OpenAI Poaches XPeng AI Infrastructure Lead with 200-Person Team, Likely to be Restructured

According to Beating, XPeng AI infrastructure head Siyu Lu is departing for OpenAI to focus on embodied AI robotics research. Lu has overseen XPeng's end-to-end AI infrastructure spanning model training, GPU clusters, custom chip compilers, model quantization, and vehicle deployment, managing approximately 200 staff members.

With no replacement yet identified, Lu's team is expected to be restructured into three or more independent teams. The position is critical to XPeng's operations, as the automaker relies on self-developed chips and must maintain its own compiler and inference runtime systems.