Near-term pressured by OEM in-house chips, but the long-term intelligent driving chip and IP licensing thesis remains

Overview

This report discusses the reasons behind Horizon Robotics’ recent share price pressure: the market is concerned about OEMs accelerating in-house development of intelligent driving chips, as well as 2026 delivery targets against a backdrop of weak auto sales. Bernstein believes these concerns create short-term sentiment pressure but do not change Horizon’s long-term competitive thesis in China’s intelligent driving chips, HSD software ecosystem, and IP licensing model.

Core views

The core view is that OEM in-house chips will not completely exclude Horizon from the value chain. Even if some customers develop their own chips, they may still license Horizon’s BPU/IP and pay royalties; at the same time, most OEMs lack sufficient scale, software accumulation, and AI/silicon talent, so third-party intelligent driving chips still have a sizable market. The report maintains an Outperform rating and sees 2026 design wins and second-half volume ramp-up as key observation points.

Analysis framework

The report is mainly based on management fireside chat and Q&A information, analyzing Horizon’s medium- to long-term investment thesis from the perspectives of business model, supply-chain landscape, feasibility of OEM in-house development, per-vehicle economics, overseas opportunities, competitive landscape, and valuation methodology.

Methodology notes

• Business ModelARM-style IP licensing + royalty model

  Customers can license Horizon’s underlying BPU/IP to develop their own chips and pay royalties after mass production.

  This model converts part of direct chip sales revenue into licensing fees and royalties, with relatively high gross margins, though the report also notes that absolute gross profit per vehicle may be lower than under the direct sales model.

• Valuation methodsEV/Sales valuation method

  Estimate the target price based on 2BF sales and an EV/Sales multiple.

  Bernstein gives a HKD 10 target price based on 10.3x EV/Sales, corresponding to RMB 11.5Bn of 2BF (2H27-1H28) sales.

• Competitive AnalysisThree types of supply-chain landscape

  The industry is divided into three supply-chain types: NVIDIA + software partners, Horizon + software, and OEM in-house chips + in-house software.

  The report believes BPU determines AI performance, and algorithm ecosystems are difficult to migrate across platforms, so self-developed chips based on Horizon BPU may still remain within the Horizon ecosystem.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Horizon Robotics（09660.HK）

  Covered company in the report; a Chinese supplier of intelligent driving chips and ADAS solutions.

  Strengths

  Has chips, BPU, HSD software, and customer ecosystem; its third-party supplier positioning suits most OEMs; it can monetize through both direct sales and IP licensing.

  Weaknesses

  Dragged in the short term by OEM in-house chips and weak industry sales; per-vehicle revenue may decline under the IP licensing model; software delivery resources are limited.

  Comparison

  The report compares it with NVIDIA, Qualcomm, Huawei, Mobileye, and OEM in-house development paths, and believes China’s local substitution trend and software ecosystem are favorable to Horizon.

  Risks

  Slower-than-expected intelligent driving penetration, higher-than-expected OEM in-house share, or delays in shipments or deployment by major customers.

Key data

• RatingOutperformBernstein maintains its Outperform rating on Horizon Robotics.
• Target priceHKD 10.00The Price Target disclosed in the report.
• Valuation basis10.3x EV/Sales; RMB 11.5Bn 2BF salesUsed to support the HKD 10 target price.
• 2026 growth paceLow in the first half, high in the second half; about 60% annual growth targetManagement said the first half is below this target but still high growth, not a decline.
• Direct chip sales gross margin40-50%Management said this is the target gross margin at the hardware level.
• IP/royalty gross profit>90% or close to pure profitThe report says royalty gross margin is high, similar to the ARM model.
• China passenger vehicle marketAbout 25Mn unitsUsed by management to estimate the serviceable market and long-tail opportunity.
• Estimated uncovered marketAbout 8Mn units, about 30-40%Includes Huawei and other OEM markets with self-developed chips.
• Horizon target long-tail share>50%Targeting the remaining roughly 60% long-tail market, competing with NVIDIA, Qualcomm, and new entrants.
• Pure overseas mass-production timingNot earlier than 2028-2029In the short term, overseas contribution comes more from China-made, globally sold products.

Impact & implications

The investment implication is that the share price may remain weak in the short term due to OEM in-house chips and weak auto sales, but if Horizon secures high-quality new design wins in 2H26/2027, ramps HSD, and proves the economics of IP licensing, the market may reassess its serviceable market and profit pool.

Risks

• Growth in intelligent driving penetration is slower than expected.
• OEM in-house chips gain a higher share than expected.
• Shipments or deployment by major customers are delayed, causing revenue growth to come in below expectations.
• If chip specification iteration slows, OEMs may prefer licensing underlying IP and developing upper-layer software themselves.
• Weak demand in the auto industry may affect the 2026 delivery pace.

What to watch

• The number and quality of new design wins around 2H26 and 2027.
• Mass-production progress of Journey 6M and Journey 6P and HSD installation volume.
• Whether OEM in-house chips adopt Horizon BPU/IP and the corresponding royalty monetization.
• Overseas ADAS orders, especially contributions from China-made, globally sold vehicle models.
• Share changes in China’s intelligent driving market for Huawei, NVIDIA, Qualcomm, Mobileye, and other new entrants.