Financial Report Analysis | Horizon's 2025: Shipments Surpass 4 Million Units, Will Profitability Turn Around?

Is AI · HSD 2.0 able to become the key catalyst driving Horizon’s profitability inflection point?

By China Jingji News reporter Chen Maoli and Zhang Shuo, Beijing

Horizon (9660.HK) recently released its 2025 full-year financial report. In 2025, Horizon delivered a “both volume and price rising” performance: its automotive-grade chip shipment volume exceeded 4 million units, up 39% year over year; total revenue was RMB 3.76 billion, up 57.7%; and ASP (average selling price) increased by more than 75% compared with 2024. Amid the industry trend in which intelligent driving is moving from “add-on” to “standard equipment,” Horizon successfully secured its position and became the absolute leader in NOA solutions in the mainstream market under 200k yuan—accounting for a 44% share among independent brands.

But another side of this performance is equally striking: its overall gross margin fell from 77.3% to 64.5%, with an adjusted operating loss of RMB 2.37 billion. Facing margin pressure alongside scale expansion is Horizon’s core contradiction today. Regarding its losses, after the financial report was released, Horizon’s founder and CEO Yu Kai told reporters from outlets including the “China Business News” that: “During the reporting period, our adjusted operating loss was RMB 2.37 billion, mainly because in 2025 we further increased our R&D investment.” The financial report shows that in 2025, Horizon’s R&D spending grew from RMB 3.16 billion to RMB 5.15 billion.

Battle for the market window to “hold the line”

In 2025, Horizon’s revenue was RMB 3.76 billion, up 57.7%. The driving force behind this growth comes from automotive solutions. In terms of revenue composition, Horizon’s business is divided into automotive solutions and non-automotive solutions, with the former being the absolute base. In 2025, revenue from automotive solutions was RMB 3.56 billion, accounting for 94.6% of total revenue.

Breaking down Horizon’s revenue composition for 2025 reveals that the internal structure of its automotive business—two key segments, namely authorization and services (technical services such as algorithm and software authorization) and product solutions—is being reshaped. The mix has moved from the former “70/30” (authorization accounting for 70%) to today’s “60/40.” In 2025, authorization and services revenue was RMB 1.94 billion, accounting for 54.5% of automotive business; product solutions revenue was RMB 1.62 billion, accounting for 45.6%, up 144.2% year over year.

The profitability of the two businesses differs markedly. In 2025, the authorization business (such as algorithm and software authorization) had a gross margin as high as 94.5%, serving as Horizon’s stable “cash-generating” source; in contrast, the full-year gross margin for product solutions was only 34.5%.

Some analysts believe the decline in gross margin for product solutions was influenced by intensifying market competition, and it is also a proactive choice by Horizon: by continuing to “generate cash” through high-gross-margin authorization business, it can support product solutions in rapidly scaling up—first building market share, then seeking improvement in profitability.

Behind this choice lies the battle for the market window. According to data from a third-party institution, in 2025, the penetration rate of China’s mid-to-high-end NOA among passenger vehicles grew from 21.6% to 42.6%, nearly doubling. Intelligent driving is shifting from “add-on” in high-end models to “standard equipment” for the mass market. In this time window, whoever can secure share in the mainstream market can effectively define the future ecosystem.

This strategy is reflected in market share. In 2025, Horizon maintained the #1 position among independent-brand ADAS solution providers with a 47.7% market share, leading the industry. At the same time, in the fast-growing autonomous-brand mid-to-high-end intelligent driving market, Horizon—thanks to a 14.4% market share—was almost tied for second place with Huawei’s 15.2% share, and together with NVIDIA formed the first tier of the market. The combined market share of these three companies reached 89%, showing a clear concentration effect at the top.

Yu Kai summarized this strategy as “secure the territory first, then deepen the cultivation.” The price is pressure on near-term profitability, but the payoff is securing a core position in the wave of intelligent driving democratization.

Chip scaling down and mid-to-high-end volume expansion

With the market-share “territory acquisition” as a foundation, Horizon’s focus in 2026 shifted to “deep cultivation”—turning scale advantages into sustainable growth momentum.

For 2026, Yu Kai offered an optimistic outlook: achieving a steeper growth curve. He said that last year the company’s expectation for revenue growth rates over the coming few years was around 50%, and the company is confident it can sustain a growth trend of both volume and price rising this year, pushing the average revenue growth rate over the next few years up to 60%.

Yu Kai pointed out that a strong product cycle, ample order backlog for designated projects, and the leadership of next-generation soft and hardware technologies will provide strong support for achieving this goal.

In terms of chip solution shipments, Horizon has made key progress. At present, the industry’s first end-to-end urban NOA solution based on a single-chip Journey 6M has been formally delivered and installed in vehicles. Yu Kai described: “By designing for extreme depth of coordinated software and hardware, the system complexity and comprehensive costs have been significantly reduced.” This year, Horizon will work with many ecosystem partners to further roll out urban NOA functionality from the 150k-yuan mainstream market to the 100k-yuan-level national car market.

In the mid-to-high-end urban intelligent driving solution area, Horizon plans to release HSD 2.0 this year. Yu Kai said the product will deliver an “absolutely leading experience.” He revealed that Horizon is currently negotiating HSD with leading Chinese automakers, and it expects HSD shipments to reach around 400k units.

Some investment research firms hold similar views, arguing that Horizon’s HSD solution has a favorable cost structure. In particular, its ability to deliver urban NOA on a single Journey 6M chip gives it differentiated advantages in price-sensitive mainstream markets. They expect 2026–2027 to be a key window period for its intelligent driving solutions to scale up in volume.

Yu Kai said: “We firmly believe that HSD is not only the core strategic product for Horizon to win in urban intelligent driving; it is also the technology foundation for the future journey toward L3/L4-level autonomous driving. In addition, the underlying AI foundation model of HSD will further empower industries such as robotics.”

Worth noting, Horizon disclosed that during the 2026 Spring Festival period, among the first models equipped with HSD, the share of autonomous driving users’ autonomous driving mileage in total mileage reached 41%. “This is a strategically meaningful number,” Yu Kai said in its communications. “In the past, everyone asked me under what conditions Horizon might consider charging for HSD through a subscription model. In my view, once the autonomous driving mileage share surpasses the critical threshold of 50%, once the mileage dominated by machines exceeds that by humans, users’ reliance on autonomous driving will become irreversible.”

He further explained: “At present, our revenue is mainly driven by new vehicle sales equipped with our products; in the future, we are likely to rely on the large installed base of vehicles equipped with HSD to charge service subscription fees across the full lifecycle.” This implies that Horizon will transition from being a supplier to becoming a service provider.

Opening up a new growth curve

In the Q&A session after the financial report release, Yu Kai elaborated on Horizon’s strategic layout under the trend toward “cockpit + driving integrated.” He believes that under the pressure of industry costs, “cockpit-driving integration” is an inevitable result of technological evolution, and 2026 will become the acceleration year for this trend.

“Cockpit-driving integration” refers to deep integration of the traditional smart cockpit domain controller and the intelligent driving domain controller. Through a single chip or a computing platform, it simultaneously carries two core functions: the cockpit (human-machine interaction, entertainment, instrument panel, etc.) and intelligent driving (perception, decision-making, planning, control).

Horizon’s entry path is from intelligent driving extending toward the cockpit. Yu Kai believes this path has advantages compared with extending from the cockpit toward intelligent driving. “When transitioning from intelligent driving to the cockpit, we generally feel it is a more ‘high-dimensional’ offensive than transitioning from the cockpit to intelligent driving,” he said. He explained that intelligent driving places far higher requirements on chip computing power, safety, and real-time performance than the cockpit does. Extending from high-threshold areas to low-threshold areas naturally gives an advantage in technical capability.

At present, companies developing cockpit-driving integration fall into three major camps: chip makers, Tier 1 suppliers, and automakers. In the chip maker tier, major players show different technical routes. Qualcomm relies on its advantage in the cockpit domain and extends from cockpit toward intelligent driving; NVIDIA covers downward from mid-to-high-end intelligent driving with its extremely strong computing power at the core; and Horizon emphasizes its unique “soft-and-hardware integration” capability, extending from intelligent driving toward the cockpit. Yu Kai commented on this: “We can leverage our soft-and-hardware integration solution to connect from chip to intelligent driving software to intelligent agent OS—this is not a field that Qualcomm and NVIDIA can reach.”

Even if Horizon has a leading advantage in the path from intelligent driving extending toward the cockpit, it still faces competitive pressure brought by automakers developing in-house. Tesla, NIO, Li Auto, and other new forces, as well as traditional automakers such as BYD and Geely, are all actively developing cockpit-driving integration solutions, which pose potential challenges to third-party suppliers including Horizon.

However, in terms of R&D and mass-production speed, Horizon has a clear advantage. Yu Kai revealed that Horizon plans to announce the specific cockpit-driving integration solution at the April press conference: “This year, we will launch an all-vehicle intelligent agent solution for cockpit-driving integration. At least in terms of memory, it is expected to help automaker customers save over a thousand yuan in cost. If you consider harnesses, cooling, PCB area, and so on, it will further help customers reduce costs.”

With high gross margin “cash infusion” from authorization business, using scale expansion to secure position with product solutions, and building technology barriers through continuous high-intensity R&D, Horizon ultimately achieves the double fulfillment of “scale + profit” in the mid-to-high-end intelligent driving market, entering a “validation period” in 2026.

(Editor: Zhang Shuo; Proofreader: Tong Haihua; Copy editor: Zhai Jun)