"Electronic Screw" Huawei: About to Be Installed in 2 Million New Cars

Gasgoo Munich- At a July 16 media briefing, Huawei's Qiankun unit revealed the scale of its smart-driving rollout: cumulative installations of Qiankun Intelligent Driving have surpassed 1.9 million vehicles and are on track to exceed 2 million by the end of August. The HarmonySpace is poised to reach its second million units around the same time. Meanwhile, the system's accumulated assisted-driving mileage has exceeded 12.8 billion kilometers, with projections exceeding 20 billion by year-end.

However, the speed of this growth matters more than the absolute numbers.

Jin Yuzhi, CEO of Yinwang

It took Qiankun Intelligent Driving 44 months to reach its first 1 million installations, but the second million is expected to take just 12 months. The HarmonySpace is accelerating at a similar pace.

What does reaching that second million in a single year signify? It's about more than simply increasing sales of cars equipped with smart driving. Information disclosed at the event suggests a convergence of factors—expanding partner brands, significant R&D investment, and the rapid accumulation of real-world road data—that are pushing Qiankun Intelligent Driving into a new phase of development. This shift represents the core significance.

Why the rush for the second million?

During the presentation, Yinwang CEO Jin Yuzhi reiterated Huawei's stance in the automotive sector: it will not build vehicles, but instead act as a supplier of intelligent components for intelligent connected cars. The goal is to become the industry's "electronic screw," helping partners build and achieve higher sales of better vehicles.

While that positioning appears unchanged, the scale effects are starting to show as the scope of cooperation widens.

Currently, Huawei Qiankun has partnered with more than 25 automotive brands across over 50 vehicle models, spanning sedans, SUVs, MPVs, and off-roaders, as well as pure electric, hybrid, and ICE segments.

To meet the diverse needs of automakers, Huawei Qiankun has established two collaboration models. The first is full-stack collaboration, involving joint participation in vehicle definition, product development, and quality improvement. The second is component-based collaboration, allowing automakers to select specific solutions like Qiankun Intelligent Driving or the HarmonySpace based on their product positioning, rather than adopting the complete package all at once.

Compared to a single-model approach, this diversified strategy covers more brands and models while lowering the barrier for automakers to adopt intelligent solutions. As the number of partner models grows, so does the installation base for Qiankun Intelligent Driving—hence the marked acceleration from the first million to the second.

From an industry perspective, this reflects the further maturation of the smart-driving supply chain. For a growing number of automakers, full in-house development is no longer the only path; integrating mature solutions can also shorten R&D cycles and accelerate product launches.

Behind the growth in scale lies a battle of long-term investment

The sustained growth of Qiankun Intelligent Driving installations is driven not only by an increase in partner models but also by consistent R&D investment.

Jin disclosed at the event that Huawei has historically invested over 100 billion yuan in automotive intelligence. In 2026 alone, specialized investment in this sector is projected to exceed 18 billion yuan, with 7 billion to 8 billion yuan dedicated solely to cloud AI computing power. Over the past three years, training compute power surged from 2.8 EFLOPS to 60 EFLOPS. Looking ahead, Huawei plans to invest another 70 billion to 80 billion yuan into computing infrastructure over the next five years.

For assisted driving, R&D investment must ultimately prove itself on real roads. The vast amount of real-world data accumulated by nearly 2 million vehicles—covering city streets, highways, and parking lots—provides the foundation for continuous training of Qiankun Intelligent Driving algorithms and product iteration.

Huawei also released safety data: under Qiankun Intelligent Driving's assisted mode, a serious collision occurs once every 8.872 million kilometers on average. Under human driving, supported by active safety capabilities, that figure is once every 5.603 million kilometers. These statistics are derived from anonymous background data and are continuously updated.

In smart driving, installation scale, usage mileage, and algorithm iteration form a virtuous cycle: more vehicles on the road mean more real-world data; that data enables continuous algorithm optimization, which is then fed back to more models via OTA updates. This dynamic distinguishes large-scale smart-driving solutions from competition based on a single vehicle model.

From feature wars to long-term capability

Beyond installation figures, the media day also highlighted the direction of Huawei Qiankun's technological evolution for the next stage.

According to the roadmap, Huawei ADS 5 will see continuous upgrades across four pillars—enhanced perception, deeper cognitive processing, precise control, and higher reliability—delivered through multiple OTA updates. These improvements span perception, decision-making, control, and system reliability. Notably, the company is adhering to its perception strategy that fuses data from cameras, millimeter-wave radars, and LiDAR.

Simultaneously, Huawei Qiankun has achieved "drive-control fusion" by deeply integrating ADS with the XMC vehicle control system, allowing it to handle extreme scenarios like tire blowouts or low-traction surfaces. Regarding system reliability, Qiankun OS implements deterministic latency scheduling, while the hardware architecture adopts full-link redundancy. The target for safe parking rates is "8 nines" (99.999999%).

On the usage ecosystem front, Huawei Qiankun now supports door-to-door capability at 1.1 million parking lots, payment services at 300,000 lots, access to 2 million charging piles, and over 3,000 car wash shops, gradually building a "one-stop parking, washing, and charging" lifestyle ecosystem.

In the HarmonySpace, the Celia voice assistant has been awakened 6.5 billion times and is evolving from a "professional assistant" to a "dedicated butler." Officials also outlined the latest OTA schedule: the HS 5.2 version will roll out in July, optimizing the intelligent assistant and contextual interactions; the all-new HS 6.0 cockpit version will arrive in September or October, upgrading in-cabin fusion perception and introducing new themes; and AI proactive suggestions 2.0 will be released by the end of the year.

The Yijing model will be the first to feature these new capabilities, including HarmonySpace 6 and the AMS in-cabin fusion perception system, offering a preview of the next-generation cockpit interaction.

Regarding the trajectory of autonomous driving, Huawei Qiankun outlined its rollout timeline: completing highway L3 pilot qualifications and launching urban low-speed L4 pilots in 2026; achieving mass commercialization of L3 in 2027; and realizing mass commercialization of urban low-speed L4 in 2028.

These developments represent more than just adding individual features; they involve the continuous refinement of the entire intelligent driving system. In past years, competition among smart vehicles centered on whether they had smart driving, large models, or LiDAR. But as intelligence becomes ubiquitous, the focus of competition is shifting.

For suppliers, the battle is no longer just about how many new features can be launched, but whether they can sustain the stable operation of large-scale vehicle fleets and continuously iterate products based on real-world driving data.

The shift from 44 months to complete the first million to a projected 12 months for the second signals more than just growth in installation figures for Huawei Qiankun; it reflects the entry of smart-driving solutions into a stage of mass application. For the entire industry, the significance of this data lies not in a new numerical milestone, but in the fact that as the user base expands, the competition for smart driving is gradually extending from product capabilities to long-term strengths like R&D, data, safety, and ecosystem.