. A certain level of warming, say X °C, is expressed as years in which the annual global average temperature is X plus or minus 0.25 °C higher than the historical climate. We obtain future climate data for Shared Socioeconomic Pathway 5-8.5, a high-future-warming scenario, enabling us to analyse a wider range of warming levels. However, our warming-level-oriented approach generates insights relevant to any climate scenario because climate realizations at a given warming level are similar across climate scenarios50.

GCMs are affected by systematic biases. To address this issue, we perform statistical downscaling and bias correction of CMIP6 atmospheric temperature and downwelling shortwave radiation based on the ERA5 reanalysis following the approach in ref. 50. This method overlays the GCM simulated climate change signal onto the ERA5 baseline after interpolation to ERA5 resolution. Here we provide a brief overview of the process. First, the same meteorological variables at hourly resolution are retrieved from ERA5 for 2000–2010, called the ERA5 baseline. Next, for each GCM and warming level, we derive the temporal changes in operating parameters by comparing hourly climate averages between 2000–2010 and the periods corresponding to the selected warming level. Finally, the time-varying fields are added at hourly intervals to each year within the ERA5 baseline. This generates 10 years of hourly bias-corrected samples for each GCM and warming level.

For more information and visualization of our climate data processing for battery simulation, see Supplementary Section 2.2.

Global cities and regional equality

With our analytical pipeline, we can calculate the accelerated degradation of an EV battery under a future warming level at any location on the globe. We apply our pipeline to quantify accelerated degradation at 300 major global cities based on population using cities’ coordinates from ref. 51. We aggregate proximate cities sharing a GCM grid cell that has the same climate signal. In analysing differences in outcomes between countries, we obtain national GDP from the World Bank World Development Indicators (https://databank.worldbank.org/source/world-development-indicators/preview/on).

Sampling uncertainties for robust results

To generate robust results against uncertainties in our study, we sample uncertainties across weather conditions and EV operations. For weather conditions, we sample interannual variability using 10 years (2000–2010) of weather data and quantify future climate impacts across several warming levels and eight GCMs. For EV operational conditions, we sample two vehicle types, three travel profiles, four charging strategies and two parking shade conditions. For each city, we robustly account for uncertainties by simulating all possible combinations of 80 years of weather data and 48 EV operation scenarios across warming levels.