In early August, Chinese EV startup XPeng Motors announced the completion of the autonomous driving intelligent computing center, Fuyao, jointly built with Alibaba Cloud in Inner Mongolia. The cloud computing system offers up to 600 PFLOPS (petaFLOPS) of arithmetic power, i.e., 6 quadrillion (1015) floating point operations per second, which will make it the second most powerful supercomputer in the world. Drawing on the computing center, XPeng can speed up AI training for its fully autonomous driving model by 170 times and expect to launch autonomous cars by 2025. XPeng's self-driving system, known as XPILOT, has already introduced L3 functions and will aim for L4/L5 technology in the future. As a startup that did not mass produce new cars until 2018, how has XPeng gained such strong ambition and execution capabilities in the past four years to do something, the introduction of autonomous driving technology, that is challenging and requires a great deal of investment?
XPeng sold a total of 90,000 units in the first eight months of this year, ranking fourth among China's new energy vehicle brands and outdoing the other two in the NXL trifecta: NIO, XPeng, and Li Auto. Nonetheless, it is still operating at a loss due to the prodigious initial R&D costs. Under the circumstance, how come XPeng still decided to make a big push in autonomous driving and is about to invest more resources in R&D? Here is what I think…
- XPeng made its debut on the New York Stock Exchange in 2020 and on the Hong Kong Stock Exchange in 2021. Price pressure makes it imperative for the Company to stand out from the rest of the auto companies in Mainland China and gain a foothold with unique product competitiveness as soon as possible.
- The Company focused on the development of EV technology in the first five years of operation, and it should be able to reap the reward from the product lineup and sales result in the next five years while the pertinent investments are being amortized considerably. It is when the Company financial wise has plenty of room to ramp up R&D for self-driving technology and become the leading group in the race of self-driving technology that is still in a state of utter chaos.
- In 2018, XPeng was accused on different occasions of hiring former employees of Tesla and Apple to obtain information about their self-driving technologies. Whether or not it is true, the accusations are a testament to XPeng's commitment to securing the top technologies in the auto industry. After the highly controversial R&D stage in the beginning, XPeng continued to build up its technological momentum rapidly, attempting to use self-driving technology as a key advantage to get ahead of other Chinese EV startups.
- With its founder He Xiaopeng from Alibaba, XPeng has close ties with Alibaba (Alibaba owns 12% of XPeng's shares) and can use Alibaba Cloud's ample technical resources to break through the current bottleneck of insufficient computing power for self-driving systems. Meanwhile, Alibaba happens to be struggling to find a way to explore new opportunities in the auto technology industry (even though it announced last year that it would invest in the development of unmanned logistics trucks, and its new joint venture with SAIC, IM Motors, has started production), so now is the perfect time.
XPeng is currently using a lot of Nvidia's AI chips to advance its R&D efforts for self-driving technology. However, US President Biden just announced a ban on the export of certain AI chips from Nvidia and AMD to China. I'm not sure if the ban will affect the future development of XPeng's self-driving system, but in any case, it is clear that XPeng must conduct a comprehensive strategic review of its supply chain management of key components for self-driving vehicles. Otherwise, its superb cloud computing power as well as networking and edge computing technologies will not achieve much without an in-car AI chip that has sufficient computing power to match.