Toread's Q1 2026 earnings report reveals a striking figure: chip business revenue reached RMB 179 million, up 206.92% year-on-year, now accounting for 36.09% of total revenue—up from less than 15% two years ago. The outdoor gear company is rewriting its valuation narrative with hard tech.

Technology Breakthrough: LLM Compression Tackles Edge Computing Bottlenecks

Deploying large language models on edge devices—phones, cars, robots—has long been constrained by bandwidth and memory limits. During inference, massive parameter and intermediate data transfers between chip and memory consume bandwidth, throttling computing power. Shanghai Tongtu Semiconductor and Horizon Robotics recently signed a strategic agreement to co-develop LLM bandwidth compression, a systematic solution to this pain point.

Tongtu has built a three-tier compression system: at the input level, it compresses video and other large data to one-fifth of its original size; at the intermediate data level, it achieves an additional 50% compression beyond Google's results; at the model parameter level, it compresses trillion-parameter models by 30%, all without retraining and plug-and-play.

At the 2026 Zhongguancun Forum, Toread Vice President and Tongtu CEO Wang Hongjian stated that Tongtu's full-stack compression aligns with Google's TurboQuant algorithm in core pain points, technology path, and industry direction, confirming that 'using compression to free computing power' is a deterministic industry trend.

Industrial Deployment: Partnership with Horizon Accelerates Commercial Closure

Horizon Robotics, a leading provider of automotive-grade AI chips and smart driving solutions, reported 2025 full-year revenue of RMB 3.758 billion, up 57.7% YoY. In 2026, Horizon launched China's first cockpit-driving integrated vehicle smart chip 'Starry 6P' and the vehicle smart OS 'Kakaxia.' The collaboration with Tongtu means LLM compression will first be validated on automotive-grade chips.

For buyers, this signals two things: bandwidth costs for edge AI deployments are likely to drop significantly, and inference efficiency on automotive chips will see real improvements. For latency-sensitive applications like autonomous driving and robotics, performance gains from compression could directly enhance product competitiveness.

Dual-Core Synergy: Chip Growth Drives End-to-End Self-Reliance

Since establishing its 'outdoor + chip' dual-core strategy in 2021, Toread has acquired five chip companies, building a product matrix covering perception, interaction, and display. The chip business's rapid growth not only provides a second growth curve but also supports smart equipment self-reliance at the underlying tech level.

During the 618 shopping festival, Toread's self-developed smart exoskeleton Crest C3 began pre-sales, achieving over 100 units sold in a single day on Tmall. Weighing only 1.8 kg with a 10:1 thrust-to-weight ratio, the product's core support comes from Tongtu's full-stack compression—end-to-end self-reliance from chip to end product is accelerating.

Wang Hongjian predicted at the Zhongguancun Forum that China's edge AI market could grow from RMB 600 billion in 2026 to RMB 2.1 trillion by 2030. Tongtu's compression capability can be applied across eight scenarios, including personal devices, smart spaces, in-vehicle systems, industrial internet, embodied robots, and AR/VR wearables.

Performance Validation: Institutional Upgrades Reflect Market Confidence

Q1 2026 results show Toread's total revenue at RMB 496 million, up 39.41% YoY; net profit attributable to parent at RMB 56.45 million, up 14.44%; and non-GAAP net profit at RMB 63.94 million, up 32.36%. Chip revenue surged 206.92% YoY, now 36.09% of total, forming a 'chip high-growth, outdoor steady-growth' pattern.

On June 17, 11 institutions surveyed Toread. Huaxi Securities maintained a 'buy' rating, forecasting 2025-2027 net profit of RMB 237/369/454 million. CICC raised 2026 revenue forecast by 29% to RMB 1.952 billion and EPS by 60% to RMB 0.15. The consensus reflects market recognition of Toread's transformation from an outdoor brand to a dual-drive 'edge AI chip + smart equipment' player.

Practical Recommendations

For Buyers - Monitor Tongtu's compression performance test data on Horizon chips, especially inference latency and bandwidth savings, which directly impact cost structures of autonomous driving and robotics products. - For enterprises planning edge LLM deployments, Tongtu's plug-and-play parameter compression can reduce hardware upgrade costs; prioritize compatibility assessment with existing models. - Toread's Crest C3 pre-sale data serves as a market acceptance indicator for consumer-grade edge AI products; buyers should track supply chain dynamics in this category.

For Foreign Trade Companies - Edge AI chips and compression tech are focal points of international tech competition; companies with relevant capabilities can leverage this to enter overseas autonomous driving and robotics supply chains. - Toread and Horizon's collaboration model (chip + algorithm co-development) can serve as a reference for tech export; foreign trade firms should explore similar 'tech licensing + joint development' business models. - China's edge AI market forecast (RMB 600 billion in 2026 to RMB 2.1 trillion by 2030) suggests parallel demand growth overseas for high-performance, low-bandwidth edge solutions; foreign trade firms should prepare product certifications and channel development early.

资料来源:中国纺织报查看原始资料
⚙
Manage your textile business with Jenny ERP
Sample · Order · Customer · Inventory · Production tracking — built for fabric mills and trading companies.
Try Free