Apple Car concept by Vanarama
Apple is working with an unspecified South Korean company to develop an autopilot processor for the much-rumoured “Apple Car.”
According to a report in Korean supply chain news site The Elec (via MacRumors), the Cupertino-based company is apparently developing the self-driving chip for its car project in partnership with a Korean OSAT (outsourced semiconductor package test) company just as it did for the M1 processor.
An OSAT firm is usually a component supplier that offers clients services such as semiconductor development, assembly, packaging, and testing. The report claims that Apple’s chip for autonomous driving capability is expected to resemble those used by Tesla. Such chips usually integrate a neural processing unit (NPU), CPU, GPU, RAM, and a camera interface for AI computation tasks.
For the record, Tesla’s Autopilot and the Full Self-Driving Beta system rely on a tailored 14nm Samsung Exynos chipset, memory, and cameras integration, and another Korean company assembled them for it. Apple is apparently taking the same route.
That being the case, Korean companies seem to be the go-to solutions in the manufacture and assembling supply chain of chips.
The report notes that Apple car’s autonomous driving set that the Korean OSAT is handling is expected to sport an NPU component, too. The NPU will be a key hub for the artificial intelligence calculations needed to monitor and manage the Apple car’s reactions to road conditions and driver input.
Apple Korea, a regional office based in South Korea, reportedly handles the partnership. “The subsidiary was given the bill of material (BOM) rights for the project and chose the South Korean OSAT firm accordingly,” the report says.
The report’s unnamed sources claim that the project began in 2021 and is expected to complete in 2023.
Other Apple partners likely will become involved with Cupertino’s car project somewhere along the way. Taiwan Semiconductor Manufacturing Company, which manufactures Apple’s A-series and M-series chipsets, seems like a likely participant.