Although in-car computing is not the highest end of chips, these are not for learning, but for execution of what was learned by way higher end server systems.
Still this shows that there is market share to be lost for American companies in the mass market of AI clients. And the logical step is to evolve these chips to server grade 5 to 10 years into the future.
The irony is that the Chinese industry actually wanted to use leading American chips, but USA has prevented that, forcing China to develop as much and as fast as they can to replace American chips. It was always the Chinese government that tried to encourage local Chinese development and use of AI chips, while most of the industry was reluctant.
But recent developments have made it clear to the Chinese industry that they have to have a homegrown option, because they can’t trust USA to deliver, despite China pays full price, and it helps American companies to stay ahead to supply to China too, as it helps pay for the necessary R&D.