With Nio and and Li Auto having joined the wave of US IPOs, Xpeng also raised $1.5 billion in an initial public offering on August 27th, becoming the third Chinese  electric-car maker in the US IPOs. 

All along, Xpeng is a good case in China's real Internet car manufacturing. He Xiaopeng, the founder and chairman of Xpeng, is a man with rich experience in the Internet field. 

Such a unique background also shaped the growth path of Xpeng. From the early Beta prototype to the engineering production car (version 1.0), to the G3 model, Xpeng has become the only one company among the new car makers that develops two platforms at the same time, and based on them, has developed two models including of SUV and sedan. 

Xpeng must be the new car-making force with the strongest iteration, or the fastest growth. At the beginning, it seems that there are always some "gaps" in Xpeng's feeling towards its products, which are not so deadly, but along the way, it becomes more and more precise and stronge".

If we were to use a simple label to quickly differentiate these car makers, we would say Nio would be the brand, Li Auto would be the product, and Xpeng's label would be "smart". This "smart" is not only reflected in the IPO prospectus, which reveals a 90% investment in research and development, as well as the building of China's full stack of self-driving capabilities, but also in He Xiaopeng's certainty on the future technological iteration of new cars.

According to the IPO prospectus, by the end of July 2020, the number of Xpeng's delivery had reached 20,707. The growing sales volume and the data behind it have made He Xiaopeng more and more confident, and his view of the car-making world more and more distinctive. 

Even though intelligence seems to be a "basic capability" of the new car-making force. But for Xpeng, which has been breaking new ground, even all-in, on the central axis of intelligence and technology, it is taking a line of attack of its own. This could be the first and sharpest way to leverage the industry's transformation with its intelligence.

What exactly is All-in's "smart"?

"Xpeng's label is smart, and smart is its core strength," He Xiaopeng said at the launch of this year's Xiaopeng P7. But what exactly exactly is All-in's "smart"?

In the early days of new car manufacturing, many would liken the new "smart car" to a "computer on wheels". But five years later, we can take a fresh look at the car companies. Who really has the ability to "build computers"?

The "smart" car, in fact, is not only many convenient and clever functions stacked, which is actually a kind of "ability" for the car company to expand.

The information disclosed in the prospectus shows that in 2018, 2019 and the six months ending June 30, 2020, Xpeng's R&D expenses were RMB 1.05 billion, RMB 2.07 billion and RMB 630 million, respectively. And the R&D expenses accounted for 89.2% and 62.9% of the company's total revenue in 2019 and the six months ended 30 June 2020, respectively.

According to the prospectus information, as of June 30, 2020, Xpeng had 3,676 employees in China and the United States. Approximately 43% (approximately 1,600) of the employees are in research and development roles, of which 66%, 17% and 17% focuses on automotive design and engineering, automated driving and intelligent operating systems, respectively.

"Currently, Xpeng has four major R&D centers, with about 1,000 engineers in the entire engineering and R&D including the automotive center, about 180 in the powertrain center, about 300 in the autopilot center, and about 300 in the internet center." Xu Jihan, vice president of Xpeng, once told GeekPark. He is mainly responsible for the formulation and management of Xpeng's overall technology development route, as well as the technical development of the E platform products, and is the overall head of the E platform products.

The term "software-defined car" seems to be the most frequently used in the industry.

However, in the context of "software-defined", the platform is the foundation to carry it. Xpeng is the only company among the new car makers that has developed two platforms at the same time, and based on them, it has developed both SUV and sedan models.

Xpeng G3 is built on the "first generation platform", codenamed "D platform", which covers a wheelbase from 2.6 meters to 2.8 meters. The second-generation platform, codenamed "E platform," covers a wheelbase from 2.8 to 3.1 meters. "Basically, we will be able to build all the models in this wheelbase range on these two platforms. This is very advantageous for our company's next step and for developing more models," Xu Jihan said.

Traditional car platforms, such as chassis, powertrain, styling, etc., are still iterating on a seven- or five-year cycle. But nowadays, the chips, sensors, electrical and electronic related hardware in cars need to be iterated at a faster pace.

Xpeng is changing the iterative changes of traditional car platform hardware. "This second-generation smart car platform (E platform) of Xpeng P7 is actually still evolving", Xu Jihan said.

The SEPA platform established by Xpeng for the characteristics of smart electric vehicles has these features, such as modularity, light weight, large variable bandwidth and high space utilization. From automatic driving, perception, computing to planning and control are all independently developed, it has formed a brand new electrical and electronic architecture and developed towards domain integration.

A dual engine drives the growth

"The intelligence of cars is following the original Moore's Law. The larger the scale and the more mature the technology, the faster the price of smart cars will fall close to that of current gasoline cars. But as the price of hardware cost decreases, the development cost of software will surely continue to increase, and the change it brings will be felt very strongly. So when hardware costs fall and software development costs rise, the new differences will emerge, and that's the time to change the automotive landscape." He Xiaopeng, chairman of Xpeng, said.

In the competition, Xpeng seems to have begun to "occupy the territory". The prerequisite for making full use of software intelligence must be the competition between "sensors" and "electrification". "I've always believed that cars will change from a single engine, such as a gasoline engine or an electric engine, to a dual engine. These two engines are not separate, but intertwined," said He Xiaopeng.

Among the smart capabilities of new car makers, it has to mention the "smart driving system". This is not only one of the essential skills for new car-making forces, but also a highly competitive field.

Currently, Xpeng's self-developed XPILOT 2.5 system has been installed in the G3 and P7. What's more, the P7 is equipped with Xpilot, Xpeng's self-developed autopilot software, and NVIDIA's Xavier autopilot chip, making it the world's first car to be equipped with NVIDIA's Xavier autopilot chip with 30 Tops. The P7 also offers more than 10 times the power of the current mainstream Mobileye Q4 chip.

The "top computing power" also brings a variety of high-level autopilot features to the P7 such as high-speed autonomous driving and parking memory parking, but these features will be updated later with full OTA capabilities.

Xpeng's smart route even earlier turned to self-development, automatic driving functions began to rapidly iterate and upgrade year by year. In 2018, it has the ability to embed Bosch's system into its products; in 2019, it developed Xpilot 2.5, and now Xpeng P7 has been equipped with Xpilot 3.0 system, which can achieve high-speed navigation, fully automatic parking, urban congestion Road Segment Assist; Xpeng plans to implement Xpilot 3.5 by 2021 and Xpilot 4.0 by 2023, with automatic assisted driving for urban road conditions.

Xpeng has been insisting on achieving end-to-end closed-loop capabilities with its own research, so with the P7's automated driving assistance system, it has achieved rapid iteration of functions by learning users' driving habits and closing the loop through data.

As of June 30, 2020, Xpeng has completed 25.1 million kilometers of road test information collection in China, and the lane keeping function has been used for 11.1 million kilometers of driving. As its products continue to iterate, Xpeng will also have self-driving technology that is most familiar with the local roads in China.

"The electric vehicle industry is undergoing rapid technological change, and the company needs to invest heavily in research and development to lead technological progress while remaining competitive in the market," the company said. In its prospectus, the company also said that it will continue to invest in improving the technology.

This is an article from WeChat official accounts GeekPark (ID: geekpark), written by Wang Xunkui, translated by Chris Yuan.