Author: Jaz
Tesla FSD welcomes a new round of price increase.
Starting from September 5, the price of Tesla’s fully self-driving system FSD in North America will increase to $15,000, a 25% increase from the original price of $12,000.
In order to recommend FSD to users, Musk specifically emphasized that “it only takes 2 minutes to upgrade to FSD through the Tesla App.”
Regarding this, some consumers expressed their anticipation to use FSD soon.
However, many consumers are not buying it, believing that if they have to pay extra fees for safety features, then Tesla’s business path will be difficult to pass.
While domestic automakers are still struggling with whether to charge for autonomous driving software or not, Musk is struggling with “how much is appropriate for the price hike.”
Cars can be cheaper, but FSD must be more expensive
Due to the impact of the epidemic and supply chain pressures, the overall vehicle prices of Tesla in North America have been on the rise.
Among them, the prices of the Model 3 and Model Y models have risen several times.
But Musk also said in the Q2 2022 earnings conference call that he hopes to lower car prices at some point in the future.
Nevertheless, cars can be cheaper, but FSD must be more expensive.
Musk once stated that FSD may be “worth nearly $100,000” and that “FSD’s price keeps increasing,” which has become a reality.
In 2015, Tesla first launched the FSD system, priced at $2,500.
The option price doubled to $5,000 in 2018.
In 2020, when the test version was launched, the price increased to $10,000, and then increased to $12,000 in January 2022. Now, after a new round of price hikes, it has reached $15,000.
In just 7 years, the price of Tesla FSD has increased 6-fold from the initial price.
Tesla’s price hike is not unexpected.
First and foremost is the cost brought about by software function upgrades.
The FSD beat 10.69.2 version update released on September 5 updated many functions.
According to publicly available Tesla version descriptions, this version update updated about 20 functions, including the addition of a new “Deep Lane Guidance” module, better simulation of delay in system and driving planning, etc.Overall, these upgrades and updates mainly focus on improving safety and comfort. For example, improving the correlation between traffic lights, signs, and taxiways, enhancing the smoothness of protected right turns, and reducing errors in deceleration and yield positions without relevant objects present.
In simple terms, with the upgrade of some functions, Tesla has reduced unnecessary deceleration, making car owners feel more comfortable and safe.
This also means that every line of code is money and every upgrade is a cost investment.
According to Tesla’s previous financial reports, Tesla’s R&D spending was as high as USD 2.591 billion last year, a year-on-year increase of 74%.
Such a huge R&D investment naturally requires FSD to provide cash flow for Tesla, coordinate the proportion of R&D and revenue, and improve gross margin.
In addition, US inflation is also a major reason for the frequent price increases of Tesla FSD.
Historical inflation in the United States has spread for nearly a year, and the most direct impact is to raise the prices of upstream raw materials such as iron and aluminum for Tesla.
Although the main result was an increase in Tesla’s hardware prices, using the appreciation space of FSD to relieve inflation pressure for Tesla was also reasonable while the price of Tesla’s entire vehicle increased.
Even if FSD has risen four times in a row, the probability of Tesla owners subscribing to FSD has not decreased but increased.
According to the public data of Tesla’s 2022 Q2 financial report, the number of users who have subscribed to FSD in North America has exceeded 100,000 as of now.
If calculated based on the average selling price of FSD in North America after September 2021, which is USD 11,000, the revenue of FSD in North America has contributed at least USD 1.1 billion to Tesla.
Moreover, according to Tesla’s statistics, the number of FSD users has gradually increased since September 2021 and has been in a high-speed growth trend for nearly half a year.
Although FSD is currently contributing higher revenue to Tesla in North America, the purchase rate of FSD in China and Europe is still very low.
Especially in China, due to domestic policy reasons, some functions of FSD are still unavailable and are only in the anticipatory stage.
According to data, the identification of traffic lights and parking signs and the automatic assisted driving function in city streets in Tesla FSD cannot be used.
Due to the high price and some functions of FSD that cannot be used, the subscription rate of FSD in China is only 1% to 2%.
Therefore, although the subscription rate of Tesla FSD is on the rise, regional policy influences and consumer education are still two important hurdles for the future appreciation of Tesla’s autonomous driving software worldwide.
Pay or Standard?
The average price of autonomous driving software for new energy vehicle companies is over CNY 30,000.Tesla, which has technical confidence in “fully autonomous driving”, has chosen to charge for its autonomous driving software to the end. However, due to factors such as policy and regulations, technical level, and corporate culture, there is more room for exploration for domestic new forces in terms of charging for autonomous driving software.
At present, the new forces are at a crossroads between charging and standard configuration.
These new forces can generally be divided into two factions:
- One faction is the software charging faction represented by NIO, Zhi Ji, and Ji Ke;
- The other faction is the standard configuration faction represented by XPeng, Li Xiang, WM Motor, and Salon Jijia.
There are big differences in views between the two sides on autonomous driving software, with the former believing that opening up subscription charging for autonomous driving services is a more sustainable model, while the latter believes that the level of charging has not yet been reached.
Among them, Wei Ma Automobile founder Shen Hui said frankly on Weibo that “let’s not talk about money for autonomous driving for the time being”, and listed reasons such as the level of autonomous driving technology is not up to the level of charging, and users’ willingness to purchase is not strong.
The main force of the standard configuration faction, Li Xiang, CEO of Li Xiang Automobile, also stated that “We will adhere to the standard configuration now and in the future.”
While the standard configuration faction firmly follows the path of standard configuration, the high-end intelligent driving of the charging faction is not cheap.
According to Autohome, the current average price of high-end intelligent driving systems in the charging faction has exceeded 32,000 yuan, with NIO Pilot’s full configuration package priced at 45,000 yuan, Zhi Ji’s IM AD Intelligent Driving package priced at 36,800 yuan, and Ji Ke’s ZAD full configuration package priced at 35,000 yuan.
Although the purchase rate of high-end intelligent driving systems has shown an overall upward trend in recent years, the price of over 30,000 yuan, which accounts for 7%-10% of the total car price, still makes ordinary car owners very hesitant.
“I don’t buy it because I don’t think it’s a complete smart car, but if I buy it, I feel like I’m wasting it. Even if I use the high-end autonomous driving function, I still have to focus on it.” Some car owners complained.
Even NIO, a group of car owners with extremely high stickiness, has to carefully divide the autonomous driving software package into three different levels – a full package for 45,000 yuan, a selected upgrade to full package service for 30,000 yuan, and a selected package for 18,000 yuan – to cater to consumers’ adaptation period for autonomous driving software fees.
In fact, the reason why domestic high-end autonomous driving software charging is rocky still needs to be discussed from the two points of technology and user perception.
Technology is the underlying logic that determines whether autonomous driving software can be charged.
At present, these high-end driving software packages are basically divided into three parts: hardware improvement, software development, and OTA upgrades.”Paid charging” usually not only enhances software functions, introduces features such as high-speed NOA and high-precision automatic parking, but sometimes also comes with more advanced or larger quantities of intelligent hardware than assisted driving, including but not limited to sensors, surround-view cameras, and new-generation ultrasonic sensors.
Hardware upgrades can be seen by consumers in real money, but has the automatic driving software upgrade really met the standard?
At present, the levels of new and established car companies are uneven.
Take the NIO Pilot full options package as an example, in addition to the content of the selected package itself, such as adaptive cruise control and road keep assist, it also adds 9 functions including leading assist, high-speed automatic driving assist, traffic jam assist and road sign recognition.
There is a qualitative difference between advanced intelligent driving and L2 assisted driving in terms of safety and comfort. Strong technical barriers can be the reason for car owners to pay for it.
However, BMW and Mercedes-Benz have included some basic software services into paid pricing, which is indeed going “off course.”
For example, BMW has already included heated seats and steering wheel in Korea and the UK as paid value-added functions. Similarly, Mercedes-Benz has paid subscriptions for unlocking rear-wheel steering and remotely starting the vehicle. This also puts the software pricing model at the forefront of public opinion.
Consumers may be willing to pay for fully automatic driving functions, but if the trend of charging for basic functions continues to spread, it will cast a shadow over the pricing model of automatic driving software.
Chinese car companies have not been able to achieve what Tesla couldn’t.
Insufficient consumer awareness and weak willingness to pay for software consumption are the second obstacle to paying for automatic driving software.
According to a report on future autonomous driving by Deloitte, the proportion of Chinese consumers unwilling to pay for automatic driving functions above USD 500 exceeds 52%, while the United States is 63%.
Overall, it is extremely difficult for car companies to charge for automatic driving functions beyond USD 500.
Although XPENG has mentioned in multiple earnings conference calls that more than 20% of XPilot 3.0 users have paid, creating more than 10 million yuan in revenue for the company.
But in May of this year, XPENG chose to “cut benefits and give away software”, turning XPilot which was worth more than 20,000 yuan into a standard configuration.
XPENG became the first car company to cancel the fee of a high-level assisted driving system after charging for it.
This move is intended to eliminate the negative impact of XPENG’s cancellation of free charging services. However, the official statement that “to achieve the widespread use of higher-level intelligent assisted driving in more users” just proves that 20% adoption rate is still far from enough, so XPENG needs to obtain more data to accelerate the landing rate of the city’s NGP function to promote the opening rate of subsequent automatic driving software.
Automatic driving software pricing: Knowing there is a tiger in the mountain, we still must go to the mountain
If you compare the three companies of Tesla, XPENG, and NIO, you will find that paying for automatic driving software is inevitable and standard configuration is just a special case.# Xpeng’s uniqueness lies in its ability to read the times, exchange users’ charging rights for replicable smart driving software, while Ideals uniqueness lies in that its software is currently out of sync with its automotive products, and it has yet to establish a user payment base.
Due to Ideals restriction by Mobileye, when Xpeng unveiled its NGP and Ideals NOP in 2021, Ideals remained inactive.
It was only at the end of 2021, in Ideals 2021 model with a dual 3-chip, that NOA assisted driving was included as a standard feature. Following this, Ideals released its AD Max intelligent driving system, which became standard in the L9 model.
Ideals, which released the L9 model three years after the success of Ideals ONE, naturally did not take any rash actions during the most critical product transition period.
Especially for Ideals, which previously did not have a software consumer base, the current stage of software charging is not that they do not want to charge, but that they cannot charge.
However, everything fate grants has already been priced in secretly.
According to media reports, new energy vehicle autonomous driving department employees have indicated that research and development costs need to be recovered through sales, and it is not ruled out that Ideals ONE has included the standard L2 level ADAS system in the vehicle sale price.
So, how can software charging be avoided from having a negative impact?
To solve this problem, automakers are finding different ways to charge.
Currently, Tesla, Xpeng and other automakers in the charging camp use a one-time purchase + subscription charging method. Generally, a one-time purchase is more favourable than paying for a subscription. But the subscription method has a sense of ‘try-before-you-buy’ and gives consumers more choice.
At the same time, automakers are charging based on mileage and duration, making it more flexible and comfortable.
To make consumers more smoothly accept payment, automakers have been “cautiously” maintaining consumer emotions, even extending the trial period to one year.
For example, the IM Zhi Ji 3.68 yuan intelligent driving package can be used by consumers for free in the first two years, and the owner must choose to spend 1.8 yuan to purchase data rights or buyout.
In the end, autonomous driving charging is not a matter of knowing that there are tigers on the mountain and leaning towards the mountain, but “must go towards the tiger”.
From the bottom-level chips to software development, software is becoming increasingly valuable.
There are two main reasons for this. The first is the shift towards the “smart” war in the automotive industry.
As the trend towards electrification lowers the barrier to entry for car manufacturers, the ultimate goal for smart automakers is the word “intelligent”.
The barrier to entry for making four wheels is not high, but the upper limit for smart vehicle makers carrying “computers” is being infinitely raised.
Currently, automotive hardware is also in the midst of consolidation, but even if it is hardware consolidation, it is based on “automatic driving software upgrades” for hardware layout.
Previously, the competitiveness of electric vehicles could have been seen mainly in who used better batteries, but now the competition is focused on the hardware configuration related to autonomous driving.# Recent NVIDIA A100 Restrictions Will Bring Challenges to Cloud-Based Training of Autonomous Driving, Says XPeng Motors CEO
According to He XPeng, CEO of XPeng Motors, as XPeng has already purchased several years of chip demand ahead of time, there will be a next generation of smart driving surpassing “fully autonomous driving” with the implementation of advanced intelligent driving assistance. He believes that the NVIDIA A100 restrictions will pose a challenge to cloud-based training of autonomous driving.
The second reason is that the cost of intelligent car hardware is decreasing, while software costs will gradually rise in the future. According to the 2020 Digital Report released by PwC, the hardware development cost for each car model is expected to decrease by 11% in 2030, while software costs will increase from 34% to 42%, with software verification and testing costs rising from 28% to 31%.
In other words, like mobile phones, cars can also rely on software to create sustainable profits. The logic is similar. Apple not only sells hardware, but also subscription fees for software services such as Final cut, Apple Music, and iCloud. Even Apple charges a 30 percent commission on application developers who earn over 1 million dollars annually through the App Store, becoming one of its key sources of revenue.
Based on this logic, we may expect that in the future, automakers will be able to have stronger vitality and business flexibility like Apple through software ecology and payment.
At present, there are still many small and large issues in automatic driving software charging, including but not limited to the mismatch between automatic driving software functions and prices, the vagueness between passenger car assisted driving and advanced autonomous driving functions, the need for time for the landing of urban intelligent driving functions, and low practicability, etc.
Especially, the current level of automatic driving software technology is still in the midst of competition. Whether it is the various names of automatic driving software or the difficult-to-standardize charging models and standards, it indirectly proves that the charging of automatic driving software still need time, mass production, market verification, and policy permission to carry out “the survival of the fittest”.
However, as He XPeng said, “At present, the effect of packaging hardware and software charging is better, but after the launch of more advanced autonomous driving, we will consider charging by time, mileage, software package + scenario service and other ways.”
With the background of “software-defined car” becoming an industry consensus, the trend towards standardized “automatic driving software charging” will be inevitable.
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.