A few days ago, Qualcomm released a white paper.
The white paper is called “Snapdragon Ride: A solution to promote the popularization of assisted driving among Chinese automakers and consumers.”
That’s right, this document talks about Qualcomm’s Auto assisted driving business.
In many people’s impression, Qualcomm seems to be equated with the smart cockpit most of the time. There are not many manufacturers who dare to take risks on cockpit chips in all kinds of new energy trams on the market. Snapdragon 8295 is synonymous with flagship smart cockpits in the current market. Some flagship models even use 2 8295.
However, in this white paper, Qualcomm told us a somewhat unexpected story. It turned out that Qualcomm has quietly made efforts in the field of assisted driving. It turned out that more than 20 car companies on the market have announced the launch or are developing models with ADAS functions based on the Snapdragon Ride platform. Qualcomm seems to be becoming a force that cannot be ignored in this wave.
After reading this white paper, I became interested in the intelligent experience empowered by the Snapdragon Ride platform.
So, on June 26, Xiao Lei took a flight to avoid the typhoon and went to Suzhou where the weather was calm and rainy, and represented Lei Technology to attend the Qualcomm Automobile Technology and Cooperation Summit.
(Photo source: Lei Technology)
According to the information provided by the official site, the theme of this summit is “We will work together, move steadily and far away”, with the purpose of presenting exciting content covering the full stack technology of smart cars, the ecology of the entire industrial chain and the full scene experience.
As for what I can see on the scene and what I have experienced here, please listen to Xiao Lei tell me.
Two supreme version platforms, shockingly debuted
In the main speech at this summit, Qualcomm teamed up with advanced Chinese automakers and ecosystem partners to demonstrate the development momentum and latest achievements of its Snapdragon digital chassis product portfolio.
The core of these achievements is the new Snapdragon Cockpit Supreme Edition platform and Snapdragon Ride Supreme Edition platform released by Qualcomm at last year’s Qualcomm Snapdragon Summit.
(Photo source: Lei Technology)
Unlike the past situation where the automotive chips lag behind the mobile side on Snapdragon 8295 and Snapdragon 8155 products, the new generation of Snapdragon cockpit supreme version platform has been tied with the mobile side for the first time, and simultaneously used the latest Oryon CPU architecture, achieving a huge leap from N-3 generation to N generation.
Therefore, in terms of performance, the Snapdragon Cockpit Supreme Edition has produced a leap-level performance improvement:
CPU performance improvement by 300%
GPU performance improvement by 300%
NPU performance improvement by 1200%
In addition, this powerful cockpit platform has also made special expansion and optimization for smart car scenarios. At the moment when high-end intelligent driving in China is in full swing, Qualcomm has further upgraded the Ride intelligent driving platform – Snapdragon Intelligent Driving Supreme Edition platform, and started the process of cabin integration, helping the automotive industry move towards an intelligent future.
Many car companies said on the spot that they planned to use the Snapdragon Auto Platform Supreme Edition in their upcoming models.
The first one to go on stage was Zhu Jiangming, CEO of Leapmotor. After briefly emphasizing his own achievements and the technical foundation of the whole region, he began to introduce his cooperation with Qualcomm. It is reported that Leapmotor’s latest models basically use the Snapdragon 8295+8650 packaged in a box, which is considered a manufacturer with far-reaching cooperation with Qualcomm in the industry.
(Photo source: Lei Technology)
Leapmotor said that Leapmotor flagship D series models will be mass-produced in the first quarter of next year. The central domain controller is equipped with two Snapdragon 8797 chips, a single-chip calculation of 640TOPS, one for smart cockpit and the other for assisted driving.
Wang Chendong, deputy general manager of SAIC and GM, also came to the stage to support Qualcomm. He said that GM will build its own cockpit and assisted driving system around Qualcomm. In 2025, two heavy-duty models will be built around SA8775 and will be upgraded to SA8397 and SA8797 in 2026.
(Photo source: Lei Technology)
From this point of view, it is time to accelerate the intelligence of joint ventures.
The competition in the car circle is fierce, and the supply chain is in full swing
In addition to the main forum, Qualcomm has also set up two major display areas for partners to display their products.
When you step into the display area, what catches your eye is a U8 that Meg Intelligently set up.
The car is equipped with Meige Intelligent’s intelligent cockpit module, based on the first generation of Snapdragon 8+ mobile platform, integrated 8-core Kryo 780CPU and Adreno730 GPU with 4nm process, supports OpenGL ES 3.2, Vulkan1.2, OpenCL 2.0, and integrates Qualcomm neural network processing unit, with AI computing power exceeding 27 Tops.
(Photo source: Lei Technology)
There were not many people on site, and after a simple queue, I was able to enter the cockpit to experience it.
It is still quite shocking to look up at the U8 cockpit. After the on-site experience, the multi-modal interaction feedback is very fast. Whether it is voice control or gesture adjustment, it is smooth enough, and there is no stickiness or no feeling of following the hand. It can even provide AI intelligent body capabilities to achieve multi-step operation of three-party applications through voice commands.
(Photo source: Lei Technology)
According to the booth staff on site, other cockpits displayed in the same venue include the automatic parking function of the cabin and the offline large-scale voice assistant. The former cannot be experienced on site. I briefly tried the latter and it is basically the Qwen 2.5-3B deployed offline, and it is still OK.
In addition to Meige Intelligence, manufacturers including Bosch, Guangtong Yuanchi, Junlian Intelligence, Yiyuan Communications, and Chengmai Technology have all displayed cockpit systems built on the Snapdragon Automobile platform. In addition to operating animation demonstrations, local big models have also become one of the hot topics today.
(Photo source: Lei Technology)
To talk about the current experience of this smart cockpit, of course, it is still inseparable from the entire vehicle audio system.
Bose and Brisonus, who are familiar to everyone, came to the scene. Although there was no environment to truly deploy an independent audio experience room, they could still understand them and Qualcomm’s on-board surround sound system from the on-site introduction.
(Photo source: Lei Technology)
As for the cabin integration, there are also plans made by several companies such as Zhuoyu, Desai SV, Visteon, Cheliantianxia, and Zhongke Chuangda.
Basically, each company has solutions based on the next-generation 8775 (5nm) and next-generation 8797 (3nm) platforms, and they all realize the two major functions of smart cockpit and assisted driving, that is, the integration of cabins.
For example, in Chelian Tianxia, they have a solution to use the Snapdragon 8775 platform and only do assisted driving. In this way, it is enough to get the computing power of 70TOPs-96TOPs.
(Photo source: Lei Technology)
As for Visteon, they only use Snapdragon 8775 to do AI and not assisted driving, so that they can deploy multimodal large models locally, and even connect to Doubao.
It is not difficult to see that Qualcomm’s on-board chips are very elastic, and can allocate different computing power to the cockpit and intelligent driving, and implement different solutions. Such a design can help Internet car companies move faster in cockpit and assisted driving. Perhaps next year, cars with a standard number of less than 300,000 yuan will be equipped with local AI assistants, high-speed assist and assisted parking functions.
It’s time to fully realize “cabin fusion”
For Qualcomm, the leader role in the smart cockpit chip market seems to be far from enough.
As the saying goes, the focus of the second half of the intelligence competition is on the driving field. The various cooperative displays in the past two days, and the release of these two chips in the domestic market, hides Qualcomm’s biggest ambition, namely, to open up the cockpit domain and the driving field.
Unlike the assisted driving computing platform in the traditional sense, by concentrating IVI (infotainment) and assisted driving tasks on the same SoC, Snapdragon Ride Flex successfully breaks the traditional functional domain division, increasing integration while reducing the system’s bill of materials, and making costs more competitive.
(Photo source: Lei Technology)
Not only that, the virtualization engine can support the simultaneous operation of multiple key levels of tasks, and achieve “uninterference” functional partitioning at the resource management and system scheduling levels, ensuring that security tasks can be continuously run even in a hybrid computing environment. The embedding of generative AI also introduces more active intelligence and user personalization capabilities to assisted driving systems.
This design provides Chinese automakers with the possibility of building an “economy, high-function vehicle” platform based on a single chip, allowing them to create more competitive differentiated and intelligent models without repeatedly investing too much resources in the hardware architecture.
In my opinion, the addition of these top manufacturers is the best proof of the strength of the Snapdragon Ride platform.
Nowadays, various chip manufacturers are trying to work in this direction.
Nvidia Thor, which has experienced many times of delays, has a design concept that it natively supports all functions of a single SoC-driven vehicle, including advanced autonomous driving, parking, driver monitoring and cockpit functions; Intel and Black Sesame Intelligent Cabin Fusion Platform Reference Design, which will be released in the second quarter of 2025.
Of course, these products are currently in a state of “promising future”. Thor has not yet confirmed that the low-end version will be used on the Xiaomi YU7 until recently; and Intel… apart from a reference design with unknown prospects, the automotive business is now full of fog.
From this perspective, Qualcomm is actually a relatively more down-to-earth side.
It can be expected that with the gradual application of these products, Qualcomm Snapdragon will inevitably face Nvidia’s competition.
After all, the trend of cabin fusion has become a consensus.
