Toread's Edge AI Strategy Takes Shape: Large Model Compression Moves from R&D to Industrial Collaboration

China's edge AI market is accelerating from concept validation to large-scale deployment, with bandwidth and memory bottlenecks in large model deployment on edge devices becoming critical variables. Shanghai Tongtu Semiconductor, a subsidiary of Toread, has signed a strategic agreement with Horizon Robotics to jointly develop large model bandwidth compression technology, directly targeting this core pain point. This move marks the official transition of Tongtu's full-stack compression technology from R&D to industrial collaborative development, providing a deployable technical solution for efficient edge AI deployment.\n\n## Technology Breakthrough: Compression Targets Memory Wall Bottlenecks\n\nWhen large models run on edge devices like automotive-grade chips, the transfer of massive parameters and intermediate data consumes significant bandwidth, creating the so-called 'memory wall' bottleneck. Tongtu's full-stack compression system optimizes at three levels: at the input data level, it compresses high-capacity data like video to one-fifth of its original size, reducing edge input bandwidth pressure; at the intermediate data level, it further compresses by 50% on top of Google's TurboQuant algorithm, alleviating memory usage during inference; at the model parameter level, it achieves 30% compression for trillion-parameter models without retraining, offering plug-and-play functionality.\n\nHorizon Robotics, a leading domestic supplier of automotive-grade AI chips, reported full-year 2025 revenue of 3.758 billion yuan, up 57.7% year-on-year, and launched the cockpit-driving integrated chip 'Starry 6P' and vehicle intelligent operating system in 2026. The partnership integrates Tongtu's compression technology with Horizon's computing platform to jointly optimize inference efficiency of large models on automotive-grade chips. This is not just a complement of technical solutions but a substantive validation of the industry direction 'using compression to free up computing power.'\n\n## Performance Confirmation: Chip Business Becomes Primary Growth Driver\n\nThe confidence in technology layout is ultimately supported by financial data. Toread's Q1 2026 report shows revenue of 496 million yuan, up 39.41% year-on-year, and net profit attributable to shareholders of 56.4457 million yuan, up 14.44%. The chip business stood out: Q1 revenue reached 179 million yuan, surging 206.92% year-on-year, accounting for 36.09% of total revenue. In contrast, the outdoor business grew only 8.02% in the same period, making the chip business the primary growth driver.\n\nInstitutional signals are also positive. On June 17, Toread received 11 institutional investor surveys. Huaxi Securities maintained a 'buy' rating, forecasting net profits 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 and EPS forecast by 60% to 0.15 yuan. These data indicate that market recognition of Toread's 'outdoor + chip' dual-business strategy is shifting from expectations to performance realization.\n\n## Strategic Depth: Full-Chain Self-Sufficiency from Chips to End Products\n\nSince establishing the dual-business strategy in 2021, Toread has acquired five chip-related companies, building a chip product matrix covering the full chain of 'perception-interaction-display.' Tongtu's full-stack compression technology is not only an independent growth pole for the chip business but also provides core technical support for Toread's second-generation intelligent exoskeleton and outdoor smart shoes. During the 618 shopping festival, the self-developed intelligent exoskeleton Crest C3 began pre-sales, achieving daily sales of over 100 units on Tmall, with a total weight of only 1.8 kg and a thrust-to-weight ratio of 10:1.\n\nFull-chain self-sufficiency from underlying chips to end products is accelerating. Tongtu's full-stack compression technology can be widely applied in eight scenarios: personal devices, smart spaces, smart vehicles, industrial internet, embodied robots, and AR/VR wearables. Wang Hongjian, Vice President of Toread, predicted at the 2026 Zhongguancun Forum that China's edge AI market size is expected to grow from 600 billion yuan in 2026 to 2.1 trillion yuan by 2030. When a technology can solve the core bottleneck of the global AI industry, complete joint development with leading companies in real commercial scenarios, and be backed by strong performance data, it is no longer a 'concept' but a real industrial variable.\n\n## Practical Recommendations\n\n### For Buyers\n- Focus on collaborative solutions between edge AI chips and compression technologies, prioritizing products that have passed automotive-grade validation to reduce deployment risks.\n- When procuring smart equipment, evaluate the actual optimization effects of full-stack compression technology on bandwidth and memory to avoid being misled by single indicators.\n- Monitor the iteration pace brought by Toread's dual-business synergy; new products like intelligent exoskeletons may accelerate penetration into industrial and consumer sectors.\n\n### For Foreign Trade Enterprises\n- Edge AI chips and compression technologies are hot topics for overseas clients; prepare technical parameters and case studies in advance to enhance quotation competitiveness.\n- Pay attention to the overseas certification progress of domestic chip companies like Horizon; automotive-grade products need to comply with local regulations when exported.\n- Use product performance data from domestic consumption events like 618 as evidence of technology maturity for overseas clients.