EB SECURITIES released a research report maintaining a "Buy" rating on HORIZONROBOT-W (09660), highlighting the company's key advancements showcased at its recent event. The high-level assisted driving system "HSD" has entered mass production, with activations surpassing 12,000 units within two weeks. The system is expected to expand into the RMB 100,000 urban new energy vehicle (NEV) market. Overseas market expansion is progressing, with mid-to-high-end product shipments likely to increase further. The debut of J6P+HSD is set to enhance both product ASP and brand strength.

Key takeaways from the "2025 Horizon Technology Ecosystem Conference" held in Shenzhen on December 8-9 include:

1. **HSD Mass Production & Expansion into Affordable Urban NOA Market** - HSD, featuring a single-stage end-to-end + VLM architecture, achieves millisecond-level latency and human-like defensive driving. - CEO Yu Kai announced that HSD activations exceeded 12,000 units within two weeks for models like DeepBlue L06 and Chery Exeed ET5. - A cost-effective urban NOA solution is set to penetrate the RMB 100,000 segment, covering 50% of China's passenger car market. Initial partners include Bosch, Zhuoyu, and QCraft.

2. **"HSD Together" Full-Stack Algorithm Services Accelerate J6 Adoption** - Unlike traditional "chip + reference algorithm" models, HSD Together offers a verified "showroom" approach with white/black-box licensing, reducing development costs. - J6 series shipments are expected to accelerate as HSD expands into lower-priced segments.

3. **New "Riemann" BPU Architecture & J7 Targeting Tesla AI5** - The Riemann architecture improves key operator performance by 10x and supports 5x greater LLM efficiency. - J7 chips aim to compete with Tesla’s next-gen AI5, while the fourth-gen compiler "Tiangong Kaiwu" cuts compilation time to minutes.

4. **Open-Source Embodied AI Models for Robotics Ecosystem** - Horizon unveiled HoloMotion (motor control) and HoloBrain (cognitive) open-source models, already adopted by Stanford and Tsinghua.

**Risks**: Slower-than-expected smart vehicle demand, ADAS/AD adoption delays, intensified chip competition, and stock volatility.