The 15th Five-Year Plan period marks the decisive stage for China to achieve the Long-Range Objectives for 2035 and build itself into an automotive powerhouse. From April 11 to 12, the High-Level Forum on Intelligent Electric Vehicle Development (2026) was held under the theme of Intelligence, Green Development, Integration and Globalization, with in-depth focus on the global industrial landscape, AI+automotive, chip computing power and other cutting-edge technologies. Through cross-sector exchanges, the forum aims to clarify industrial thinking, build a sound and orderly policy and market system, strengthen global collaboration, and enable China’s new energy vehicles to achieve steady and long-term growth during the critical window of high-quality development.
Jin Yuzhi, Senior Vice President of Huawei and CEO of Yinwang Automotive, delivered a speech at the forum. He noted that in 2025, the penetration rate of L2 assisted driving in China’s new energy vehicles priced above 100,000 yuan exceeded 90%, and the pilot access for L3 autonomous driving is progressing in an orderly manner. Meanwhile, relevant legislation in the United States is accelerating, speeding up the global commercialization of autonomous driving. 2026 is expected to become the first year of global autonomous driving.
Jin Yuzhi reaffirmed Huawei’s strategic positioning: Huawei will never make cars itself, but is committed to being an electronic component provider in the era of intelligent connected vehicles, empowering automakers based on 30 years of ICT technology accumulation. At present, Huawei has partnered with more than 25 automaker brands, launched over 35 mass-produced models, and equipped more than 1.4 million sets of assisted driving systems in total.
Looking ahead, Huawei put forward two core viewpoints:First, L3 is an indispensable stage toward high-level autonomous driving. Full-scenario L3 rollout for consumers should be accelerated to accumulate data and build trust; for L4, orderly verification is recommended starting from B2B and low-speed scenarios.Second, the current iteration cycle of intelligent hardware (2–3 years) is seriously mismatched with the vehicle lifecycle (10–15 years). Industry consensus and policy support are urgently needed to enable hardware upgradability and meet users’ demand for always-new experience.
Full Speech Transcript
Today I would like to share with you how to accelerate the transition from assisted driving to autonomous driving in the era of electrified future mobility.
Regarding the popularity of assisted driving in China’s new energy vehicles in 2025, the penetration rate of L2 and above assisted driving in models priced above 100,000 yuan has exceeded 90%. Moreover, the higher the vehicle price, the more prevalent such configurations are; models priced above 500,000 yuan are basically equipped with L2 and above functions as standard.
Meanwhile, China has been promoting the conditional approval of L3 autonomous driving, with relevant policies, laws and regulations advancing steadily. Last year, two models were approved for road trials with conditional L3 assisted driving. Starting this year, L3 models will be approved and allowed on roads in batches.
This year, the United States is accelerating the enactment of legislation related to autonomous driving, which is now under a full vote in the House of Representatives. The commercial deployment of autonomous driving systems in the U.S. is accelerating. Two key points in the U.S. Autonomous Vehicle Safety Act deserve special attention: one is allowing automakers to increase the number of autonomous vehicles on roads to 90,000 per year; the other is transforming the fragmented regulation across 50 states into unified federal management. Once these provisions take effect, the commercialization of autonomous driving in the U.S. is expected to advance rapidly.
Judging from the progress in China and the U.S., two major tech powers, we can conclude that 2026 will mark the beginning of the global autonomous driving era.
Huawei’s Strategy in the Automotive Industry
Huawei’s strategy in the automotive industry is clear, consistent and unwavering. We are committed to being an electronic component provider in the era of intelligent connected vehicles. We will not compete in the industry, but empower it. Our positioning is to help global automotive partners build and sell better vehicles — we do not manufacture complete vehicles.
With this positioning, our business boundaries are very clear. Over the past 30 years, Huawei has deeply cultivated ICT technologies and become a world-leading global ICT supplier, with absolute leading positions in 5G, optical communications and data communications. We apply these core technologies rooted in decades of ICT expertise to the automotive industry.
For example:
Huawei’s QIANKUN intelligent assisted driving system uses on-board computing chips derived from Huawei’s Ascend AI chip accumulation in computing.
Technologies behind the Harmony cockpit are extended from Huawei’s decades of experience in consumer mobile terminals — the in-vehicle infotainment system can be seen as a large smartphone with a big screen and tablet functions.
Vehicle control technologies draw on our leading data communication expertise.
In-vehicle optical technologies are adapted from Huawei’s globally leading optical communication technologies.
Our positioning and business scope are clear: apply our 30 years of core technologies and experience in delivering high-quality, highly reliable products to the automotive sector.
Our principle is to be user-centric and create value for customers. We provide end consumers with more advanced assisted driving and future autonomous driving systems, as well as user-friendly cockpit systems. We see ourselves only as an electronic component provider — our success depends entirely on the success of our automotive partners. Only when our partners achieve volume growth and market success can we succeed.
To meet the needs of automaker partners, we offer diverse cooperation models:
Full-stack solutions
Dual-intelligence or triple-intelligence solutions
Pure component supply
For full-stack cooperation, we have in-depth partnerships with JAC Zunjie, Seres AITO, Shanjie, Zhijie, Xiangjie, Avatr, Dongfeng Yijing and GAC Qijing. We participate in product definition and help improve overall vehicle quality based on partners’ needs.
We also provide triple-intelligence solutions to FAW Hongqi and Voyah. We welcome partners that only adopt single-intelligence modules, such as supplying assisted driving systems to Audi, and providing Harmony cockpit components for Nissan and Toyota models.
Driven by this user-centric strategy of empowering partners to build and sell better cars, by 2025 Huawei had cooperated with more than 25 automaker brands, launched over 35 mass-produced models, and equipped a cumulative total of 1.4 million assisted driving systems by the end of last year over the past six to seven years, with rapid growth momentum. We sincerely thank users for their trust in our assisted driving systems and automaker partners for their support.
Safety and Quality: The Foundation of Trust
Safety and quality are the cornerstones of consumer trust and the fundamental pillar of automotive intelligence. Without this pillar, everything else is meaningless. We make no compromises on safety.
We have continuously iterated our systems:
2022: Launch of the first ADS version
2023: ADS 2.0, enabling large-scale commercial use in urban areas nationwide
2024: Pioneered parking valet capabilities from spot to spot
2025: Released the brand-new WEWA world model architecture and upgraded Spot-to-Spot 2.0
Each year we drive the rapid development of assisted driving in China’s automotive industry, and our new technologies become benchmarks for the whole sector. In this process, we have gradually built user trust. Today, more and more car buyers clearly recognize: when buying a car, choose one with assisted driving — and for assisted driving, choose Huawei QIANKUN, especially for high-end vehicles.
Last year’s ADS 4.0 adopted the new WEWA architecture, featuring a world engine on the cloud and a world behavior model on the vehicle side. Last year, cloud computing power reached 4.5 exaFLOPS, an extremely powerful computing system. Through generative diffusion and safety reinforcement, combined with multi-modal perception, native foundation model training and MoE (Mixture of Experts) systems, we deliver smoother and safer driving experiences.
After the upgrade to ADS 4.1 before last Spring Festival, users gave highly positive feedback. During the holiday, users accumulated massive assisted driving mileage. In the past two weeks, we held the Huawei QIANKUN AD Open Competition, which was extremely popular with 150,000 participants. To enter the top three, contestants had to drive over 17,000 kilometers in two weeks — more than 1,000 kilometers per day, counting only urban mileage, not highways. This public open competition verified the safety, stability and user experience of the ADS system. We will hold such competitions annually to let users enjoy the system to the fullest.
Multi-Sensor Fusion for Uncompromising Safety
Beyond leading algorithms and computing power, we insist on a multi-sensor fusion perception solution because safety and quality always come first.
High-definition cameras excel at understanding environmental elements
LiDAR delivers accurate ranging and distance resolution under varying light
Millimeter-wave radars perform well in rain, fog and speed measurement
No single sensor is perfect, but multi-sensor fusion enables 360-degree coverage without blind spots. Contrary to the misunderstanding that QIANKUN AD only uses LiDAR, our vehicles are equipped with up to 11 high-definition cameras for full front, rear, surround and side view coverage.
We also pioneered distributed 4D millimeter-wave radar, derived from Huawei’s 20+ years of RF expertise in distributed base stations. LiDAR has evolved from 96-line to 192-line and now to 896-line, launched in March this year, delivering four times higher resolution than 192-line LiDAR, reaching image-grade performance. It can accurately detect small targets above 14 cm at a distance of 120 meters, laying the groundwork for future high-speed autonomous driving.
We also introduced the industry’s first in-cabin LiDAR-camera hybrid system (LIMERA), integrating camera and LiDAR in a quiet, weather-resistant module.
On safety, we keep advancing from CAS 2.0 to 3.0 and 4.0, achieving full-speed, full-direction, full-target and full-scenario protection, elevating vehicle active safety to a new level. By the end of March this year, CAS 4.0 had helped avoid 4.79 million potential collisions.
Starting this year, we publish the Huawei QIANKUN AD ADS Safety Mobility Report monthly on our official website.
Total assisted driving mileage has exceeded 9.5 billion kilometers and will surpass 10 billion kilometers this year.
The national average for severe collisions (seatbelt pretensioner deployment or airbag ignition) is once every 1.8 million kilometers for human driving.
For vehicles with Huawei ADS (even under human control), severe collisions occur once every 5.17 million kilometers — 2.87 times safer than human driving.
Under full ADS assisted driving, severe collisions occur once every 7.57 million kilometers — 4.2 times safer than human driving.
We publish this report to present transparent, verifiable data to users. Data never lies. We call on the whole industry — automakers and autonomous driving system providers alike — to publish safety data openly, rather than just marketing claims. This also helps the industry accumulate data to determine when full autonomous driving becomes feasible.
Two Key Proposals for the Future
We are ready to join hands with industry partners to advance the intelligent development of the automotive industry. We put forward two core proposals:
1. L3 is an indispensable step to L4/L5 full autonomous driving
L3 cannot be skipped.
Safety perspective: Safety must be quantified with public data. To achieve L4 fully driverless operation, safety needs to be at least 10 times that of human driving, requiring massive data accumulation.
User perspective: Users need time to accept and adapt to a vehicle without a steering wheel or human driver. Even today’s L2 systems are not fully trusted by many users.
Regulatory perspective: Laws, regulations and insurance systems need experience and data to mature.
A critical shift occurs from L2 to L3: L3 is conditional autonomous driving, where liability for accidents shifts to automakers and autonomous driving providers, not the user. L3 keeps the driver in the seat with extended takeover time, acting as supervised autonomous driving for safe validation.
We recommend accelerating L3 deployment directly to consumers in full scenarios and regions to accumulate data and build trust.
For L4, despite technological leadership, we suggest initial validation in B2B scenarios such as Robotaxi and limited low-speed environments (e.g., 40 km/h on public roads, 20 km/h in campuses), ensuring safe and orderly industry progress.
2. Resolve the lifecycle mismatch between intelligent hardware and vehicles
Intelligent hardware evolves every 2–3 years following Moore’s Law, while vehicles have a 10–15 year lifecycle. Software can stay fresh via OTA, but intelligent hardware cannot yet be upgraded like batteries. Users expect hardware to be upgradable for 2–3 generations during vehicle ownership.
We call on the industry to reach consensus and promote policy and regulatory support to enable intelligent hardware upgradability, fostering the healthy development of the intelligent automotive industry and better meeting user expectations.
Huawei QIANKUN AD Vision
Huawei QIANKUN AD aims to bring intelligence to every vehicle, make travel safer and life better. We stand ready to work with all industry partners to advance the healthy, high-quality intelligent development of China’s automotive industry, accelerate the arrival of full autonomous driving, and lead global development in a safer manner.