Will reshoring manufacturing of advanced electric vehicle battery support renewable energy transition and climate targets?

Recent global logistics and geopolitical challenges draw attention to the potential raw material shortages for electric vehicle (EV) batteries. Here, we analyze the long-term energy and sustainability prospects to ensure a secure and resilient midstream and downstream value chain for the U.S. EV battery market amid uncertain market expansion and evolving battery technologies. With current battery technologies, reshoring and ally-shoring the midstream and downstream EV battery manufacturing will reduce the carbon footprint by 15% and energy use by 5 to 7%. While next-generation cobalt-free battery technologies will achieve up to 27% carbon emission reduction, transitioning to 54% less carbon-intensive blade lithium iron phosphate may diminish the mitigation benefits of supply chain restructuring. Our findings underscore the importance of adopting nickel from secondary sources and nickel-rich ores. However, the advantages of restructuring the U.S. EV battery supply chain depend on projected battery technology advancements.

EV sales by the model in the United States (57). Battery capacity for each model is obtained from an Argonne National Laboratory study (60). Electric range and energy efficiency are extracted from fueleconomy.gov (59). Battery chemistry data are obtained from both existing journal articles and industry reports (11,(66)(67)(68)(69). As lithium nickel manganese cobalt oxide (NMC622) dominates the battery chemistry found for the US BEV fleet, all NMC LIBs with unknown compositions of nickel, manganese, and cobalt are assumed to be NMC622. Acronyms: lithium nickel cobalt aluminum oxide (NCA); lithium manganese oxide (LMO).  is the largest LIB cell and component producer, following China, Japan, Korea, and the US (70). We attribute all rest-of-world (RoW) production capacity to the EU for simplification.    Table S6 Reallocated production capacity of LIB components by region for the ally-shoring scenario (17). The production capacity of China is allocated to other regions proportional to their current production capacities. The production capacity of LIB cells and packs are not reallocated for the ally-shoring scenario because no LIB cell or pack of the US EV fleet with over 1,000vehicle sales is produced in China.

Interrupted nickel supply and US domestic sourcing of critical materials
Given the recent geopolitical tensions over the global critical material supply chain, we conduct sensitivity analysis on the nickel supply from Russia for both scenarios and domestic sourcing of critical materials for the US manufacturing scenario. We assess the impact of no nickel supply from Russia by reallocating its shares of cobalt and nickel mining capacities as well as the nickel refining capacity to other top miners and refiners mentioned in Table S7. Moreover, we model the impact of 100% domestic sourcing of critical materials, including lithium, nickel, and cobalt, by adapting the mining and refining processes of Canada to the US. Specifically, NERC regions where mines are located in the US are selected for the power grid; Transportation distances are recalculated. According to the US's recent Defense Production Act, lithium is projected to be mainly produced from geothermal brine, with about 400,000 t lithium carbonate equivalent annual production capacity from Berkshire Hathaway Energy Renewables, Controlled Thermal Resources, and EnergySource Minerals (71). Meanwhile, the only operating lithium mine in the US produces 60,000 t lithium carbonate equivalent per year. Therefore, we adopt the life cycle inventory of lithium extraction using the geothermal brine, considering future demand growth and potential supply from geothermal brine based on existing works (33,72). We consider a weighted sum of the environmental impacts of lithium produced from both the conventional pathway and geothermal brine, following their respective projected annual production capacities.
The most influential factors to the manufacturing cost of automotive LIBs for the US case include the battery pack energy capacity and EV type. When LIB pack energy capacity is less than 19 kWh, the cost of purchased items takes the largest proportion of the manufacturing costs (Fig.  S9b). Otherwise, material cost dominates LIB's manufacturing costs. This is primarily because the amount of purchased items for 1 kWh LIB is larger with lower energy capacity. Purchased items include terminals, cell container, module enclosure, battery jacket, battery management system, thermal management system, and other auxiliary hardware. The amount of purchased items is determined by several design parameters, including the number of cells per module, number of cells and modules per battery pack, dimension of cells, modules and packs, energy capacity, and battery chemistry. While the number of cells per module and the number of cells and modules per battery pack are not accessible for most LIBs in different EV models, for a fair comparison, we adopt the default settings from the BatPac model and keep these parameters constant regardless of the energy capacity and battery chemistry of LIBs (61). The dimension of cells, modules, and packs are highly correlated with the energy capacity and battery chemistry of LIBs. EV type has an impact on the cell thickness as BEVs and PHEVs have different requirements for batteries. Specifically, the electrode thickness is determined by the sustained power requirement for BEVs and acceleration power requirement for PHEVs (61). In addition to the lower unit purchased item cost for high-capacity LIBs, the economies of scale for capital equipment, plant area, and working time are considered. Therefore, LIBs with greater energy capacity achieve lower unit manufacturing costs, as shown in Fig. S9c.
Material cost also takes advantage of volume discounts, but the benefits are minor, and the unit material cost remains almost constant with increasing energy capacity. In terms of the battery chemistry, LIBs with higher specific energy density tend to attain lower unit manufacturing costs, as shown in Fig. S8. However, due to the regional differences in producing cells and packs, manufacturing the high-performance NCA LIBs for BEV models of the US EV fleet gains less cost advantage over other types of LIBs. Despite the factors to raise the unit manufacturing cost of NCA LIBs, with a higher specific energy density, its unit manufacturing cost is still slightly lower than that of NMC622 LIBs.
The spatial variation in the EV battery value chain lies in the costs of cathode active materials, labor, construction, and transportation. These costs vary according to the production regions of LIB components, cells, and packs for each EV model. As cathode active materials are sourced from China, South Korea, Japan, the US, and the EU with respective shares (Table S3), spatial variation is not presented across different EV models. Among all the EV models, LIB cells and packs produced in South Korea, Poland, and Hungary achieve substantial cost benefits due to low labor costs. Only 1.6% of the LIB manufacturing cost can be attributed to transportation. However, the container shipping rates keep surging throughout the COVID-19 pandemic due to the lockdown measures and expanding electronic commerce. The sensitivity analysis result shows that the LIB manufacturing cost can be increased by up to 1.5% with a 166% higher container shipping rate (Fig. S10). The effect of spatial variation in building cost on LIB's manufacturing cost is subtle. Fig. S10d shows that the LIB manufacturing cost of US manufacturing ($164/kWh) and allyshoring ($155/kWh) scenario is 11% and 5% higher than the US EV fleet case ($147/kWh), respectively. The increase in the LIB manufacturing cost can be mainly attributed to the geographic variation in cathode active material cost and labor cost. Cathode active material cost and labor cost explain around 80% and 20% of the differences in the LIB manufacturing costs among these scenarios. As shown in Table S3, 42% of the LIB cathodes are produced in China, and the cathode active material cost in China is 21%-42% lower than that in South Korea, the EU, Japan, and the US. The lower cathode active material cost can be mostly explained by the lower material, labor, and building cost in China. First, China is one of the largest producers of valuable metals used in LIBs, which accounts for 60%, 72%, and 16% of the world's production capacity for refined lithium, cobalt, and nickel, respectively (17). Therefore, there exist advantages in the material cost of refined lithium, cobalt, and nickel for local LIB manufacturers in China (72). Second, labor cost in China is 50%-88% lower than in the US, EU, South Korea, and Japan, as shown in Table S19. Third, as building cost is driven by labor cost (72), China has the lowest building cost among the LIB cathode production regions. Fig. S10d also indicates that the production of LIB components, cells, and packs exclusively in the Far East or the EU could be more cost-effective than in the US. Therefore, the increasing proportion of LIB component, cell, and pack production in the US, as shown in Fig. S10a-c, causes the worse economic performances of the US manufacturing and allyshoring scenarios compared to the US EV fleet case. Soaring nickel price at up to $48/kg, two times higher than the pre-pandemic nickel price, can lead to a 16-21% increase in the EV battery manufacturing cost, which would delay the vehicle electrification due to demand-side challenges. When sourcing key critical minerals and materials domestically, the price of nickel exported from Canada to the US is used as a proxy for the domestic nickel price, which is lower than the nickel price exported from other countries, possibly due to lower transportation costs and higher nickel ore grade in Canada. Owing to the favorable nickel price and economic performance of lithium sourcing from geothermal brine, US's LIB products may gain cost advantages from domestic sourcing of critical materials.
To fairly allocate the climate change mitigation responsibilities, we calculate the border carbon adjustments required to achieve parity between LIBs produced in the US and their counterparts imported from the EU and main Asia suppliers. We compare these break-even border carbon adjustments with the carbon capture and storage costs reported in the existing literature and discuss the feasibility of avoiding carbon leakage.

Fig. S9
Economic performance of LIBs for the pre-pandemic US EV fleet. a, Breakdowns of manufacturing cost by the manufacturing process and cost type. Breakdowns of material costs are also presented. b, Trends for shares of different types of LIB manufacturing costs with increasing battery pack energy capacity. NMC622 and NCA are chosen because they represent 98% of the US EV LIB market. c, Trends for different types of manufacturing costs per kWh energy capacity with increasing battery pack energy capacity. d, Trends for different types of manufacturing costs per pack with increasing battery pack energy capacity. Acronyms: battery management system (BMS).

Fig. S10
Comparison of manufacturing costs across the US EV fleet case, US manufacturing scenario, and ally-shoring scenario. a, LIB manufacturing costs by the country for the US EV fleet case. b, LIB manufacturing costs by country for the US manufacturing scenario. c, LIB manufacturing costs by the country for the ally-shoring scenario. d, Comparison of LIB manufacturing costs across different scenarios, including US EV fleet case, US manufacturing scenario, ally-shoring scenario, and sensitivity analyses for manufacturing LIBs from components to packs in EU and main Asia suppliers. For a-c, the costs for all battery materials or battery assembly processes are aggregated by the production region. The sensitivity analysis results based on the record-high nickel price are presented as solid brown lines. Acronyms: battery management system (BMS).

Break-even border carbon adjustment analysis
A border carbon adjustment is proposed as a type of border tax adjustment to avoid carbon leakage and advance domestic competitiveness. In this study, we calculate the break-even border carbon adjustments for automotive LIBs imported from different regions to the US following the previous literature (85). In order to eliminate trade discrimination, the importing country should not charge border tax adjustments on the imported products in excess of taxes on domestic products, according to the WTO rules (51). In other words, if a border carbon adjustment is imposed on imported products, it should also be charged on domestic products in addition to the existing border carbon adjustments. Moreover, the border carbon adjustments priced in the exporting regions should be compensated by the importing country (52). Based on the above rules, the equation for calculating the break-even border carbon adjustments between the US and the importing regions, including the EU and the major LIB suppliers in the Far East, are revised as shown in equation (1), where I denotes the set of importing countries, indexed by i. These regions are considered as they are the largest LIB producers in the world (16). In equation (1), BCAUS,i denotes the break-even border carbon adjustment between the US and region i, MCUS and MCi represent the manufacturing cost of LIBs in the US and region i, respectively; CPUS and CPi are the domestic carbon price in the US and the importing region i; GWPUS and GWPi are the carbon footprint calculated for manufacturing EV LIBs in the US and the region i.
Note that there is no current domestic carbon tax in the US, China, and South Korea (86). Furthermore, to the best of our knowledge, Japan and European countries where main battery factories are located (i.e., Poland and Hungary) do not levy a carbon tax on EVs (86)(87)(88). However, Emissions Trading System (ETS) has been implemented in many countries as carbon pricing initiatives though none of them are targeting EV battery suppliers at this moment. Therefore, to investigate the potential implications of carbon pricing initiatives on the break-even border carbon adjustments, we conduct a sensitivity analysis on carbon pricing based on the World Bank statistics (86,89). As shown in Fig. S18, 2019 and 2020 carbon prices and border carbon adjustments ranged $0-$127/t CO2 eq. in the EU, $1-$12/t CO2 eq. in China, $8-$17/t CO2 eq. in the US, and $22-$33/t CO2 eq. in South Korea, $3-$6/t CO2 eq. in Japan (86,89). The results are shown in Fig.  S19-SFig. S21.
Border carbon adjustments are the charge levied on the embodied emissions of imports to secure competitive neutrality across countries with differential carbon prices and can be considered to facilitate mitigating climate change. According to World Trade Organization (WTO), in order to ensure competitive equality, countries cannot adjust the taxes on imported products in excess of domestic taxation (51). In other words, a country should impose the same border carbon adjustment on both its domestic products and imported products. To understand the trade-offs between manufacturing costs and climate change of EV batteries, we calculate the break-even border carbon adjustments to be imposed on automotive LIBs consumed in the US, at which the summation of manufacturing cost and carbon price for LIBs manufactured domestically in the US and imported from other regions are equal. A baseline carbon price of $0 per t CO2 is considered as none of the assessed regions levies carbon taxes on EV or EV batteries at this moment. Fig.  S11a shows that the break-even border carbon adjustment for automotive LIBs consumed in the US varies across the sources of the imported LIBs, EV type, and the battery chemistry. Specifically, among the largest LIB producers, the EU achieves negative break-even border carbon adjustments as LIB manufacturing in the EU is less cost-effective than in the US due to the lower embodied greenhouse gas emissions of LIBs from the EU suppliers. LIBs from their Asia suppliers are more carbon-intensive but more cost-effective than their US counterparts. The break-even border carbon adjustment between the US and China is the highest for all types of LIBs. A breakeven border carbon adjustment at $68 to $1080 per t CO2 on EV batteries imported from major Asia suppliers is reported. In particular, the break-even border carbon adjustments between the US and the LIB exporting countries in Asia range from $65 to $1051 per t CO2 eq. for BEV LIBs and $110 to $1543 per t CO2 eq. for PHEV LIBs. The differences in break-even border carbon adjustments by EV type can be attributed to the previously mentioned low battery pack energy density of PHEV LIBs and, consequently, more material and energy input to LIB manufacturing. The highest break-even border carbon adjustment is $1735 per t CO2 for NCA LIBs in PHEVs because there is only one PHEV model for NCA LIBs, and its battery pack energy capacity is the lowest among LIBs in all EV models. The geopolitical tensions over the supply and price of nickel do not significantly change the break-even border carbon adjustments. On the contrary, raising domestic sourcing of key critical minerals and materials, according to the US's recent Defense Production Act, can reduce the carbon footprint of domestic LIB products by 37%, thus lowering the break-even border carbon adjustments between the US and the LIB exporting Asian countries.
The break-even border carbon adjustments between the US and major Northeast Asian economies are within the range of carbon capture and storage costs in the existing studies, as presented in Fig. S11c. The carbon capture costs vary substantially by technology, industry, location, and the inclusion of carbon storage (90)(91)(92)(93)(94)(95)(96). Microsoft and Stripe proposed a biospherebased carbon storage cost of $16/t CO2 eq. and a geosphere-based storage cost of $141/t CO2 eq. with a range of $20-10,000/t CO2 eq. (97). The results of many studies also supported that biosphere-based carbon storage costs less than geosphere-based storage (90)(91)(92)(93)(94)(95).
It is worth mentioning that the scope of break-even analysis is limited to regulating carbon emissions of LIB manufacturing on the supply side, so it omits how the demand side would respond to the imposed border carbon adjustment as the cost would increase accordingly as shown in Fig. S11b. In detail, an increase in EV cost and consequently the price may stimulate the demand for conventional internal combustion engine vehicles and consequently deteriorate the efforts to reduce climate change impacts. Moreover, opponents of border carbon adjustment view it as a protectionist and discriminatory policy measure to the developing countries that export cheaper but more carbon-intensive products (52,53).

Fig. S11
Comparison of break-even border carbon adjustments to carbon capture cost reported by previous studies. a, Break-even border carbon adjustments for the US on LIBs imported from major Northeast Asian economies and EU. b, Total cost of EV batteries considering the manufacturing cost and the break-even border carbon adjustment rate based on the baseline carbon price of $0/t CO2. c, Carbon capture costs reported by existing studies. Finkenrath did not include the carbon storage cost (96). Acronyms: China (CN); the United States (US); Europe (RER); European Union (EU); Organization for Economic Co-operation and Development (OECD).

Sensitivity analysis
In this section, sensitivity analysis results on the LIB manufacturing cost, carbon footprint, cumulative energy demand (CED), mineral resource scarcity, and break-even border carbon adjustment are provided. Sensitivity analysis of the container shipping rate shows that LIB manufacturing cost can be increased by up to 1.5% with a 166% higher container shipping rate (Fig. S12). Sensitivity analysis of temporal variation in the power grid implies that the clean energy transitions from 2019 to 2050 can reduce the carbon footprint by up to 5% and CED by up to 3% for the US manufacturing scenario, as shown in Fig. S13-SFig. S14. For the ally-shoring scenario, the reduction in carbon footprint and CED are up to 7% and 4%, respectively. The sensitivity analysis result of spatial variation in the power grid suggests that there will be up to 11% and 7% reduction in the carbon footprint and CED if the LIB is produced in the region with the least carbon-and energy-intensive power grid. The result suggests that relocating the midstream and downstream automotive LIB value chain in the Northeast Power Coordinating Council (NPCC) and Western Electricity Coordinating Council (WECC) region of the US will be the least carbonand energy-intensive by 2050, respectively. On the contrary, the power grid in the Reliability First Corporation (RFC) region of the US will be the most carbon-and energy-intensive in 2050. Fig.  S15-SFig. S17 demonstrate the sensitivity analysis results of the recycled content for valuable metals on the carbon footprint, CED, and mineral resource scarcity of the US EV fleet case, US manufacturing scenario, and ally-shoring scenario. For the US EV fleet, using 100% secondary nickel would reduce the carbon footprint by 26-37%, CED by 23-33%, and mineral resource scarcity by 27-43%, suggesting a greater emission reduction benefit compared to adopting 100% recycled cobalt, aluminum, copper, or steel. Among the three scenarios, the US manufacturing scenario achieves the most reduction benefits in the carbon footprint and CED, while the allyshoring scenario can mitigate the most surplus ore potential by adopting 100% secondary nickel. By adopting 100% secondary aluminum for manufacturing automotive LIBs, both the US EV fleet case and ally-shoring scenario have the potential to achieve around 10% reduction in the carbon footprint and CED. Fig. S18 depicts the worst-and best-case carbon price for the sensitivity analysis on break-even border carbon adjustments. The results (Fig. S19-SFig. S21) imply limited impacts of variation in existing carbon price on the break-even carbon border adjustment, compared to the more influential battery pack energy capacity, battery chemistry, and production regions. Table S26 provides the material cost of cathode active material production for sensitivity analysis on nickel supply from Russia and domestic sourcing of critical battery materials. The record-high nickel price during the recent Russia-Ukraine war is adopted as the worst case for the US manufacturing and ally-shoring scenarios without nickel supply from Russia. For the domestic sourcing of critical battery materials, we consider the unit prices of NiSO4 and CoSO4 exported from Canada to the US as proxies for the prices of domestic originated metals; in terms of the lithium price, we calculate the weighted sum of the lithium price exported from Canada to the US and the projected lithium price considering future demand growth and potential supply from geothermal brine given the respective production capacities of lithium extraction from conventional and geothermal brines.  shows that both the carbon footprint and CED are reduced gradually from 2019 to 2050. For the US manufacturing scenario, the reduction in carbon footprint varies from 12.1% for an average EV battery manufactured in 2019 to 15.0% for its counterpart manufactured in 2050. Similarly, the reduction in CED varies from 2.9% to 5.4%. Moreover, for the ally-shoring scenario, the reduction in carbon footprint varies from 7.9% for an average EV battery manufactured in 2019 to 14.5% for its counterpart manufactured in 2050. Similarly, the reduction in CED varies from 3.6% to 7.2%. Compared to the US manufacturing scenario, the ally-shoring scenario achieves more reduction in both carbon footprint and CED from varying the power grid from 2019 to 2050.         Table S27 Total cost of EV batteries considering the manufacturing cost and the break-even border carbon adjustment rate based on the baseline carbon price of $0/t CO 2 .