Quick Summary
Covering the latest research from top Wall Street investment banks

1Q26 China NOA penetration temporarily fell to 32%, Horizon gained share against the backdrop as its target price was cut

Institution
Bernstein
Date
2026-05-11
Authors
Stacy A. Rasgon, Dien Wang, Alrick Shaw, Arpad von Nemes
Company
Horizon Robotics
Ticker
9660.HK
Industry
Semiconductors
Rating
Horizon Robotics: Outperform; Black Sesame: Underperform; NVDA: Outperform; QCOM: Market-Perform
NeutralLow confidenceIn the short term, weak Chinese passenger and EV sales, higher storage prices, and intensifying L2++ competition dampened sector sentiment, while Horizon continues to gain share in L2+ and L2++, supported by new design wins at BYD and the OTA path, sustaining long-term growth.
AuthorsStacy A. Rasgon, Dien Wang, Alrick Shaw, Arpad von Nemes
Target priceHorizon Robotics HKD 10; Black Sesame HKD 16; NVDA $300; QCOM $140
Asset classesEquity
Business segmentsSmart driving SoC、L1-L2 ADAS、L2+ NOA、L2++ Urban NOA、LiDAR
Research firm divisions/subsidiariesBernstein(Other)

AI summary card

1Q26 China NOA penetration temporarily fell to 32%, Horizon gained share against the backdrop as its target price was cut

Bernstein sees weaker China EV and passenger-car sales in 1Q26 pulling NOA penetration down from 36% in 4Q25 to 32%, but L2++ demand remains more resilient, and Horizon continues to gain share in both L2+ and L2++.

Horizon Robotics maintains Outperform but cuts the target price to HKD 10; Black Sesame maintains Underperform and the HKD 16 target price; NVDA is Outperform with a target of $300; QCOM is Market-Perform with a target of $140.
Smart Driving ChipsNOAL2+L2++Horizon RoboticsBlack SesameNVIDIAChina EV
  • China passenger-car sales in 1Q26 were 4.05 million vehicles, down 18% year-over-year and 36% sequentially; NOA (L2+ and L2++) penetration fell from 36% in 4Q25 to 32%.
  • L2+ penetration declined by 3 percentage points QoQ to 13%, while L2++ declined only 1 percentage point to 19%, showing greater resilience in urban NOA demand.
  • Horizon held up relatively well in the weak sector backdrop, with L2+ share rising 7 percentage points QoQ and L2++ share rising 4 percentage points QoQ to 5%.
  • NVIDIA's L2+ share fell from 45% to 35%, while L2++ remains leading but share declined from 54% to 50%.
  • LiDAR penetration declined from about 17% in 4Q25 to 15% in 1Q26, with Hesai remaining the leader at 51% share.

Report interpretation

Overview

This report uses Gasgoo automotive insurance registration-based data to track China smart-driving chip industry penetration, supplier share, OEM procurement changes, and LiDAR fitment in 1Q26. It notes that the rollback of EV subsidies, weak EV sales, rising storage prices, and seasonal softness in passenger-vehicle demand jointly pressured NOA penetration, but L2++ remains the more valuable and sustainable upgrade direction in the medium to long term.

Core views

Core views include: first, 1Q26 NOA penetration declined to 32% in the short term, mainly due to sales mix shifts rather than a reversal of the long-term smart-driving trend; second, L2++ is more resilient than L2+, and urban NOA experiences are more effective at lifting consumer willingness to pay; third, Horizon continues to expand share in L2+ and L2++ with its J6P, dual J6M, and single J6M solutions, supported by an OTA upgrade path; fourth, NVIDIA still dominates outsourced L2++ solutions, but in L2+ it faces competition from Horizon's higher cost-efficiency alternatives; fifth, Black Sesame remains weak on adoption because of customer concentration, slower design-in, and R&D cash-flow pressure.

Analysis framework

The report uses vehicle data based on end-user sales and insurance registration statistics, measuring smart-driving penetration by L1-L2, L2+, and L2++, and further breaks down the competitive landscape by supplier, OEM customer, TOPS compute, and LiDAR fitment. It also combines OEM sales mix, outsourcing preferences, chip configuration, design-win progress, and valuation multiples to assess investment implications for Horizon, Black Sesame, NVIDIA, Qualcomm, and relevant automakers.

Methodology notes

  • 行业跟踪Gasgoo终端销量与保险注册口径

    Counts vehicles equipped with smart-driving functions based on vehicles sold to end users.

    This methodology lags chip vendor shipments to OEMs or Tier-1 suppliers by about 3 to 5 months, but it reflects medium- and long-term industry trends and competitive dynamics fairly well.

  • 产品分层ADAS智能驾驶等级分层

    Segments smart-driving functions into None/L0, L1-L2, L2+, and L2++.

    The report defines NOA focus on L2+ and L2++, where L2++ is closer to urban NOA capability and is considered the main battleground before L3.

  • 竞争分析车辆搭载量与TOPS双口径份额

    Measures supplier share using both the number of equipped vehicles and installed compute measured in TOPS.

    Vehicle-count share reflects penetration breadth, while TOPS share reflects the weight of higher-compute solutions; in L2++, self-developed solutions from Tesla, Huawei, NIO, and Xpeng have higher compute, so the share structure under the TOPS lens differs from the vehicle-count lens.

Asset mapping & comparison

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

  • Horizon Robotics
    A key beneficiary in China’s smart-driving chip chain; Bernstein maintains Outperform but cuts the target price to HKD 10.
    Strengths
    L2+ and L2++ share rose even as the sector weakened; J6P, dual J6M, and single J6M cover different price points; HSD and J6P are already adopted by 10 OEMs and more than 20 models; BYD L2++ design wins and the OTA upgrade path support faster penetration.
    Weaknesses
    J6 family growth momentum is affected by weak passenger-car demand and delayed launches of key customer models; absolute J6P install base remains below NVIDIA Thor and Huawei 610; heightened near-term competition and higher R&D spending pressure margins.
    Comparison
    Compared with NVIDIA, Horizon is more competitive on cost efficiency in L2+; compared with Black Sesame, Horizon has stronger software and hardware-software integration capabilities and faster customer expansion.
    Risks
    EV sales recovery slower than expected, sustained storage cost pressure, escalating L2++ competition, delayed design-in, and valuation multiple compression.
  • Black Sesame
    A China smart-driving chip supplier, with Bernstein maintaining Underperform and an HKD 16 target price.
    Strengths
    Moving from L2+ to L2++, and has disclosed one OEM confirming A2000 for L2+; added embodied-AI solution business is also expected to contribute to revenue.
    Weaknesses
    L2+ customers are mainly limited to Geely and Dongfeng, and Geely share has been eroded by Horizon; A1000 shipments declined for three consecutive quarters; software capability is lacking, design wins are slow, scale is limited, and R&D cash-flow pressure is significant.
    Comparison
    Compared with Horizon, Black Sesame is weaker on customer breadth, software capability, and speed of market adoption.
    Risks
    Continued slow market adoption, financing dilution, delayed L2++ deployment, and persistent cash-flow pressure.
  • NVIDIA
    Dominant supplier of outsourced L2++ third-party solutions and a covered U.S. semiconductor stock.
    Strengths
    L2++ remains a leading share position; it has strong brand and ecosystem advantages in high-compute smart-driving solutions; Bernstein maintains Outperform with a $300 target price.
    Weaknesses
    L2+ share fell from 45% to 35% due to substitution by higher cost-efficiency solutions such as Horizon J6M; L2++ share also fell from 54% to 50%.
    Comparison
    In high-end L2++ solutions, it remains stronger than domestic third-party suppliers, but faces pressure from Horizon in the cost-sensitive L2+ market.
    Risks
    OEM shifts to self-developed chips or domestic third-party suppliers, pricing competition, and cyclicality in model sales.
  • Qualcomm
    An L2+ smart-driving chip participant, rated Market-Perform by the report with a $140 target price.
    Strengths
    Customer base is expanding from Leapmotor to an additional five OEMs, and L2+ share has modestly risen from 11% to 12%.
    Weaknesses
    No visible L2++ deployments yet; the report views the SA8775P Urban NOA path at 72 TOPS as unlikely to become mainstream.
    Comparison
    Compared with Horizon and NVIDIA, Qualcomm has lower visibility in high-tier L2++ solutions.
    Risks
    Sub-100 TOPS solutions may fail to meet consumer expectations for flexible and safe L2++ experiences, limiting smart-driving competitiveness.
  • Hesai
    A structural beneficiary in China LiDAR.
    Strengths
    Hesai reached a 51% share in China LiDAR in 1Q26 and is well positioned as LiDAR usage exceeds 80% in urban NOA models.
    Weaknesses
    Overall LiDAR penetration in China declined to 15% in 1Q26 as EV sales weakened.
    Comparison
    Leads Huawei, RoboSense, and Soyond.
    Risks
    EV sales mix volatility, slower urban NOA penetration pace, and competition from non-lidar alternatives.

Key data

  • 1Q26 China passenger vehicle sales4.05 million vehiclesdown 18% year-over-year and down 36% sequentially.
  • 1Q26 NOA penetration32%down from 36% in 4Q25; NOA includes L2+ and L2++.
  • 1Q26 L2+ penetration13%down 3 percentage points sequentially, mainly due to sales mix changes and weaker sales of BYD-related L2+ models.
  • 1Q26 L2++ penetration19%down 1 percentage point sequentially, a smaller decline than L2+, indicating stronger resilience of urban NOA demand.
  • Horizon L2+ share change+7 percentage points QoQMeasured by equipped vehicle count, Horizon still gained share despite weaker absolute sales.
  • Horizon L2++ share changefrom 1% to 5%up 4 percentage points QoQ, with J6P shipments up 24% QoQ.
  • NVIDIA L2+ sharefrom 45% to 35%Horizon J6M, as a higher cost-efficiency alternative, pressures Orin-N.
  • NVIDIA L2++ sharefrom 54% to 50%still remains dominant, but faces competition from OEM self-developed chips and domestic third-party suppliers.
  • 1Q26 LiDAR penetration15%below about 17% in 4Q25; Hesai leads with a 51% share, followed by Huawei, RoboSense, and Soyond at 23%, 16%, and 10% respectively.

Impact & implications

In the short term, weak EV sales, changes in subsidy policy, rising storage costs, and fragile sector sentiment are weighing on valuations and earnings expectations of smart-driving chip companies, so Bernstein lowered Horizon's target price and reduced its 2027E and 2028E revenue and operating margin assumptions. In the longer term, urban NOA has stronger user value and legacy OEM penetration remains low; if Horizon can successfully upgrade from an L1-L2 supplier to a L2+ and above supplier for OEMs such as BYD and Geely, it should continue to benefit from domestic substitution and AD upgrading.

Risks

  • EV sales recovery slower than expected, continuing to weigh on NOA and LiDAR penetration.
  • Persistent increases in storage prices, adding hardware cost pressure for smart-driving.
  • Further adjustments to EV subsidy policy affecting new-energy vehicle sales mix.
  • Heightened L2++ chip competition, with Momenta, Black Sesame, and OEM self-developed solutions entering the market.
  • Horizon J6P and HSD key model launches or design-in progressing slower than expected.
  • Black Sesame's customer expansion and A2000 deployment lag expectations and it may face dilution.

What to watch

  • Whether 2Q26 EV sales recovery drives a rebound in NOA penetration.
  • Whether L2++ penetration continues to be more resilient than L2+.
  • Horizon's pace of new design wins and install ramp at OEMs such as BYD, Geely, Chery, Changan, and Li Auto.
  • The scaling speed of Urban NOA via OTA on single J6M.
  • Whether NVIDIA's share in L2++ outsourcing solutions continues to be eroded by OEM self-developed chips and domestic suppliers.
  • LiDAR penetration in urban NOA models and whether Hesai can maintain its leading share.
Zhejiang ICP No. 2022035445-5
Disclaimer: Market data, charts, indicators, research views, and other information provided on this website are intended solely for information display, research communication, and educational reference. They should not be regarded as personalized investment advice, securities recommendations, trading instructions, solicitations, or guarantees of return. While we strive to improve the reliability of our data and content, such information may still be subject to delays, errors, incompleteness, or untimely updates due to source differences, methodological limitations, system processing, or market volatility. Users should exercise independent judgment based on their own circumstances and bear all risks and responsibilities arising from the use of this website.

Settings

Sign in to view recent logins