企業網站
Lately, BYD’s chairman Wang Chuanfu has been making late-night calls to the head of the company’s new technology institute. Their conversations focus on one critical subject: “intelligentization.” This shift is notable, as their discussions previously centered more on powertrain technologies.
This change signals a clear evolution in BYD’s technological strategy. Having secured its leadership position in the new energy vehicle (NEV) market through its supply chain, BYD is now charging forward into the realm of intelligentization.
The day after the Denza Z9 GT presale launch, 36Kr interviewed Yang Dongsheng, the director of BYD’s new technology institute. Yang, who joined BYD in 2005, has held various roles, including deputy manager of the chassis department as well as general manager of product and technology planning. He now serves BYD as a vice president and is a key figure in driving the company’s technological advancements.
During the interview, Yang emphasized that intelligentization is now BYD’s top priority.
The development of advanced driver assistance systems (ADAS) has marked the beginning of the second half of the intelligentization era. From urban navigation to the capability of driving on any road, and from integrating battery electric vehicles and transformer technologies to deploying end-to-end (E2E) large models in vehicles, every automaker now sees ADAS as the key to winning the next round of competition. No one wants to be left behind.
Even BYD, once criticized for lagging in this area, is now making significant strides in intelligent driving. Unlike new entrants that focus on software algorithms and computing power, BYD builds its intelligent driving capabilities on top of its mature electrification technologies.
At a recent launch event, Yang introduced BYD’s definition of intelligent driving. While many automakers have achieved human-like driving assistance in urban environments—a limited form of autonomous driving—BYD aims for a broader approach. This includes the Xuanji architecture and electrification, which together can achieve a superior level of driving assistance.
Yang provided an example to illustrate this: blowout control. BYD assesses the vehicle’s posture in the event of a blowout and controls wheel speed, suspension, and tire pressure through the vehicle’s controller. “We truly achieve end-to-end control, from tire sensors to the brain’s response and strategy, down to the tire’s control,” Yang explained. “Others may stop at sensing and control—we go from actuator to actuator.”
BYD’s intelligent driving team is over 4,000 strong, far exceeding the industry average. However, Yang clarified that the core algorithm team is made up of about 1,000 members. The larger team size reflects BYD’s deep integration of its supply chain. “We even handle the UI design for cabin matching in-house,” Yang said.
Looking to the future, BYD’s goal is to make ADAS widely available in lower-end models. Yang believes that reducing traffic accidents and increasing safety are essential needs, and BYD aims to make these benefits accessible to more consumers.
“In the next two years, advanced driver assistance systems will be integrated into low- to mid-end models,” Yang said. Following market research, BYD plans to identify the most commonly used ADAS features by consumers and incorporate them into a comprehensive solution for lower-end models.
While BYD is advancing its intelligent driving capabilities, it is also enhancing its system-level capabilities, leveraging its strong vertical integration.
Yang explained that BYD takes a holistic approach to system design, avoiding a parts-centric mindset. He emphasized that future innovation must come from system-wide innovation, which is one of BYD’s strengths. “Other automakers lack both a supply chain and a system innovation team. Our system innovation team has 15,000 members.”
Using blowout control as an example, Yang highlighted that all the hardware design, software design, communication protocols, and algorithms were developed by BYD. “No suppliers were involved—the electronic control software and intelligent driving software were all developed in-house.”
R&D is the bedrock of BYD, and Wang remains actively involved in the frontlines of technology. According to Yang, Wang’s recent focus has been on intelligent driving, requiring vehicle design to align with intelligent driving standards.
“When we design vehicles or electronic and electrical architectures, many details need to be adjusted according to intelligent driving standards. Our technical team previously found it challenging to promote standardization and platformization, but Wang oversees these efforts directly,” Yang said.
Earlier in 2023, during an earnings call, Wang stated that the next three years would be a decisive battle for scale, cost, and technology, predicting that the market share of joint venture brands would drop to 10% within 3–5 years.
To win this battle, maintaining technological leadership is essential. In 2023, BYD’s R&D investment reached RMB 39.918 billion (USD 5.6 billion), doubling from the previous year. Yang revealed that BYD has established a R&D system that spans from fundamental research to pre-research and product conversion.
With competition getting ever tighter in the automotive industry, Yang relayed Wang’s expectations: “To stay ahead in this marathon, we must continuously maintain technological leadership. Deep technology reserves and a diverse strategy are key.”
The insights above were derived from an interview conducted by 36Kr with Yang. The following transcript is a translation of that interview and has been edited and consolidated for brevity and clarity.
36Kr: How long did it take to develop the Yisifang technology? How many of such projects does BYD undertake each year?
Yang Dongsheng (YD): Some components took three years to develop, but the entire system has been in the works for even longer. When we took back Denza, we needed to find a technical direction for it, and that process took about five to six years.
Every brand needs a technological label, and Denza’s image is that of a luxury car. At that time, we already had Yisifang, as well as DM and pure electric platforms, so we had to determine which direction Denza should take. We noticed that BBA (BMW, Mercedes-Benz, and Audi) not only had luxury branding but also a dedicated user base, so we identified a unique characteristic for Denza.
Since Yisifang was already being used in the Yangwang brand, which strives for extreme performance at a high cost, Denza needed to make performance its selling point. Enhanced vehicle handling can also solve safety and control issues, so we decided on a three-motor solution.
But three motors alone aren’t enough. To improve handling, the rear wheels must have dual steering. After much analysis by our PhD researchers in vehicle dynamics, it was clear that relying solely on front-wheel steering wasn’t sufficient. Independent rear-wheel steering was essential to achieve the extreme performance we aimed for, leading to the development of this powertrain solution.
36Kr: You mentioned that vehicle departments propose requirements, and then you innovate the technology, which also involves the battery and other departments. How does BYD coordinate among different departments?
YD: BYD is quite unique in this regard, which might be hard for outsiders to understand. For instance, our new technology institute is responsible for technological innovation, like DM. I come from a powertrain background, and we work on intelligent technologies, new electric platforms, and more.
We operate like a system research institute, developing system solutions. The components departments are responsible for implementation—once we complete a system solution, the components departments handle the design and structural development. They act as our suppliers, so to speak.
36Kr: BYD’s intelligentization is often said to be built on its advantages in electrification technology. But so far, the differentiation resulting from this combination is not very apparent compared to other automakers. Why is that?
YD: Actually, we have already showcased this. Take blowout control, for example—this is a typical example of the integration of intelligentization and electrification, and it’s a unique differentiator for BYD. Only BYD can do this because we control all the hardware design, software design, communication protocols, and algorithms ourselves. All the code and strategies for the vehicle controller are defined by us, with no involvement from suppliers. We integrate intelligent driving software with the vehicle controller, making it a single ‘brain.’ This level of integration—from sensing to control—is something no other company has.
For instance, in the event of a blowout, you need to assess the vehicle’s posture and the changes in tire characteristics. Through the vehicle controller, we manage wheel speed, suspension, and tire pressure, gaining a clear understanding of the blowout’s impact on traction. We then make adjustments accordingly. We truly achieve E2E control—from tire sensors to the brain’s response and strategy, down to tire control. Others may stop at sensing and control. We go from actuator to actuator.
36Kr: Wang Chuanfu mentioned that BYD’s intelligentization team has 4,000 people, while Tesla only has 200. Can BYD’s strategy of leveraging large numbers in ADAS development still be effective? How is the advantage of team size maximized?
YD: BYD has many vehicle models, and our deep supply chain integration also requires a larger workforce. For example, developing domain controllers with high-computing AI chips alone requires about 600–700 people. Huawei also has a similar setup, but structures its team differently. We even handle the development of the base software and middleware in-house, which is included in the 4,000-person team. The core algorithm team doesn’t need to be that large—our team of about 1,000 people is sufficient.
However, BYD has many models, and ADAS development is particularly labor-intensive. It requires a lot of manpower for engineering implementation. We currently have 40–50 vehicle models with different configurations that need to be developed and integrated with the cabin. Unlike other companies that outsource to external suppliers, we handle even the UI design for cabin matching in-house.
Additionally, when it comes to vehicle-level coordination, like the moose test we showcased, it requires the entire vehicle to act as a brain, sending numerous signals to the powertrain and suspension systems. This workload is significant and sets us apart from others.
36Kr: You mentioned E2E earlier. Once E2E models are deployed in vehicles, doesn’t that reduce the need for as many people? Wouldn’t future competition focus more on algorithms, computing power, and data? Will we see adjustments in the algorithm team?
YD: This mainly concerns the software team members, but they can focus on other tasks. In the early stages, E2E perception models—whether using LiDAR, cameras, or millimeter-wave radar—are all about perception. But how do we handle the integration of other sensor inputs?
The vehicle also has signals from the motor and suspension systems. With so many signals, we can’t rush into full integration. We have to create a smaller, initial network and experiment with it. The software team can continue exploring by integrating various past models. For example, we’re now adding post-fusion capabilities, and we’re shifting talent to explore post-fusion integration.
36Kr: How does BYD view the importance of intelligentization internally? What is Wang’s attitude toward ADAS?
YD: It’s highly prioritized—our top priority. Now, when I get calls late at night, Wang is asking about ADAS. In the past, he might have asked more about powertrain details, but now it’s all about intelligentization.
36Kr: Does Wang set any specific goals for you? For example, when will the in-house developed ADAS be available in vehicles, or when will advanced ADAS be available in lower-end models?
YD: Our goal is clear: currently, advanced ADAS is available in vehicles priced around RMB 200,000 (USD 28,000) or above. BYD wants to make advanced ADAS available to ordinary consumers. Reducing traffic accidents and increasing safety are essential needs, and we hope to bring these benefits to lower-priced vehicles, even those priced below RMB 100,000 (USD 14,000).
So, our future goal is to leverage BYD’s vertical integration capabilities and algorithm improvements to ensure that advanced ADAS is available in low- to mid-end vehicles.
36Kr Are there set timelines, such as when models like the Seagull, an entry-level vehicle, will be equipped with advanced ADAS?
YD: For low- to mid-end models, we will definitely achieve this within two years because there’s a process involved. First, consumers’ understanding of ADAS is evolving. We believe we are currently experiencing a second ‘handshake’ between ADAS and consumers. The first handshake, which involved concepts like autonomous driving and unmanned driving, felt very distant to consumers and was ultimately unsuccessful.
This second handshake focuses on the capabilities of advanced ADAS, such as automatic emergency braking (AEB) and parking assistance, which we believe is successful. Consumers are now seeing that advanced ADAS can improve safety and convenience.
At this point, we can precisely identify which advanced ADAS features consumers like and which are merely attention-grabbing. We’ll identify the 20 most popular features from an initial set of 100 and focus on perfecting those before compressing the overall solution to fit low- to mid-end models.
36Kr: Wang has always been active in frontline management. It’s said that he used to personally decide how battery cells should be placed. Does he still get involved in such detailed decisions?
YD: What you mentioned isn’t even detailed enough.
36Kr: How detailed does he get?
YD: For example, with our intelligent driving solutions, he has been paying close attention to how these can be made more scientific. BYD is now pushing for platformization and standardization, not just in sensor configurations but also in braking, steering, and driving systems. Even the placement of components must be standardized.
Now, when designing vehicles or electronic and electrical architectures, many details need to be adjusted according to intelligent driving standards. Our technical team previously found it challenging to promote standardization and platformization, but Wang directly oversees these efforts.
36Kr: BYD has developed its ADAS in-house while working with external suppliers. How are these approaches balanced? When will in-house developed solutions be deployed in vehicles?
YD: ADAS algorithm iterations are happening very quickly. It’s still uncertain which company will lead in the future. BYD has many vehicle models, so we are standardizing and platformizing our entire vehicle lineup. This way, more algorithm companies can join in. We maintain an open stance—we want to be strong and differentiate ourselves, but we also can’t close ourselves off. Collaboration is essential to identifying the best solutions.
BYD is like a training ground. If a supplier can secure orders and sustain cooperation with BYD, it proves that the company is strong enough. BYD has no reason to abandon its best partners because, ultimately, we want the best products.
36Kr: BYD is currently engaged in a three-year battle. While other automakers seem to be struggling, BYD still has some room to maneuver. In this situation, what approach will it take to win this battle?
YD: The market is highly dynamic, with competition over technology, configurations, pricing, and even company leadership. This is a manifestation of market prosperity. Chinese industries, including the internet, mobile phones, and home appliances, have all gone through similar processes. The difference now is that we live in an era of information proliferation. Past competition might have been less perceptible to consumers, but now they are more aware of it.
Wang Chuanfu said that at this moment, deep technology reserves and a multitude of strategies are crucial. First, we must maintain our technological leadership. BYD has transformed everything, from powertrains to chassis to intelligentization systems. This is not a simple replacement but a complete transformation. We must continue to solidify our foundation, and we have invested in talent and indefinite R&D to prepare us.
36Kr: What is BYD’s tolerance for errors in technology development?
YD: We categorize this into two types. One is fundamental research that may not be commercialized in the short term—we call this mechanism research. Mechanism research doesn’t require tolerance for errors because it’s about scientific discovery, and it can keep your products ahead for a long time.
If it’s product engineering leadership, it’s relatively easy to be surpassed, including software algorithms. Software algorithms, at best, lead for six months. That’s why we invest heavily in fundamental research. Internally, we don’t assess these projects.
36Kr: Can BYD keep investing indefinitely?
YD: I wouldn’t say indefinitely, it’s more about encouragement. Our product teams and pre-research teams are incentivized. If you conduct research or pre-research that leads to a product, your recognition will increase significantly.
So, those involved must consider market needs. Internally, there’s a self-driven motivation because the ultimate recognition comes from your research being applied in products. I can wait two years, even three years, but whether it takes two years or ten years for your research to be implemented in a product will determine your sense of accomplishment and recognition.
36Kr: The fifth-generation DM system relies on the integration of multiple components to achieve such efficiency. After battery charging speed is improved, will this system still have room for further optimization? Is this also the direction for the exploration of the sixth-generation DM?
YD: Yes, you’re right on point. But we take a holistic approach, rather than focusing on individual components. Future innovation can only come from system innovation. It’s difficult to innovate with a single component, and this is BYD’s advantage. Other automakers lack both a supply chain and a system innovation team, while our system innovation team has 15,000 members. The future lies in optimizing efficiency under real-world conditions because even if you achieve high efficiency in a single engine, consumers may not be able to experience it.
When you consider the conversion between mechanical energy, chemical energy, and electrical energy in a car, the conversion efficiencies vary, and the component systems differ. That’s why we need to build a complex model, something that’s not visible to outsiders. We have a huge simulation team that models these components, so when we need to modify something, the team can make changes directly, and we can see how the system’s direction changes after the modifications. This system capability is something others can’t see within BYD.
36Kr: When developing the fifth-generation DM system, which requires system coordination, how many people are needed to support the battery development?
YD: Battery development involves some of the core work, with two main groups: one for cell design and one for pack design. The cell design team isn’t large, with around 100–200 people, while the pack team is larger due to the engineering aspects, totaling about 400–500 people.
For batteries, we need to ensure low-temperature performance, high-temperature performance, state of charge (SOC) performance, and post-degradation performance, among other factors. We provide the battery team with precise requirements, and they design according to those needs, with many factors to consider. If we worked with external suppliers, it would be difficult to coordinate so well, and many things would remain confidential.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Han Yongchang for 36Kr.
