Horizon Faces Short-term Pressure, Long-term Logic Unchanged
AI summary card
Horizon Faces Short-term Pressure, Long-term Logic Unchanged
OEM self-developed chips bring short-term sentiment pressure, but Horizon consolidates its niche through dual models of IP licensing + chip sales, and the long-term growth logic remains undamaged.
- Maintain 'Outperform' rating with target price of HK$10, representing an 84% upside potential from the current stock price
- OEM self-developed chips are not a threat to Horizon, but a business model shift: from selling chips to IP licensing + royalties
- The company adopts a business model similar to ARM, with IP licensing gross margin exceeding 90%, resulting in a better profit structure
- 2026 guidance is 'low in the first half, high in the second half', with full-year still expected to achieve approximately 60% growth
- Mid-to-low-end OEMs still rely on third-party Urban NOA chips, and Horizon occupies the best position in this market
- Overseas ADAS exports are performing well, with pure overseas mass production expected to start in 2028-2029
Report interpretation
Overview
Bernstein released an earnings review report on Horizon Robotics (9660.HK), pointing out that although the stock price has recently been under pressure—mainly due to market concerns about OEMs accelerating self-development of smart driving chips and whether the 2026 delivery target can be achieved against the backdrop of weak auto sales—the company's long-term investment logic remains solid. The institution believes that OEM self-developed chips will not significantly erode Horizon's addressable market, but instead drive its transformation towards a higher-margin IP licensing model, similar to ARM's business model. Therefore, the 'Outperform' rating and HK$10 target price are maintained.
Core views
The short-term pressure faced by Horizon mainly comes from two aspects: first, leading OEMs such as BYD, Li Auto, XPeng, NIO, Tesla, and Huawei are all deploying self-developed chips; second, domestic auto sales in the first half of 2026 fell by about 20% year-on-year, raising doubts about the full-year delivery target. However, management clearly stated that the success threshold for OEM self-developed chips is extremely high, requiring simultaneous scale effect (annual production exceeding 1.5 million units), deep software capabilities (algorithm and engineering practice), and true tech company genes, with only a few companies likely to succeed. More importantly, even if OEMs choose to self-develop chips, they often develop based on Horizon's BPU (AI core) and continue to use its HSD smart driving software ecosystem. Volkswagen's adoption of Horizon's IP licensing model through the CARIZON joint venture project is an example. Under this model, Horizon charges a one-time IP licensing fee (gross margin nearly 100%) and royalties after mass production (gross margin over 90%). Although per-vehicle revenue decreases, overall profitability is better, and customer stickiness is stronger. The company's 2026 guidance is 'low in the first half, high in the second half', relative to its own annualized growth target of about 60%. Growth in the first half will be lower than this level, but not negative growth. In the second half, with the volume of new designated projects, growth will significantly rebound. Mid-to-low-end OEMs have strong demand for third-party Urban NOA chips, and Horizon's Journey 6M and 6P product portfolio can effectively cover this market. Additionally, the company is steadily advancing ADAS exports in overseas markets through the 'China production, global sales' model. Currently, only Mobileye and Horizon have the capability to serve global automakers on a large scale.
Analysis framework
Bernstein's analysis revolves around 'short-term disturbances vs. long-term value'. First, it identifies the core of market concerns (OEM self-developed chips), then argues through dissecting industry barriers (scale, software, technology iteration speed) that self-development is difficult to popularize. Next, it focuses on explaining the evolution of Horizon's business model—from pure chip sales to 'chip + IP licensing' dual drive, analogous to ARM's model to illustrate the improvement in profit quality. Finally, combined with the 2026 delivery rhythm, mid-to-low-end market demand, and overseas progress, it verifies the sustainability of the long-term growth path.
Methodology notes
Competition in the smart driving chip industry depends not only on hardware performance but also on software ecosystem and algorithm adaptation capability
The report points out that since BPU defines AI performance, the algorithm layer naturally grows within the Horizon ecosystem, making it extremely difficult to移植 algorithms from other ecosystems (e.g., NVIDIA) to BPU. This means that even if OEMs self-develop chips, it is hard to脱离 Horizon's software system, thus ensuring Horizon's core position in the industry chain.
Valuation based on forward sales EV/Sales
The report uses a 10.3x enterprise value to sales (EV/Sales) ratio, benchmarking against Horizon's expected sales of RMB 11.5 billion in the second half of 2027 to the first half of 2028 (2BF), resulting in a target price of HK$10, reflecting pricing for远期成长性.
Horizon's competitive advantage lies in its integrated hardware-software capabilities and the ecosystem lock-in effect of BPU IP
The report emphasizes that software is the core of smart driving, and hardware is just the carrier. Horizon provides an integrated SoC and algorithm solution, building difficult-to-replicate engineering know-how (analogous to CATL's manufacturing process), forming a technological moat.
Key data
- Target Price10.00 HKDBased on 10.3x EV/Sales valuation
- Current Price5.43 HKDAs of June 1, 2026
- Implied Upside Potential84%Target price increase from current price
- 2026 Growth Target~60%Annualized growth target, guidance 'low in the first half, high in the second half'
- IP Licensing Gross Margin>90%Royalty portion is接近纯利润
- Direct Chip Sales Gross Margin40-50%Target managed at the hardware level
Impact & implications
The report believes that the market is overly concerned about the impact of OEM self-developed chips on Horizon, which actually verifies Horizon's technological leadership and opens up a new high-profit business model for it. In the short term, the stock price may remain low due to sentiment factors, but with the landing of new designated projects in the second half and the ramp-up of delivery volume, fundamentals will gradually materialize. In the long run, Horizon aims to capture over 50% share in China's approximately 60% 'long-tail market' (non-leading OEM self-development camp), with broad growth space.
Risks
- Smart driving penetration rate increases slower than expected
- OEM self-developed chip progress exceeds expectations, capturing more market share
- Shipment or deployment delays to major customers, leading to pressure on revenue growth
What to watch
- Quantity and quality of new designated projects in the second half of 2026
- Penetration of Journey 6 series chips in mid-to-low-end OEMs
- Breakthrough progress in pure overseas localization mass production projects (expected 2028-2029)