EV Makers Switch to In-house Chip Production Amid Shortage
EV companies are making chip design an in-house matter—regaining supply chain control and reducing reliance on third-party manufacturers.
Many electric vehicle (EV) manufacturers are switching to in-house chip designs to reduce their reliance on the global semiconductor supply chain. In fact, Gartner estimates that approximately half of the top 10 automotive OEMs will design their own chips by 2025.
After COVID-19-related slowdowns, EV makers have come to grips with their reliance on third-party silicon and are looking to change their production model to prevent global chip shortages from impacting their timelines in the future.
A GM/Cruise autonomous Chevy Bolt using in-house custom chip designs to accomplish self-driving tasks. Image used courtesy of Cruise
In-house production aligns well with the legislative goals of the Chips and Science Act (CHIPS+ Act) recently passed, which aims to buoy domestic chip production. Even so, this switch in sourcing yields both benefits and tradeoffs.
The Semiconductor Drought
The COVID-19 pandemic dealt a massive blow to the global semiconductor market. Work-from-home requirements caused many material refineries, processing plants, and semiconductor foundries to face an in-person labor shortage, reducing the amount of supply available while demand increased near the end of 2020. As such, the number of available chips has plummeted, increasing lead times and creating design constraints for vehicle manufacturers.
A graph showing a history of lead times for commercial chips. As the lead time grows, engineers cannot reliably source components for their designs, increasing production time. Image used courtesy of Forbes
Many different design sectors, including the automotive industry, have typically relied on chips produced by third-party manufacturers such as NVIDIA or Texas Instruments. This business model helps push back hefty R&D costs and allows for innovation at the application level. However, with automotive manufacturers such as Volkswagen believing that demand will outpace supply until 2024, a paradigm shift for the automotive industry may provide a better solution than simply weathering the storm.
The OEM-Foundry Direct Model
Instead of using commercial chips in new designs, numerous EV manufacturers such as GM, Tesla, and China’s Nio hope to bring chip design back in-house. The benefits of in-house chip design are numerous, with the greatest being the ability to design exactly to spec with maximum efficiency. GM’s autonomous driving unit Cruise, for example, produced a board with 70% less mass, 60% less power use, and 90% reduced cost thanks to in-house chip design. Using advanced materials such as silicon carbide or gallium nitride also opens the door to even greater performance, as highlighted by Tesla’s improved energy efficiency.
This performance boost is not due to comparatively outstanding chip design capabilities in the auto industry but rather a smaller list of end-use cases. In an automotive setting, chip designers must design their devices to accomplish only the task at hand, whereas dedicated chip manufacturers must focus on more general use cases to appeal to a broader market, potentially sacrificing performance.
A plate of custom-designed Dune chips developed by GM’s autonomous vehicle unit Cruise. Image used courtesy of Cruise via Reuters
In-house chip design does come with its own tradeoffs, however. Namely, R&D and initial setup costs associated with custom chip design can be quite high. To regain these initial investments, GM's Cruise plans to scale up the production of vehicles using multiple custom chips.
Scaling Up Custom Chip Design
The CHIPS+ Act, which bolsters domestic semiconductor manufacturing and reduces reliance on foreign manufacturers, is already affecting EV momentum. $2 billion of the $52.7 billion bill has been earmarked for legacy chips used in automobiles and defense systems.
Although fully autonomous vehicles may still be years in the making, custom chip design provides greater computational headroom for advanced self-driving algorithms and tremendous savings in overall cost.
