In October 2013, shortly after a parting of ways with Google, Tesla Autopilot brought on its first employee, Drew Gray, whose title at the time was Sr. Engineering Manager. After less than a year of development, Tesla released the first generation of Autopilot in October 2014. A year later, in October 2015, Tesla finally pushed the V7.0 software version to its vehicle owners, which was also the starting point for Autopilot.
While ordinary users were still immersed in the fresh experience brought by the rapid development of technology, Autopilot brought a good question to the investment community: What will future cars be like and what are the sustainable profit points?
In Tesla’s Q4 2020 earnings call, Musk said that the percentage of China’s FSD package orders, which is priced at 64,000 yuan, was only 1%-2%. At the same time, in NIO’s earnings call, Li Bin mentioned that the percentage of the two Pilot packages, priced at 15,000 yuan and 39,000 yuan respectively, was 50%.
From the data, it can be seen easily that if auxiliary driving systems are sold with a one-time purchase price, the price is the core factor affecting their installation rate. Interestingly, both companies mentioned in their earnings calls that they plan to take measures in 2021 to increase the installation rate of their own auxiliary driving systems.
Today we will talk about how car companies are “just right” about their autonomous driving software.
Should Software and Hardware Be Separated?
In August 2020, XPENG mentioned in its U.S. IPO prospectus that it would explore a software charging model in 2021 to make software revenue one of the company’s sources of revenue. In November, XPENG released its next-generation autonomous driving hardware, which adopts urban road HD Map and LiDAR. He XPeng also said that he wanted to “confuse the people in the west so that they can’t find the east.”
For serious and conservative capital markets, they need exciting stories to stimulate the dreams of investors overnight. Last year, many were looking for the next Tesla in the stock market. And at that time, XPENG, which focuses on intelligent driving, “bumped into the barrel of capital.” In the next few days, XPENG’s stock price also reached a historical high of $74.49.In the Q1 2021 conference call of XPeng Motors, as stated in the prospectus, the company recognized XPILOT software revenue for the first time in the overall vehicle revenue. As of the first quarter of this year, XPeng Motors has earned 80 million yuan from assisted driving software, with 50 million yuan in 2020 and 30 million yuan in the first quarter of 2021. Software revenue accounts for 2.5% of total revenue.
Of course, XPeng Motors also explained the composition of the 80 million yuan revenue, and let’s take a closer look.
Firstly, XPeng mentioned that since the delivery started in June 2020, up to the first quarter of 2021, XPeng P7 has delivered more than 23,000 vehicles. The cumulative paid rate of XPILOT 3.0 is more than 20%, and this figure reached 25% in March. However, the increase in NGP option rate is not significant. The NGP option rate in 2020 was 16%, and in 2021, it was 18.8%.
It should be noted that XPeng Motors is one of the few car companies that adopts “separate purchase of hardware and software.” The sales of the P7 model allow users to choose between the 2.5 hardware model and the 3.0 hardware model, with a price difference of 10,000 yuan.
It is worth noting that XPeng mentioned the cumulative paid rate of XPILOT 3.0 in March was 25%. With the data that XPeng P7 has delivered more than 23,000 vehicles, what XPeng actually means is that about 5,750 users selected the 3.0 hardware model, which means XPeng’s revenue in this area is approximately 57.5 million yuan.
For the other data, users who chose the NGP option increased to 18.8% this year. The potential meaning is that of the 5,750 users, approximately 1,092 users paid to choose the NGP option, and XPeng’s revenue in this area is approximately 21.84 million yuan.
The sum of the two data is almost the same as the claimed 80 million yuan revenue, but this also reveals a problem. That is, the number of XPeng P7s that can actually use NGP on the road is only about 1,092, and the penetration rate is 4.7% based on the data of cumulative delivery of 23,000 vehicles. According to XPeng’s official data, nearly 75% of P7 users cannot pay to upgrade NGP because they have chosen the 2.5 hardware model. This means that if this data continues to develop, XPeng’s software revenue growth can only be found in the remaining 20% of users.## Are You Paying the Bill?
Among the car companies that embed advanced driving hardware in advance and charge for software activation, Tesla and NIO are the most representative. And I want to share an “old story” first.
Apart from selling coffee and Starbucks-branded merchandise, the only other product sold by Starbucks is Evian mineral water, which is priced as high as ¥22. This bottle of water is usually displayed prominently, but how many “philanthropists” actually spend ¥22 on a bottle of Evian mineral water at Starbucks all year round?
When I saw the Starbucks staff mix a small cup of coffee into a cup of water to make my ¥28 Americano, I understood that the coffee itself only costs ¥6. Starbucks also told me that the biggest difference between an Americano and water was that it had that small cup of premium coffee.
Of course, I believe that Li Bin, who has publicly declared his love for the Starbucks brand culture on many occasions, must also understand this principle. To avoid the NP fully-equipped package looking too expensive, they launched a “NIO-branded mineral water” package, no, it’s the NP Selected Package. But unlike Starbucks, NIO’s “water” sells.
There are two key data points. First, according to the Q2 2020 NIO earnings conference call, Li Bin mentioned that the ratio of users choosing the NP fully-equipped package at a cost of over 30,000 yuan under normal sales conditions is approximately 25%. Second, according to the Q1 2021 earnings conference call, Li Bin mentioned that the combined ratio of the NP Selected and fully-equipped packages exceeds 50%, which means that the audience for the NP Selected Package is also about 25%.
Of course, NIO also mentioned more important information, that the average gross profit margin of the NP package is about 8,000 yuan. If we combine this with the delivery data of about 20,000 vehicles in the first quarter of NIO and make a reasonable inference, NIO’s “software gross profit” in the first quarter of this year is about CNY 800 million, which is also an important reason why NIO’s vehicle gross profit margin rose to 21.2% in the first quarter.
In comparison, Tesla’s FSD promotion is more diversified. During the Q4 2020 Tesla earnings call, Musk stated that the optional rate for FSD in China was only 1%-2%. In fact, before that, Tesla China team had hoped to change the low FSD optional rate in China through some marketing activities. For example, in September last year, Tesla’s official website briefly offered the EAP Enhanced Version of Automatic Assistance Driving for users to choose from, and provided a 3-year interest-free financial plan.Since the Tesla China team was mentioned in the financial report conference call, it can be foreseen that they will bear a more challenging software “revenue generating” task. Therefore, during the 5th day of the Chinese New Year and Labor Day, they respectively gave users 10 days and 45 days of free EAP experience time. Of course, one of the reasons why FSD has a low option rate in China besides its high price is because all models come standard with AP.
Because Tesla did not disclose any information about the FSD option rate in China in the Q1 2021 financial report conference call, we can only make an “irresponsible inference” from the phenomenon that Tesla China gave users a longer experience time again, which is that perhaps Tesla China achieved some effect from the first giveaway of EAP experience time, but did not reach expectations, so they decided to do it again with a “big” one.
From Tesla and NIO’s behavior, we can foresee that when car companies embed hardware in advance, even if users do not have assisted driving in the early stage, car companies can still promote software penetration through some marketing activities. Software pricing is critical to this.
According to data from the China Automotive Technology and Research Center, NIO’s average transaction price is 430,000 yuan, so the price of the NP selected package is about 3.48% of the total car price, while the full package is about 9% of the total car price. The later upgrade cost for the NP selected package is about 4.1% of the total car price, while the full package becomes 10.4%.
According to data from the China Automotive Technology and Research Center, Tesla’s average transaction price is 284,300 yuan, of which EAP accounts for 11% of the total car price, while FSD accounts for 22.5% of the total car price. This may be the root cause of the low option rate.
Looking at the relatively complex XPeng, based on Q1 financial report, the total revenue was 2.95 billion yuan and the total delivery was 13,340 units, which means XPeng’s average selling price is about 221,000 yuan. The NGP priced at 20,000 yuan actually accounts for about 9% of the total car price.
But if you add the cost of buying the 3.0 hardware model for 10,000 yuan in order to purchase the NGP, the cost of this assisted driving system actually reaches 13.57% of the total car price. And the cost of upgrading NGP in the later period is even higher. After more than 6 months, the cost of upgrading NGP will increase to 36,000 yuan, which will reach 16.2% of the total car price.
Taking all the data into account, I believe that if the pricing of assisted driving is in the form of a one-time payment and controlled within 10% of the car price, it is relatively reasonable.
## Use Your Imagination
In the eyes of primary market investors, there are only two types of companies: companies that make money and companies that are valuable. Of course, a company that makes money may not be valuable, and a company that is valuable may not necessarily make money.
For capital, valuable companies must first meet three criteria: the industry in which the company is located must have the ability to attract large cash flows; the product of the company must have great differentiation and subversion; the company’s product must be able to achieve “first” in a specific area, as long as there is imagination space.
Automatic driving is undoubtedly an industry full of imagination for capital. According to a CITIC Securities research report, the landing scenarios for high-level automatic driving are extensive. If estimated by potential scale, it is expected that the Chinese high-speed intercity logistics market will reach RMB 3.3 trillion, the automatic driving travel service market will be nearly RMB 1.7 trillion, the unmanned mining area driving market will be nearly RMB 670 billion, the unmanned terminal delivery market will reach RMB 170 billion, and the autonomous guest parking market will be about RMB 80 billion.
Chen Mo, CEO of Tusimple, said in an interview with “LatePost” that for the segmented market of self-driving trucks, the number of truck tractors in the United States is close to 3 million and the number of heavy-duty trucks in China exceeds 9 million. At present, their L4 self-driving virtual truck drivers have a net income of about $50,000 per year. This means that even if they occupy only 1% of the American market and operate 30,000 trucks, they will also have a revenue of 1.5 billion U.S. dollars per year.
Such a billing method is similar to the subscription-based model announced by Li Bin for NIO’s NAD autonomous driving system that will be adopted flexibly and charged on a monthly basis during the 2021 NIO Day.
Of course, this practice of “porting” the billing method of video and music software to the automotive industry can reduce the cost for users and seems to be beneficial for improving user software usage rate for car companies. This means that the fundamental factor that determines the software income of a car company has become the user size. From this perspective, Tesla should be the most suitable car company to implement subscription-based services.
According to Forbes, it is estimated that 60% of the value in future autonomous cars will depend on software, while the value of software is only about 10% today. This is also the reason why electric cars are “money magnets” at present.After all, the business society and public welfare charity are two contradictory opposites. No investor invests a lot of money just to get a “good person card”. How to commercialize and monetize is what they are more concerned about.
Final words
In fact, there is still a “minority” ideal car among all car companies that sell assisted driving software, and their approach is to directly equip all assisted driving software and hardware as standard.
Don’t get me wrong, I’m not saying it’s free, but when car companies directly make the hardware and software systems into standard equipment, if there is enough production and sales volume to drive the formation of economies of scale, it is possible to spread the hardware cost and software development cost thin enough. Whether subscription or standard equipment is better for users and investors in the future remains an unknown problem.
Finally, I want to say that the Japanese thinker Fukuzawa Yukichi mentioned in “An Outline of a Theory of Civilization” that “all the progress of civilization since ancient times started with an unscrupulous heresy.”
Fortunately, electric vehicles have already emerged from this period.
This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.