L2++ Penetration Rises to 26%, Horizon Robotics Leads the L2+ Market for the First Time

Overview

The report is based on China auto insurance registration data provided by Gasgoo and tracks, by quarter, intelligent driving levels, chip supplier shares, OEM customer mix, and LiDAR penetration for vehicles sold to end users. In Q2 2026, as Urban NOA deployment accelerated, China’s NOA penetration rebounded significantly, with L2++ becoming the main growth driver. Horizon Robotics surpassed NVIDIA in L2+ installed-vehicle share for the first time, but its L2++ share remained below expectations.

Core views

China’s intelligent driving market is upgrading from basic highway assisted driving to Urban NOA, driving continued growth in semiconductor content per vehicle and computing-power demand. Incremental growth over the next 1 to 2 years will mainly come from exploratory OEMs such as BYD and Geely, while traditional OEMs have larger long-term penetration potential and are more dependent on third-party chips. Horizon Robotics became No. 1 in the L2+ market thanks to J6M volume ramp-up and share gains at BYD and Geely, but further re-rating depends on HSD model sales performance and whether it can secure BYD’s L2++ supply share. NVIDIA still leads L2++ with a 52% installed-vehicle share, but it is facing combined challenges from OEM in-house solutions, Horizon Robotics, and Qualcomm.

Analysis framework

The report uses end-user auto insurance registrations as the basis, first classifying intelligent driving levels into L1-L2, L2+, and L2++, then comparing chip supplier shares from two dimensions: installed vehicle count and total computing power. It also divides OEMs into leaders, explorers, and traditional manufacturers to assess penetration potential, outsourcing tendencies, and potential market expansion paths across different customer groups, and combines supplier wallet share at key OEMs to form company-level judgments.

Methodology notes

• Data trackingEnd-user insurance registration tracking

  Uses data on vehicles actually sold to end users and registered for insurance to measure intelligent driving configurations.

  This method is closer to end-user demand, but there is a lag between chip shipments to OEMs or Tier-1 suppliers and final vehicle sales. Different sections of the report describe the lag as 3 to 5 months or 1 to 2 quarters, so single-quarter installed-vehicle data should not be directly treated as chipmakers’ revenue for the same period.

• Product tieringIntelligent driving level classification

  Classifies vehicles into no ADAS or L0, L1-L2, L2+, and L2++.

  L2+ mainly corresponds to highway NOA, while L2++ mainly corresponds to city NOA; both are close in functionality to higher-level autonomous driving, but the driver remains responsible, so they still fall under ADAS.

• Customer segmentationThree-category OEM grouping

  Divides OEMs into leaders, explorers, and traditional manufacturers based on intelligent driving penetration and outsourcing preference.

  Leaders mainly include Tesla and Huawei, whose L2++ penetration has exceeded 80%; explorers include BYD, Geely, and emerging local manufacturers, and are the main source of recent incremental growth; traditional manufacturers have lower penetration, but larger long-term potential and greater reliance on third-party solutions.

• Competitive landscapeDual measurement by installed vehicles and TOPS

  Compares chip supplier market positions using both installed vehicle count and total installed computing power.

  Installed vehicle count reflects platform coverage scale, while total TOPS reflects computing-power value content; combining the two can reduce the bias of judging high-computing-power platform competitiveness solely by chip quantity.

• Valuation methodsForward enterprise value multiple method

  Estimates target value by multiplying forward revenue by the enterprise value-to-sales multiple.

  Horizon Robotics’ target price uses a 10.3x enterprise value-to-sales multiple, corresponding to RMB 11.5 billion of revenue during the 2BF period, namely the second half of 2027 to the first half of 2028.

Asset mapping & comparison

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

• Horizon Robotics (9660.HK)

  Core beneficiary and L2+ market leader

  Strengths

  Has integrated SoC and algorithm capabilities, J6M is ramping up, it has gained more share at BYD and Geely, and its balance sheet supports continued R&D investment.

  Weaknesses

  L2++ share fell to 3%, and HSD-related model sales and design-win progress were slower than expected.

  Comparison

  L2+ installed-vehicle share was 32%, surpassing NVIDIA’s 28% for the first time; L2++ share significantly lagged NVIDIA’s 52%.

  Risks

  BYD partnership scale falls short of expectations, HSD model sales are weak, and OEM in-house chips and new entrants intensify competition.

• NVIDIA (NVDA)

  Current L2++ market leader

  Strengths

  Its high-computing-power platform is mature, and it still had a 52% L2++ installed-vehicle share in Q2 2026.

  Weaknesses

  Its L2+ and L2++ shares are both declining, and it has been surpassed by Horizon Robotics in L2+.

  Comparison

  Maintains a significant lead in L2++, but faces challenges from third-party suppliers and OEM in-house solutions.

  Risks

  BYD shifts to a multi-supplier system, OEM in-house substitution, and volume ramp-up of new platforms such as Qualcomm.

• Qualcomm (QCOM)

  Fast-growing intelligent driving chip supplier

  Strengths

  SA8797P offers more than 600 TOPS of computing power and has entered the leading L2++ platform camp through deployment across Leapmotor’s full model lineup.

  Weaknesses

  Its market-share base is relatively low, and its customer structure remains highly dependent on Leapmotor.

  Comparison

  L2+ share rose to 20%, growing faster than major competitors, and it has officially entered the L2++ market.

  Risks

  High customer concentration, volatility in Leapmotor sales, and new platform ramp-up falling short of expectations.

• Black Sesame (2533.HK)

  Local intelligent driving chip supplier under share pressure

  Strengths

  Has deployed L2+ chips and is attempting to expand into L2++.

  Weaknesses

  Software capabilities are relatively insufficient, it lacks new OEM customers, continues to lose wallet share at Geely, and has not yet secured an L2++ vehicle program nomination.

  Comparison

  Compared with Horizon Robotics, its customer acquisition, scaling, and software-hardware integration capabilities are weaker.

  Risks

  R&D burden and cash-flow pressure may lead to frequent financing and shareholder dilution.

• Hesai (HSAI.US, 2525.HK)

  Major beneficiary of rising LiDAR penetration

  Strengths

  Maintains the lead with a 47% China market share, and has a strong global business and technology position.

  Weaknesses

  Demand remains highly correlated with Urban NOA model sales and LiDAR configuration rates.

  Comparison

  Market share is higher than RoboSense’s 23%, Huawei’s 21%, and Seyond’s 9%.

  Risks

  Price competition, substitution by LiDAR-free solutions, and intelligent vehicle sales falling short of expectations.

• BYD

  Largest potential incremental source of future Urban NOA demand

  Strengths

  Has large sales scale and is shifting its intelligent driving focus from L2+ to L2++.

  Weaknesses

  Its supply system is still in an adjustment period, and the final allocation to third-party suppliers remains unclear.

  Comparison

  Supplier strategy is shifting from emphasizing NVIDIA to a coexistence of NVIDIA, Horizon Robotics, and in-house chips.

  Risks

  Urban NOA deployment pace or end-market model sales fall below expectations.

Key data

• NOA penetration42.6%In Q2 2026, it rebounded significantly from 32% in Q1.
• L2++ penetration26%Increased 7 percentage points quarter over quarter, reaching a record high.
• L2+ penetration16%Increased 3 percentage points quarter over quarter.
• L1-L2 penetration37%The report’s detailed section shows a 4-percentage-point quarter-over-quarter decline, reflecting consumer migration toward higher-level functions.
• NOA model sales growthUp 17.6% year over yearOver the same period, total China passenger-vehicle sales were 22% lower than in Q2 2025.
• Horizon Robotics’ L2+ share32%Calculated by installed vehicle count, it surpassed NVIDIA for the first time and became No. 1 in the market.
• NVIDIA’s L2+ share28%Facing competition from Horizon Robotics, Qualcomm, and OEM in-house solutions.
• Qualcomm’s L2+ share20%Mainly driven by Leapmotor sales and platform deployment.
• NVIDIA’s L2++ share52%Still No. 1 in the market, but its leadership position is being challenged.
• Horizon Robotics’ L2++ share3%Down from 5% in the previous quarter, with design-win progress slower than expected.
• China LiDAR penetration20%Recovered from about 15% in Q1 2026.
• Hesai LiDAR share47%Ahead of RoboSense at 23%, Huawei at 21%, and Seyond at 9%.

Impact & implications

Urban NOA is evolving from a high-end differentiating feature to a mainstream function, which will increase intelligent driving chip value content per vehicle and computing-power demand, and expand the potential market for high-computing-power SoCs and LiDAR. Horizon Robotics is the main beneficiary of L2+ share gains; if it secures meaningful supply share from BYD, the report estimates its L2+ share could rise from 32% to 55% to 60%, and even a limited BYD allocation in the L2++ market could bring its current 3% share close to tripling. By contrast, Black Sesame faces insufficient customer expansion and loss of Geely wallet share. Although NVIDIA still leads L2++, its automotive intelligent driving share is under pressure; Qualcomm and Hesai benefit respectively from high-computing-power platform volume ramp-up and rising LiDAR penetration.

Risks

• The scale, timetable, or model sales performance of Horizon Robotics’ partnership with BYD is below expectations.
• OEM in-house chips accelerate penetration, compressing the potential market for third-party intelligent driving SoC suppliers.
• China passenger-vehicle demand remains weak, affecting installed volumes of high-level intelligent driving models.
• There is a lag of several months between chip shipments and end-user insurance registrations, so quarterly data may not promptly reflect suppliers’ current operating conditions.
• Competition and pricing pressure in intelligent driving chips intensify, potentially weakening the profit contribution from market-share gains.
• Changes in safety, regulation, or consumer acceptance of L2++ and Urban NOA may affect the pace of adoption.
• Black Sesame’s R&D and cash-flow pressure may lead to further financing and shareholder dilution.
• Qualcomm’s reliance on a small number of customers such as Leapmotor may amplify the impact of single-customer sales volatility.

What to watch

• Further disclosure and actual model launches for Horizon Robotics’ partnership with BYD in the second half of 2026.
• Whether Horizon Robotics can obtain meaningful wallet share in BYD’s Urban NOA supply system.
• The installed-vehicle ramp-up of the J6M and J6B platforms, and end-user sales performance of HSD models.
• Whether NVIDIA’s 52% share in the L2++ market can withstand challenges from third-party suppliers and OEM in-house solutions.
• The pace of Qualcomm SA8797P volume ramp-up and whether its customer structure can further diversify beyond Leapmotor.
• Whether Geely continues shifting L2+ share from Black Sesame to Horizon Robotics, as well as its L2++ supplier selection.
• Whether L2++ penetration among exploratory OEMs can continue rising, and when traditional OEMs will accelerate NOA adoption.
• Whether LiDAR penetration and Hesai’s market share can continue improving as Urban NOA becomes more widespread.