Black Sesame Intelligence Unveils Automotive-Grade AI Chips and Embodied Intelligence Platform at WAIC 2026

On July 17, 2026, the 2026 World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance opened in Shanghai. At the event, Black Sesame Technologies showcased its full lineup of core chips, proprietary algorithms, and embodied intelligence applications spanning advanced intelligent driving, integrated cockpit-driving systems, and robotics. In the domain of intelligent driving chips, the company exhibited the Huashan A2000 family, featuring a peak single-chip computing power of up to 1,000 TOPS and supporting multi-chip collaborative scaling to meet the full-scenario computational demands of Level 4 autonomous driving. On-site, Black Sesame also demonstrated its FAD Tianyan L3 autonomous driving domain controller powered by dual A2000 chips, highlighting two key capabilities: one-touch urban and highway NOA (Navigate on Autopilot) enabled by its proprietary end-to-end algorithm, and real-time on-device interactive inference of the Qwen VLM large vision-language model running on the Huashan A2000 platform. Built on Black Sesame’s self-developed Jiushao NPU architecture, the chip supports mixed-precision computation—including INT4, INT8, FP8, FP16, and FP32—without requiring quantization, delivering lower power consumption and higher throughput at equivalent compute performance. Additionally, it integrates seven core technologies: a high-performance in-house ISP, a triad functional safety system, high-speed inter-chip interconnect technology, and the Shanhai AI toolchain, among others. Together, these enable a tiered coverage from intelligent cockpit agents to L3/L4 advanced autonomous driving, with further extensibility into embodied intelligence and broader edge AI applications. In the field of embodied intelligence, Black Sesame launched the SesameX multidimensional embodied intelligence computing platform, repurposing automotive-grade AI compute for robotics scenarios. The platform comprises three self-developed core modules—Kalos, Aura, and Liora—with computing power ranging from 48 TOPS to nearly 600 TOPS. With edge-side AI compute as its core focus, Black Sesame is driving the extension of automotive-grade technologies into diverse application domains, covering the full perception-decision-execution pipeline in the physical world. The company plans to continue deepening its expertise in edge AI inference chips, strengthening ecosystem collaboration, and accelerating the expansion of its compute capabilities from automotive into a wider range of industries.