Horizon Robotics Jumps to No. 2 in Urban NOA and Guides Full-Year Revenue Above RMB 5 Billion
August 31, 2026 — 2026 Interim Results Conference Call
Horizon Robotics used its first-half 2026 results call to argue that it is no longer a latecomer in advanced assisted driving. Founder and CEO Kai Yu said urban NOA computing-platform share among Chinese domestic brands rose to around 23% in the first six months, up roughly 5 percentage points, lifting the company from third to second place. The only player still ahead is a U.S. technology company whose share fell by nearly 10 percentage points versus full-year 2025. “In this arena, Horizon is on the offensive, not defensive,” Yu said. Highway NOA only entered mass production last year and urban solutions only at the end of last year, so the six-month share grab is the core new fact for investors, even if it remains a domestic-brand slice of the market rather than an all-China number.
ADAS precedent and the 2027 “combined No. 1” claim
Yu framed the advanced-driving fight as a replay of ADAS. Horizon’s ADAS share was less than 5% in 2022 and broke 50% in the first half of 2026, twice the No. 2 player. “Going from under 5% to 50% took us only 4 years. I think it will be the same in the advanced market.” The 2027 endgame he described is not Horizon chips alone. Direct SoC shipments plus automaker in-house chips that license Horizon IP are supposed to take the No. 1 position in the advanced segment together. That is a convenient definition. It credits both product wins and the ARM-plus-Android licensing overlay, so reported “share” can keep rising even where an OEM stamps its own silicon. The claim is still useful: it tells investors management is explicitly underwriting a two-track model rather than a pure merchant-chip story.
HSD coverage now spans top Chinese brands and the two largest JVs
Software is the proof point management wants the market to price. HSD 1.0 launched last November and HSD 2.0 at the end of June. Yu said third-party evaluations already put Horizon ahead on takeover ratio, and investor ride-alongs in Beijing, Shanghai and Shenzhen were used to sell smoothness in rush-hour hospitals, tight U-turns and urban villages. More commercially relevant, HSD is already on vehicles from Chery and Changan and, according to Yu, is available across the top five highest-volume Chinese brands. J6M is expected to account for 70% to 80% of entry-level AD shipments at the largest new-energy vehicle maker, with high-compute SoCs plus HSD starting mass production on that customer’s advanced platform this year.
The joint-venture channel is the second new commercial fact. Toyota and Volkswagen are the two largest JV groups in China, and Yu said HSD has become the design for both. Through Carizon, the Volkswagen joint venture, Journey 6-based solutions are slated for seven all-new FAW-Volkswagen, SAIC-Volkswagen and ID models this year, then a broader rollout onto Volkswagen’s mainstream CEA architecture next year, about 20 models in total. Toyota’s Journey 6 program entered mass production in the first half on GAC Toyota’s highest-volume entry-level model, with Denso expected to take the stack onto more mainstream platforms and contribute sales the year after next. CFO Lei Wang added that high-compute chips plus HSD recently broke through at two leading Chinese automakers, with those design wins starting production at the end of this year and “significant revenue contribution next year.” No other supplier, Wang argued, sits across the top five local brands and the top two JVs at once. That positioning is real; conversion into volume and mix is still a 2027 story.
First-half results: 33% growth, 66% gross margin, and a back-half-loaded year
Against CPCA data showing domestic passenger-vehicle retail down more than 20% year-on-year in the first half, Horizon recognized revenue of RMB 2.055 billion, up 32.9%, at a 66% gross margin. Product shipments rose 12.1% to about 2.2 million units despite the auto slump. License and services, including Carizon and one of the world’s largest auto-parts groups, reached RMB 1.1 billion, up 52.7%, at a 90.4% gross margin. Product-solutions gross margin was 36.2%, depressed by bundling domain controllers with core products to speed HSD ramps. Excluding that one-time mix, product gross margin was 48.1%, up 3 percentage points year-on-year. Combined R&D, selling and administrative expense growth of 23.3% lagged revenue, which Yu cited as emerging operating leverage. The absolute picture is less flattering. R&D including share-based compensation was slightly more than RMB 2.7 billion and SG&A was RMB 634 million, or RMB 3.3 billion together. Operating loss was RMB 1.7 billion, or RMB 1.3 billion on an adjusted non-IFRS basis. IFRS profit of RMB 3.8 billion was an accounting artifact of fair-value gains on a convertible loan. Adjusted net loss was RMB 1.67 billion after stripping that gain, adding back SBC and deducting the one-time impact of deconsolidating D-Robotics. Management is confident of full-year revenue of more than CNY 5 billion, which implies a sharply heavier second half than the first. Chip shipments are guided at around 5 million units this year and more than 7 million next year, with J6M volume at the large NEV customer as the unit driver and higher-value 6M, 6P, 6H and HSD as the revenue mix driver.
Price war as a feature, not a bug, and the 2028 breakeven line
Yu was unusually direct on OEM margin pressure. He called the vehicle price war “a very good opportunity for Horizon” because cost-effective high-performance stacks become more attractive as in-house R&D becomes unaffordable. In the endgame he expects only 20% or less of OEMs to sustain in-house capability, with the rest buying from independent suppliers. That is the bull case for a merchant-plus-IP vendor; it also assumes Horizon keeps winning when Chinese OEMs are cutting every line item. On returns, Yu refused to throttle R&D. He said today’s revenue is the fruit of prior-year spend, R&D is the seed of future statements, and saturated investment is required because a low ceiling would collapse the industry into price competition. A thought experiment he offered: RMB 10 billion of revenue at a 60% gross margin produces RMB 6 billion of gross profit that, if poured into SoCs and AI, would exceed what most automakers can fund. Breakeven remains “about 2028.” Investors should treat that as an objective, not a bridge. H1 already shows R&D running well above revenue, and cloud training costs were the bulk of the RMB 650 million expense increase.
Volkswagen capital ties, overseas ADAS beachhead, and Journey 7
The Volkswagen relationship deepened on both product and capital. Horizon redeemed a portion of convertible bonds held by Volkswagen, cutting equity dilution by 4.9%. CARIAD converted early into a 9.9% strategic stake and agreed to a 12-month lockup. That is a cleaner shareholder register and a tighter industrial alignment; it also concentrates a large, locked strategic holder just as HSD is supposed to scale on CEA. Overseas, Yu said only three intelligent-driving suppliers can support the Chinese groups that dominate exports, and Horizon has 24 brands and nearly 60 export models across the top six export groups. J6B has secured 20 million units of orders. Bosch and neueHCT have each built a new generation of entry-level products on J6B, with design wins described in the tens of millions and millions of units across China, Europe, Australia and New Zealand, Latin America, the Middle East and Asia Pacific. The strategy is explicit: own overseas ADAS first, then upgrade the same customers to J6E and J6M as advanced demand appears in France, Germany and Japan. J6E and J6M are already designed into nominated export models, but Yu conceded the overseas advanced ramp is slower than China.
Journey 7, aimed at L3 and L4, is the next hardware step. R&D is “progressing smoothly,” with tape-out expected early in the second quarter of 2027. The chip is described as natively optimized for a larger-parameter next-generation HSD model and a local LLM cockpit model. Leading automakers and Tier 1s have already approached the company, which Yu read as confirmation of demand. Horizon is also testing L4 robotaxi with a partner. Yu’s point that private L4 and robotaxi share the same foundation is directionally right; it does not shorten the regulatory or SOP gap between a 2027 tape-out and material revenue. Parallel to autos, the ARM-plus-Android licensing model is being pushed into robots via associate D-Robotics, which Yu said has reached more than 100 robot categories, more than 400 customers and more than half of China’s embodied-intelligence firms. Open-source embodied models have been tested by NVIDIA, Meta, Microsoft, ByteDance, Alibaba, Unitree and several universities. That is optionality, not 2026 earnings power, especially after D-Robotics was deconsolidated.
The industry backdrop Yu described is internally consistent even when it is self-serving. Intelligent assisted-driving penetration in China passenger vehicles hit 76.1%, up 8.5 percentage points from 2025, with joint-venture penetration above 80% for the first time and urban NOA on about 23% of all new cars sold. JV brands have stabilized near 35% share. Horizon wants to be the Chinese supplier that JVs use instead of imported stacks, and the software-plus-SoC vendor that captures OEMs who cannot fund a steep technology curve. The new information from the call is that HSD is no longer a slide: it has design wins at the highest-volume local brands and at Volkswagen and Toyota, license revenue is already the profit engine, and management is willing to stay deep in the red until about 2028 to keep that curve steep. The risk is unchanged. A 2026 revenue target above RMB 5 billion, a 2027 combined-share claim that includes other people’s chips, and a 2028 breakeven all still have to survive a weak China retail market and a product gross margin that only looks healthy after backing out the DCU bundling used to buy the ramp.