GM Uses AI Model and Mini Factory to Halve Development and Testing Time for Cheaper EV Battery Cells
GM is using an AI model and a new mini factory to build and test cheaper EV battery cells in half the time.
The facility, located at the Battery Cell Development Center (BCDC) in Warren, Michigan, covers approximately 500,000 square feet and represents an investment of about $900 million. It serves as a bridge between laboratory prototypes and mass production, capable of producing around 2,500 cells per day.
The AI model is based on the company's 10 years of battery data, using large language models and physical simulations (with over 150 million CPU hours just for LMR chemistry) to predict outcomes of dozens of chemical and process variations. A digital twin replicates facility details to shorten debugging and ramp-up times, with the overall end-to-end process expected to be halved, previously estimated to shorten new chemistry development by up to a year.
The focus chemistry is lithium manganese rich (LMR) prismatic cells, which have energy densities close to current NMC but costs closer to LFP. The goal is to achieve over 400 miles of range for full-size trucks and SUVs while reducing battery costs by thousands of dollars per vehicle, with mass production planned to start in 2028 at the Ultium plant in Tennessee.
Simulations have halved the slurry mixing time without affecting quality, and testing operating costs are significantly lower than those of full-scale plants. The facility will also support other chemistries, such as sodium-ion.
The initiative is driven by the reduction of EV costs and supply chain autonomy, with resources shifting from traditional overseas iterations to internal AI simulations and pilot production lines. Beneficiaries include GM, which accelerates chemical validation and cost control, while traditional battery development paths relying on long-cycle external suppliers face pressure.
The BCDC is expected to produce its first batch of LMR test cells in the fourth quarter of this year, strengthening the closed-loop capability of U.S. battery development from R&D to mass production.
Source: Public Information
ABAB AI Insight
Since establishing the Wallace Battery Cell Innovation Center in 2022, GM has continuously reduced its reliance on Asian suppliers through vertical integration. Historical paths include a joint venture with LG Energy Solution for Ultium and over $5 billion invested in U.S. battery operations, with the recent $900 million BCDC filling the gap between laboratory and mass production.
In terms of capital strategy, the company is investing computational resources and pilot production lines into AI-driven virtualization and digital twins, shifting resources from physical prototype iterations to rapid simulations based on historical data. The motivation is to compress the development cycle for new chemistries (like LMR) by half and bring them to market a year earlier, while also lowering battery costs per vehicle to enhance EV competitiveness.
Similar cases include other automakers using digital twins to accelerate battery or engine development. Currently, GM is in an expansion phase, moving from high-nickel chemistries to lower-cost options like LMR and LFP, while challenging China's dominance in LFP.
This fundamentally represents a technological substitution and industrial chain reconstruction: battery development is shifting from long-cycle physical testing to AI prediction and internal piloting. The mechanism is that data accumulation allows virtualization to replace some physical iterations, transferring pricing power from external cell suppliers to automakers with simulation capabilities and vertical production lines, making cheaper, high-range batteries an attainable path rather than relying on imports.
ABAB News · Cognitive Law
AI simulation halves time, pilot factory connects R&D and mass production
Ten years of data accumulation, chemical changes predicted in days
Battery costs decrease, vertical integration is key