Turing moves 10% of its AI training workload to AMD chips, signaling potential competition for Nvidia in the AI hardware market.
Japanese self-driving startup Turing has shifted 10% of its AI training workload from Nvidia to AMD chips, marking its first diversification away from Nvidia hardware. The move follows AMD Ventures taking an undisclosed stake in Turing, which has yet to commercialize its technology but raised $79 million last year at a $600 million valuation.
Turing previously relied entirely on Nvidia for training and inference since its launch five years ago. Executives cited cost savings and supply diversification as key reasons for the shift, noting AMD’s proximity to Nvidia’s headquarters in Santa Clara offered a viable alternative. The company aims to deploy robotaxi fleets by 2028 and partnered with auto supplier Denso last year to advance production.
AMD shares rose roughly 3% in premarket trading on the news, reflecting investor enthusiasm for any sign of competition against Nvidia’s dominance in AI hardware.