The life-and-death elimination race in China's intelligent driving industry has entered its second half.

Brands remaining in the competition are surrounded by the spotlight. Momenta's third-party market share surpassed 61%, Huawei's ADS-equipped vehicles exceeded 1.4 million units cumulatively, Horizon Robotics' chip shipments surpassed 10 million units, and DeepRoute.ai's market share soared to 38% in a single month in October 2025.

Meanwhile, companies exiting the race have essentially received "terminal illness notices." Haomo AI has Complete shutdown (completely halted operations), Jingchi Tech has entered bankruptcy restructuring, Zhongzhihang was ordered into liquidation for failing to pay RMB 15,000, and HoloMatic has initiated bankruptcy proceedings.

By the end of 2025, China's L2-level passenger vehicle penetration reached 66.1%, while urban NOA (Navigate on Autopilot) functionality penetration hit 15.1%. Guojin Securities predicts that urban NOA hardware configuration penetration will rise from 16% in 2025 to 25% in 2026, with annual equipped sales potentially reaching 5.45 million units.

The market pie is growing, but the number of players eating it is shrinking rapidly.

| Two Extremes Coexist |

Shanghai-based Momenta emerged as a rising star. CAAM reports show that from January to November 2025, Momenta's urban NOA deployments reached 414,400 units, accounting for 61.06% of the third-party supplier market. Combined with second-place Huawei HI Mode (~19.76%), the two firms command over 80% market share, forming a solid "duopoly."

Technologically, Momenta pioneered mass production of "end-to-end + reinforcement learning" systems. In Q3 2025, its Momenta R6 reinforcement learning model became China's first truly mass-produced end-to-end reinforcement learning system, capable of self-optimization and strategy exploration in virtual environments.

Currently, 8 of the world's top 10 automakers collaborate with Momenta. Vehicles using Momenta's solutions have driven over 6.5 billion kilometers across more than 160 models globally.

Rumors suggest Momenta has confidentially submitted its IPO prospectus to the Hong Kong Stock Exchange, actively pursuing a listing within the year with an expected valuation exceeding RMB 100 billion.

Meanwhile, Huawei and Horizon Robotics are charging ahead in their respective domains.

In November 2025, Huawei launched ADS 4.0—the world's first intelligent driving system achieving full-scenario "parking spot to parking spot" connectivity. Over 500,000 parking lots nationwide now support this function, with projections to exceed 1 million by the end of 2026.

In April 2026, Huawei unveiled ADS 5.0 and the world's first 896-line LiDAR. By April 2026, cumulative Huawei ADS-assisted driving mileage surpassed 9.56 billion kilometers, with safety mileage under intelligent driving being 4.2 times that of human driving.

Huawei offers three flexible cooperation tiers: full-stack solutions (Harmony Intelligent Driving Mode), system packages (HI Mode deep empowerment), and component supply. This flexibility alleviates automakers' concerns about "sovereignty transfer."

From initial deployment in models priced above RMB 300,000 like Seres, to lowering the threshold to RMB 135,900 through cooperation with Deepal in 2026, Huawei has transformed high-end intelligent driving from a "luxury" to a "standard feature" for mainstream households through rapid technological democratization.

Horizon Robotics dominates the chip track ( track : competitive arena).

This Hong Kong-listed company shipped over 4 million chip solutions in 2025 (+38.8% YoY), generating RMB 3.76 billion in revenue (+57.7% YoY) while maintaining a 64.5% gross margin.

By 2026, cumulative shipments of Horizon's Journey series chips exceeded 10 million units, including 1.8 million high/mid-tier intelligent driving chip solutions (+nearly 5x YoY), accounting for 45% of total shipments.

Horizon's full-scenario urban assisted driving solution HSD entered mass production in November 2025, pioneering urban NOA functionality in mainstream RMB 150,000-level vehicles. In April 2026, the fourth HSD-equipped model—Chery Forthing T9L—delivered approximately 25,000 units just eight weeks after launch. By April 2026, HSD had secured design wins from 10 automakers across over 20 models.

However, the track 's (competitive arena's) brutality exceeds imagination. While survivors race ahead with capital and technological advantages, former star players are quietly collapsing.

Haomo AI, founded in 2019 as Great Wall Motor's intelligent driving R&D unit, reached a USD 1 billion valuation in 2021 to become one of the brightest intelligent driving unicorns. Yet on November 22, 2025, Haomo AI announced a company-wide work suspension.

The turning point came in 2024 when Great Wall Motor exclusively invested USD 100 million in DeepRoute.ai and partnered with DJI Automotive. After losing its core client's financial support, Haomo AI's urban NOA solutions failed to breakthrough. Having long bet on "high-definition maps + rule-based algorithms," it fell behind during the industry's shift toward "mapless intelligent driving."

Haomo AI's collapse isn't an isolated case. Since 2025, HoloMatic has entered bankruptcy proceedings, while Qcraft has faced massive layoffs, reducing its workforce from over 500 at its peak to around 200—a cut exceeding 50%. These intelligent driving unicorns once valued at hundreds of millions or even billions of USD are now falling before commercialization thresholds.

HoloMatic's downfall was particularly emblematic. Founded by Ni Kai, a Baidu Autonomous Driving veteran, and deeply tied to GAC Group, the company reached a USD 1.3 billion valuation after 2023 funding. However, with 93% of revenue dependent on GAC, it failed to secure new funding after major 2024 layoffs, ultimately entering bankruptcy proceedings in February 2025.

Another Baidu-linked casualty was Zhongzhihang, founded by Wang Jing (former Baidu Senior Vice President and first General Manager of Baidu's Autonomous Driving Business Unit).

Zhongzhihang bet on "vehicle-infrastructure cooperation," but this path required government-led infrastructure investments costing billions per city, making short-term commercialization nearly impossible. The company was ordered into liquidation after failing to pay RMB 15,000 in labor arbitration awards.

Jingchi Tech's collapse was equally shocking.

In April 2025, Shanghai's Pudong New Area People's Court formally accepted Jingchi Tech's bankruptcy application. Founded in 2013, the company raised over RMB 2.2 billion and once exceeded a RMB 9 billion valuation as China's earliest mass-producing local Tier 1 ADAS supplier.

Jingchi Tech's fatal flaw was its single-client structure. In 2023, its top five clients (Seres, Seres, Changan, Voyah, FAW) contributed 93% of revenue. When Seres fully switched to Huawei's solutions and Li Auto removed Jingchi from its supplier list, core revenue plummeted.

| The Code of Survival |

Haomo AI's primary lesson was its misguided technological bet.

The company long adhered to "high-definition maps + rule-based algorithms," rejecting end-to-end algorithm proposals in early 2023 as "periodically uncontrollable." A reassessment only came in late 2024, by which time it lagged competitors by at least one technological generation, directly causing core client losses.

After Great Wall Motor partnered with DeepRoute.ai, Haomo AI lost even the opportunity to participate in next-generation model definitions.

Jingchi Tech collapsed due to client concentration risks, while Zhongzhihang failed because short-term commercialization proved impossible—a fundamental conflict with capital expectations for rapid returns.

These failures offer profound lessons. In intelligent driving, a single technological misstep, loss of a core client, or commercial model mismatch can become the final straw.

What, then, did the survivors do right? Three companies offer contrasting survival strategies.

Momenta's logic lies in its positioning (positioning) as a "pure software algorithm supplier."

The company adheres to one principle: no car manufacturing, no hardware production—only providing "the smartest brain." This "Android model" offers clear advantages. It doesn't compete with automakers for hardware profits or force them to buy specific sensors, avoiding brand dominance conflicts. Consequently, Momenta collaborates with over 30 global automakers including Toyota, Mercedes-Benz, and BYD, covering vehicles from RMB 100,000 family cars to RMB 500,000 luxury models.

More critically, it leverages a data flywheel. Mass-produced vehicles continuously collect real-world driving data to fuel algorithm iterations, attracting more automaker partnerships and generating even more data. Once this virtuous cycle starts, latecomers cannot catch up.

Huawei's breakthrough stems from its "no car manufacturing" differentiation. Precisely this positioning allows Huawei to partner with all automakers rather than compete against any.

Horizon Robotics survives through its "software-hardware integration" ecosystem strategy. In 2025, over 95% of chip shipments were co-delivered with ecosystem partners. Horizon competes head-to-head with NVIDIA and Huawei, with the trio controlling ~90% of the mid/high-tier intelligent driving market.

Meanwhile, the rapidly rising DeepRoute.ai is disrupting the "duopoly."

In October 2025, DeepRoute.ai's urban NOA market share soared to 38% in a single month—a 2.7x increase, the fastest among mainstream players. By end-2025, cumulative urban NOA deployments exceeded 200,000 units across 15 mainstream models. Great Wall Motor, one of its core shareholders, invested USD 100 million in 2024.

| One More Round of Carnage Awaits |

The "Matthew Effect" will intensify in 2026's intelligent driving market.

January-October 2025 data shows the top three third-party urban NOA players command 99% market share—Momenta (~54%), DeepRoute.ai (~23%), and Huawei HI (~22%)—effectively clearing Tail enterprises (tail-end companies). Former star unicorns like HoloMatic, Jingchi Tech, and Haomo AI have collapsed, accelerating industry consolidation.

Momenta CEO Cao Xudong explicitly judges: "The 2026 urban assisted driving landscape will be set, with perhaps only 2-3 players remaining in China and 3-4 globally."

Even survivors face no respite.

Momenta contends with scale-versus-profitability paradoxes. Despite its 61% third-party market share, its 414,400 deployments as of November 2025 lag Huawei ADS's 600,000+ units across just 35 models. This suggests Momenta's many partnerships yield few high-volume models, limiting actual deployments.

More critically, end-to-end models' explainability and accident liability remain unresolved. When accidents occur, algorithm decision-making processes are difficult to trace—a potential obstacle to Momenta's expansion.

Momenta's core clients are traditional automakers, but as new forces pursue "full-stack self-research," traditional automakers are building their own intelligent driving teams. Whether these automakers will reduce external supplier reliance remains uncertain.

Huawei faces high-investment and persistent-loss pressures.

Huawei's Intelligent Automotive Solutions BU has invested over RMB 50 billion cumulatively, with 2024 alone exceeding RMB 10 billion. Huawei ADS aims to turn profitable by 2027, but this target remains uncertain amid fierce price wars and rapid technological iteration.

Huawei must also counter competitors on all fronts: Tesla FSD's continuous evolution, XPENG's XNGP iterations, and BYD's rapidly scaling "Divine Eye" self-developed solution. Every rival eyes dominance.

Notably, Huawei uses two data models: Harmony Intelligent Driving Mode (for brands like Seres and Luxeed with deep Huawei ties, using full-stack self-developed solutions) and HI Mode (providing intelligent driving solutions to traditional automakers). In 2025, cumulative ADS deployments reached 1.4 million+ units (including both models), but HI Mode deployments numbered just 134,100 units under third-party supplier metrics.

The competition's decisive factor will shift from "technological existence" to "technological efficiency"—who can deliver better experiences at lower costs? Who can iterate faster? Who can accumulate more road data through more vehicles?

Momenta, Huawei, and DeepRoute.ai lead the first tier. These firms have established "technology-mass production-data" virtuous cycles. Momenta boasts the broadest model coverage and go overseas (overseas expansion) advantages, Huawei leverages Harmony Intelligent Driving's ecosystem closed loop (closed loop) and brand premium, while DeepRoute.ai excels in deep binding strategies and rapid scaling—each with unique moats.

Horizon Robotics, DJI Automotive, and XPENG form the second tier. Horizon dominates chips but needs software algorithm strengthening; DJI targets mid/low-end markets with cost-effective solutions; XPENG persists with full-stack self-research, leading in certain functional experiences.

Overall, China's intelligent driving industry faces one more round of carnage in its survival race.

Image sourced from the internet. Rights reserved for original creator.