XPENG Motors has proudly declared the official activation of its third Turing AI chip, a milestone that propels the company into the 'Triple-Chip Era' for its proprietary chip development. This advancement means that production models will transition from a 'dual-chip' to a 'triple-chip' comprehensive configuration. The Turing AI chip, a fully in-house-developed, high-performance computing chip tailored for onboard applications, is poised to commence mass production and integration into vehicles in the third quarter of 2025. Cumulative shipments are anticipated to surpass 200,000 units. From the second quarter of 2026 onwards, all XPENG models will adopt this chip, with an ambitious annual shipment target nearing 1 million units. Presently, the triple-chip solution has undergone thorough validation. The three chips are designated for perception, decision-making, and cross-domain integration tasks, respectively, ensuring computational redundancy and functional hierarchy. Varying chip counts correspond to distinct capability tiers. The essence of this 'onboard superintelligent agent' resides in the cross-domain integration capabilities of the second-generation VLA (Vision-Language-Action) and VLM (Vision-Language-Model). The system boasts three primary capabilities: cross-domain holistic orchestration, native multimodality, and onboard large-scale models. It revolutionizes the cabin voice architecture, enabling voice-controlled vehicle operations and delegating certain L4-level functionalities from Robotaxi systems. At 15:00 on August 27, XPENG will host a Physical AI Sharing and Second-Generation VLA New Version Experience Day. The XPENG G9L will make its global premiere with this iteration, marking the first significant version update of the second-generation VLA since its mass production and rollout in March of this year.