This article is based on publicly available information and intended solely for informational purposes, not constituting investment advice.

BYD Automobile Manufacturing Plant beneath the Qinling Mountains in Xi'an

In our report, "The Main Battle Lines for BYD, Xpeng, and Others in 2025," we emphasized that intelligent driving will be the strategic fulcrum for domestic automakers in 2025. This year marks the dawn of equal rights in intelligent driving.

Amidst this wave, BYD, with its substantial shipment volume, has demonstrated its commitment to fully embracing intelligent driving. Long-time followers believe this signifies the onset of BYD's second growth trajectory, yet the challenges it faces are unprecedented.

Intelligent driving diverges from electrification. Success in the era of electrification does not guarantee success in the era of intelligence. The prospects for BYD's "equal rights in intelligent driving" strategy cannot be predicted by simply extrapolating from past experiences.

01 Rooted in Manufacturing, BYD Deserves Respect

We often praise the craftsmanship spirit of Germany and Japan, but the Chinese manufacturing industry, epitomized by BYD, has propelled this spirit to new heights—extreme cost-effectiveness. In the fiercely competitive mid-to-low-end passenger car market, BYD maintains a profit of 10,000 yuan per vehicle.

Since its inception in 1995, BYD's 30-year journey has been centered on manufacturing technology. Despite transitioning from the fringes to the spotlight in the past three years, its success is attributed to its manufacturing DNA, advocating "technology is king, innovation is fundamental," exemplified by:

In the debate between ternary and lithium iron phosphate batteries, BYD, as a former battery manufacturer, took an unconventional approach and launched the Blade Battery with great success. Its subsidiary Fudi became the only successful case of a wholly-owned vehicle manufacturer being supported.

Regarding automotive architectures, when Volkswagen aimed to dominate the new energy market with the ID series, BYD's DM architecture emerged as the most successful.

Regarding vertical integration, Wang Chuanfu's early entrepreneurial experience and the supply challenges encountered during BYD's initial stages of car manufacturing laid the foundation for its vertical integration of the industrial chain. Therefore, competitors lack confidence when competing with BYD on price.

Regarding route disputes, whether it's pure electric or extended range, BYD has proven that hybrid is the optimal choice.

Regarding brand matrices, while Geely has launched Zeekr, Yinhe, Geometry, Lynk & Co, among others, BYD boasts its main Ocean and Dynasty series, along with mid-to-high-end breakthrough brands Tenzhi, Fangchengbao, and Yangwang, covering a price range from 50,000 to 1 million yuan.

Regarding brand exports, while Chery leads and new forces trail, BYD, the eldest brother, established a subsidiary in Europe as early as 1995.

This explains why, amidst the ups and downs of the lithium battery new energy industry in the past three years, BYD has emerged as the company with the best stock price performance. The core reason is that it has become the king in three price bands in the mid-to-low-end market due to its manufacturing advantages.

Figure: Domestic market share ranking by price band Source: HTI

The "hexagonal warrior" BYD has its biggest weakness in intelligent driving.

Previously, it was illogical for automakers with an average price of less than 150,000 yuan to invest heavily in intelligence. However, as BYD strives to break into the high-end market, intelligence becomes a necessity. Furthermore, 2025 is a year of rapid growth in intelligent driving. This industry trend aligns with BYD's high-end strategy, making its full embrace of intelligent driving inevitable.

Since BYD is China's top-selling new energy vehicle manufacturer, its transition to intelligent driving will inevitably have a significant impact on the industrial chain. It is understood that many supply chain enterprises have already received substantial orders from BYD.

02 Recruitment and Investment Cooperation: Heavy Bets on Intelligent Driving

BYD's gamble on intelligent driving is comprehensive.

1. Recruitment

The market has long believed that BYD focuses on the mid-to-low-end market, and intelligent driving involves high costs and technical risks, so its layout in this field has been relatively restrained. In reality, BYD did not hastily embark on intelligent driving to chase this year's hot topic. It has already been heavily recruiting in this field.

In 2022, BYD's intelligent driving team at the Planning Institute comprised only around 100 people. According to the annual report of the listed company, the total number of employees in 2022 was 570,000, with an average salary of 139,000 yuan, typical of a manufacturing company.

In less than two years, to fully embrace intelligent driving, BYD's intelligent driving team has expanded to 4,000 people. This number is comparable to the scale of the intelligent driving team of a leading brand.

At the beginning of 2024, BYD reintegrated its intelligent driving R&D resources and merged multiple R&D teams into the "New Automotive Technology Institute." Around the same time, at BYD's Dream Day, it unveiled a new strategy for the intelligent development of new energy vehicles—intelligent vehicles.

On December 16, 2024, BYD announced the establishment of the Advanced Technology R&D Center, integrating employees from the former Vehicle Simulation Technology Department, Big Data Platform Department, and other departments, including the AI Laboratory, AI Supercomputing Development Department, and Big Data Platform Department, focusing on the research and development of AI algorithms, AI infrastructure, and large model technologies and platforms.

There is no need to confirm the news from the industrial chain. Judging from these actions, BYD is fully prepared to compete in intelligent driving. An example is the "high-tech label" affixed to its new cars at the beginning of this year.

2. Learning from Market Leaders

Xiaomi's Lei Jun is not shy about "learning from others" in car manufacturing, ultimately finding a way out by paying tribute to every automaker, highlighting his interpersonal skills. For BYD, which has always advocated vertical integration, its accumulation in intelligent driving is relatively weak. Therefore, it has to directly learn from leaders in the industrial chain, though it has kept this internal.

However, this can be discerned through some of the company's investment and cooperation activities:

In February 2021, BYD completed a strategic investment in Horizon Robotics (9660.HK), a leading intelligent driving chip company. Starting from 2023, BYD's Song, Han, Tang, and other models in the Dynasty series, as well as the Haiyang series of Seals, all feature basic intelligent driving functions based on Journey 2.

In December 2021, BYD and Momenta, a domestic unicorn in autonomous driving solutions, officially announced the establishment of a joint venture, DiPai Zhixing, with BYD holding a majority stake and having the dominant power.

In December 2021, BYD invested in RoboSense, a lidar company, and reached a strategic cooperation agreement. In January 2022, BYD invested in VisionChip Microelectronics to further deploy lidar suitable for autonomous driving.

In 2022, BYD announced a cooperation with NVIDIA in the direction of intelligent driving. Starting from 2023, it began equipping the NVIDIA DRIVE Hyperion platform, and in 2024, it will equip the Drive Thor chip and use NVIDIA chips for cloud training.

In 2024, BYD signed an intelligent driving cooperation agreement with Huawei, primarily targeting Fangchengbao, with the first product, the Bao 8, equipped with Huawei's Kunlun intelligent driving ADS 3.0.

In terms of external promotion, BYD claims to walk on two legs of "open cooperation + in-depth self-research" and will gradually strengthen its full-stack self-research capabilities, including computing chips, algorithms, large models, etc.

For example, the company's high-level intelligent driving system that began mass production in 2023 is named "Sky God's Eye" (another intelligent driving system is named Dipilot). The company promotes its architecture system definition and full-stack self-research, and its large model technology has achieved a leap in perception algorithms from 2D to 3D, improving the system's perception, decision-making, and execution capabilities through end-to-end learning.

Although BYD has historically excelled in vertical integration, it has never ventured into high-performance computing chips. Its subsidiary BYD Semiconductor primarily focuses on power semiconductors and MCUs, and its CIS with certain computing power attributes is only average, let alone computing chips with 10 TOPS or even 100 TOPS of performance. As for algorithms, software, and models, these are not the company's traditional strengths.

3. Trial Runs Starting with High-End Brands

Since 2021, BYD's intelligent driving has been primarily tested on mid-to-high-end models. BYD's current brands equipped with high-level intelligent functions such as NOA are primarily Yangwang, Fangchengbao, Tenzhi series, as well as a few high-end models of Dynasty and Ocean:

Based on the Horizon single Journey 5 solution, it is mass-produced on models like the Han, priced at around 200,000 yuan.

Based on the NVIDIA single Orin X paired with dual lidar solution, it is mass-produced on the Tenzhi N7, priced at around 300,000 yuan.

Based on the NVIDIA dual Orin X paired with triple lidar solution, it is equipped on the Yangwang model, priced at 1.1 million yuan.

However, according to statistics, BYD's penetration rate of high-speed NOA/city NOA intelligent driving in 2024 was only 1.0%/0.5%, indicating it is still in its infancy and significantly lags behind the overall industry penetration rate.

4. Comprehensive Rollout

Historically, BYD's assisted driving systems primarily include Dipilot and "Sky God's Eye," capable of achieving L2 and L2+ levels of autonomous driving, respectively. The Dipilot assisted driving system was launched in 2020, partially utilizing self-developed chips and software algorithms while relying on Bosch, and has been used in the Han EV model, among others. The high-level intelligent driving system Sky God's Eye was launched in 2023.

To diminish the gap between different models, BYD will no longer disclose specific chip brands and models in the future. Instead, it will collectively refer to its intelligent driving computing platforms as DiPilotxxx, with the size of the suffix number implying computing power capabilities. We estimate that its low-computing-power platforms will use a mix of NVIDIA and Horizon chips, while mid-to-high-level computing power will still rely fully on NVIDIA.

According to an exclusive interview by DearAuto with Mr. Zhang Zhuo, General Manager of BYD's Ocean Network, BYD will comprehensively equip future models priced at 150,000 yuan and below with self-developed intelligent driving systems, promoting the popularization of intelligent driving technology. This goes a step further than Xiaopeng's P7+ in promoting equal rights in intelligent driving, particularly in the 100,000 yuan price range, which is the largest shipment interval for cars. This move will undoubtedly attract significant market attention.

We speculate that in models priced below 200,000 yuan, the intelligent driving mentioned may be limited by cost and can only be equipped with high-speed NOA at most, while mid-to-high-end models will be equipped with both city and high-speed NOA.

03 The Road to Intelligence: Steep and Arduous

Based on BYD's past achievements, most fans and investors are optimistic about its foray into intelligent driving. In the past, BYD has solved all survival-related problems, and intelligent driving, as an added bonus, seems a natural fit for BYD to tackle.

However, we believe that while BYD is a pivotal player in promoting equal rights in intelligent driving, we are conservative about the magnitude of its success. The market's current optimistic predictions for BYD's intelligent driving, based on its historical success, need to be tempered. This is because the current optimism linearly extrapolates BYD's past traditional advantages in manufacturing and branding to intelligent driving.

A closer look at all the new challenges BYD has faced throughout its history reveals that they have been "manufacturing-related" issues that the company could solve using its existing expertise, such as batteries and architectures mentioned earlier. However, intelligent driving involves computing power and algorithms. In this regard, BYD is no different from other fuel vehicle manufacturers and can be said to be starting from scratch, which explains the large-scale cooperation with the industrial chain mentioned earlier.

Let's still consider the model student Tesla's standard operations. From the time Musk entered the electric vehicle market, it has been a parallel path of intelligence and electrification, rather than making up for intelligence after electrification.

As early as 2013, Tesla embarked on its exploration of autonomous driving technology. In October 2014, it forged a partnership with Mobileye to unveil the first-generation autonomous driving hardware, HW1.0, which was initially integrated into the Model S. By 2016, Tesla had shifted gears, collaborating with NVIDIA to launch the HW2.0 system, powered by the NVIDIA Drive PX2 platform. Fast forward to 2019, when Tesla introduced the HW3.0 system, marking a transition to self-developed hardware with the launch of its first fully in-house FSD chip. Since then, Tesla has maintained full control over chips, algorithms, and computing power.

It's noteworthy that Tesla first achieved prominence in electrification, capitalizing on the success of the Model 3 in 2016 to solidify its position as the world's leading automaker.

However, Tesla's early forays into intelligent driving have consistently fallen short of Musk's ambitious market projections, underscoring the immense challenges inherent in this field. It wasn't until 2023 that Tesla unveiled FSD V12, becoming the first to offer an end-to-end autonomous driving system, featuring a closed-loop architecture from perception to decision-making, thereby hinting at commercial maturity.

BYD and Tesla are poised to become the undisputed leaders in new energy vehicles, with both aiming to achieve annual shipments of 5 million vehicles. While Tesla has prioritized intelligent driving from the outset and addressed chip self-research by recruiting legendary talent like Jim Keller, BYD's strategy focuses on volume expansion first, with intelligent driving catching up later. Even its pursuit of parity in intelligent driving feels somewhat hurried.

Opinions diverge on whether the correct starting point or timely intervention is more crucial. However, for intelligent driving, a solid foundation may pave the way for smoother progress. Speaking of which, Xpeng, perhaps overlooked by many, has demonstrated remarkable progress and could emerge as a surprise contender in the era of intelligent driving, deserving renewed attention.

Examining the intelligent driving trajectories of Tesla and Huawei reveals that such technological advancements do not happen overnight. For BYD, which lacks a strong software and computing chip presence, the path to catching up, despite its full commitment to intelligent driving, will be arduous. While market-ready standardized solutions offer a certain level of performance, the core technologies remain proprietary.

Furthermore, integrating intelligent driving systems adds 3,000 to 5,000 yuan to the cost of each vehicle. In the current automotive market, passing this cost on to consumers through price hikes is impractical, and profitability through service fees is even more elusive. Thus, promoting intelligent driving becomes a necessary expenditure, inevitably compressing profit margins per vehicle. For BYD, which aims to sell 5.5 million vehicles by 2025, this represents a significant hidden cost.