Intel senior vice president Cho Min-sung announced the roadmap for the upcoming Razor Lake-AX system-on-chip, scheduled for release in 2028. The company expects the new architecture to deliver 700 TOPS of AI performance, which is approximately 3.88 times the capability of the current Panther Lake platform.
The Razor Lake-AX chip is designed for high-performance applications, specifically targeting the physical AI sector, including robotics and automation. Intel plans to integrate a unified large-capacity memory pool, potentially utilizing a Memory on Package design similar to the one implemented in the Lunar Lake series.
The 700 TOPS target for Razor Lake-AX positions it against competitors like the Nvidia Jetson AGX Orin, which offers 275 TOPS. Intel also plans to launch the Nova Lake platform next year, which is expected to reach the same 700 TOPS performance milestone as the company shifts focus toward specialized AI hardware.
The industry landscape remains competitive as Intel faces pressure from Nvidia, which provides the Jetson Thor T2000 and T3000 platforms with up to 865 TFLOPS of FP4 AI performance, and AMD with its Ryzen AI MAX+ Halo series. Intel aims to expand its presence in the physical AI market through these upcoming architectures.
