The core obstacle to edge AI large model deployment—the memory wall bottleneck—is being broken by a cross-industry player from the textile sector. Toread's subsidiary Shanghai Tongtu Semiconductor recently signed a strategic agreement with Horizon Robotics, a leading domestic automotive-grade AI chip company, to jointly develop large model bandwidth compression technology, directly targeting the transmission bottleneck between computing power and storage in edge scenarios such as autonomous driving and robotics.
Technology Breakthrough: Full-Stack Compression Unlocks Edge Deployment
Running large models on edge devices consumes enormous bandwidth resources due to the movement of massive parameters and intermediate data, a widely recognized industry pain point. Tongtu has built a full-stack compression system covering three layers: input data, intermediate data, and model parameters, forming a systematic technical barrier. Specifically, for high-bandwidth input data like video, the compression ratio reaches 5x; for intermediate data such as KV Cache generated during inference, it reduces by an additional 50% over Google's TurboQuant algorithm; for billion-parameter models, it achieves 30% parameter compression without retraining.
Horizon Robotics, China's first company to achieve mass production of automotive-grade AI chips, reported 2025 revenue of 3.758 billion yuan, up 57.7% year-on-year. In 2026, it launched the 'Starry 6P' cabin-driving integrated chip and the 'Kakaxia' operating system. In this collaboration, Tongtu's full-stack compression technology will directly integrate with Horizon's intelligent driving computing platform to optimize inference bandwidth and memory usage.
Performance Validation: Chip First Curve Grows Over 200%
Toread's Q1 2026 report shows revenue of 496 million yuan, up 39.41% year-on-year; net profit of 56.4457 million yuan, up 14.44%; and non-GAAP net profit of 63.9395 million yuan, up 32.36%. Chip business revenue reached 179 million yuan, surging 206.92% year-on-year, accounting for 36.09% of total revenue. Analysts from Huaxi Securities maintain a 'buy' rating, forecasting net profit of 237 million, 369 million, and 454 million yuan for 2025-2027. CICC raised its 2026 revenue forecast by 29% to 1.952 billion yuan.
Strategic Depth: Dual-Business Synergy and the Edge AI Blue Ocean
Since establishing the 'outdoor + chip' dual-business strategy in 2021, Toread has acquired five chip-related companies, building a product matrix covering 'perception-interaction-display'. Tongtu's compression technology not only drives chip business growth but also supports Toread's second-generation smart exoskeleton and outdoor smart shoes. During the 618 shopping festival, the Crest C3 smart exoskeleton sold over 100 units in a single day on Tmall, weighing only 1.8 kg with a 10:1 thrust-to-weight ratio.
Toread's vice president Wang Hongjian predicted at the 2026 Zhongguancun Forum that China's edge AI market will grow from 600 billion yuan in 2026 to 2.1 trillion yuan by 2030. Tongtu's compression capability can be applied to eight major scenarios including personal devices, smart spaces, connected vehicles, industrial internet, embodied robots, and AR/VR wearables.