In a report released today, Matrix Partners China, a leading venture capital (VC) firm that has backed major Chinese tech firms at their early stages, says it is not surprised by the popularity of electric vehicles in China today, as it very much recognizes the underlying revolution that smart electric vehicles will bring.

According to the report, Matrix Partners China has been systematically investing in China since 2016 and has invested in more than 10 companies, three of which have gone public since then.

Two of the core OEMs invested in, Li Auto and Xpeng, "are like high points in a war, allowing us to take a more holistic view of the entire smart electric vehicle industry."

It's worth noting that the three best-known Chinese EV companies listed in the US are Nio, Li Auto, and Xpeng, and Matrix Partners China seems to have missed out on Nio, whose shares are up more than 900% this year.

Here is a translation of the original version of their report.

The entire electric vehicle sector has been on a rampage recently, with Tesla recording a mythical 1,000% rise in five years, most of which was achieved this year.

Li Auto and Xpeng, invested by Matrix Partners China, have risen 86% and 100% respectively since their IPO; BYD, a leader in new energy vehicles on A-shares, has also risen 280% from the beginning of the year.

Why are the shares of electric car companies soaring? Keep in mind that Tesla has only delivered a cumulative total of just over a million to this day, a fraction of the tens of millions delivered by traditional car giant Toyota, but already twice the market capitalization.

From a more global perspective, the total market capitalization of listed electric car makers is approaching $500 billion, and with the addition of battery producers, EVs and their chain companies are worth at least $700 billion (excluding raw material suppliers such as lithium mines), which is almost three-fifths of that of traditional carmakers.

But electric vehicle production capacity is still only a few hundred thousand units per year, while traditional automakers produce more than 80 million fuel-efficient vehicles per year.

We're not surprised by the explosive state of EVs today, as we very much recognize the underlying revolution that smart EVs are about to bring.

We have been systematically investing in layouts since 2016, and have invested in more than 10 companies since then, three of which have gone public.

The two core OEMs in which we have invested (Li Auto and Xpeng) are like vantage points in a war, allowing us to take a more holistic view of the entire smart electric vehicle industry.

In the upstream and downstream industry chain, we believe there are also abundant opportunities, such as automotive chips and automated driving. To this day, we are still systematically laying out the entire industry.

In this research, we first analyze what Tesla's $400 billion market cap means today. How did such a high valuation come about? And what is the underlying revolution led by the smart electric car? Which tracks will generate big new opportunities?

First, let's take a look at Tesla's valuation. The entire electric car sector has risen wildly, thanks to Tesla.

Tesla delivers roughly 1% of the cars of Toyota, the world's largest car company, but twice the market capitalization of Toyota, is this a bubble? Even if a bubble exists, why are investors so crazy about EVs?

Aswath Damodaran, a professor at New York University who is well versed in valuation, bought Tesla at $180 and gave it a valuation of $430, and he eventually sold it at $640. But Tesla continued to rise, going as high as $2,500.

The electric car frenzy caused a lot of difficulties for investment bank analysts, who kept looking for reasons to raise their target price, but eventually ran out of reasons and had to write "Tesla shares are getting higher and higher, and the rise itself makes it likely that it will continue to rise.

While Damodaran's target price is too low, his thoughts help us understand how the high valuation of the electric car company came about.

Four levers drive company valuations: growth, earnings, investment efficiency, and risk, with risk including the cost of capital and probability of bankruptcy, which Damodaran assumes a 7% cost of capital and a 10% probability of bankruptcy.

Growth depends primarily on revenue and needs to start with revenue projections ten years out. For the auto industry as a whole, the total market is $2.5 trillion, but growth is slow at 3.5%.

The biggest car companies in the world are Toyota and Volkswagen, both with revenues of $300 billion, followed by Mercedes-Benz at about $200 billion, and then BMW at about $100 billion. Tesla is currently only $25 billion.

So for now, let's just think of Tesla as a successful car company that will be able to deliver more EV sales than BMW, but less than Mercedes in 10 years. So let's assume that in 10 years' time it will have revenue between BMW and Mercedes, will be able to reach $125 billion, and will have a higher profit margin than the average car company set at 12%, which ultimately corresponds to $430 per share.

But today we wouldn't think of Tesla as just a car company, but a tech company in the car business. Tech giants like FAANG (Facebook, Amazon, Apple, Netflix, Google) actually have smaller revenues, even though their market capitalization is much higher than traditional car companies like Toyota or Volkswagen.

Tech companies have the advantage of high profit margins, with FAANG averaging 20% profit margins. So car companies have high revenue but low margins, and tech companies have low revenue but high margins.

If Tesla were a tech company in the car business, this story would be worth a lot more immediately, because the market would be predicting "car company revenue + tech company profit margin".

If you give Tesla the revenue level of BMW, the valuation would rise to $459 per share, and if you give Toyota the revenue level, it would rise further to $1,200 per share, based on the profit margins and investment efficiency of a typical tech company.

All things being equal, Tesla is a revolutionary company, capable of generating as much revenue as the world's largest automakers, while having the profitability and capital efficiency of a top-tier technology company. "It's $2,105/share.On November 10, 2020, Tesla's share price will be $2,106/share.

So when Tesla rises like crazy, you need to ask yourself: do you firmly believe that smart electric cars will disrupt fuel-efficient cars in the next decade? Can smart electric car companies like Tesla, Li Auto, and Xpeng, which emphasize their technology attributes, become companies with revenues on par with Toyota, profit margins on par with Google, and capital efficiencies that far exceed those of any manufacturing company in history, while also having a low cost of capital at the level of a car company, in the future?

If you believe that, then you should front-load your judgment and invest early into the game. The best window for venture capital to enter a new field is the eve of a major change that's about to occur in that industry, and we think 2016 is in the middle of such a window.

Here's our analysis of why we think smart electric vehicles will be such an opportunity. What deeper changes has it brought about?

In the first half of 2020, the traditional giant Volkswagen Group mobilized nearly 10,000 research and development staff to form a special team in the 74th hall of the Volkswagen headquarters in Wolfsburg to "close the development" day and night.

This frenzied battle is the traditional car company's death struggle in the face of the overwhelming attack of the new car manufacturing forces - they are building their own smart car software operating system vw.OS, but numerous bugs have appeared.

The difficulty in mobilizing such a large research and development team from the Volkswagen Group for this battle lies in the need to completely restructure the electrical and electronic architecture of the entire vehicle. In the world of traditional fuel cars, the car is just a hardware product, with little to do with the Internet or intelligence.

Although there are many electronic components in a conventional car, they are barely connected to each other.

A traditional car has more than 70 ECUs (electronic control units), but these ECUs only serve specific car parts, such as the specific ECU that controls the lights, or the one that controls the airbags, the one that controls the air conditioning, the one that controls the brakes, and so on.

This results in each component being independent, with numerous suppliers to build specific sensors, actuators, operating systems, and applications for specific components that don't yet intersect with each other.

So while a traditional car may look like it has a lot of chips, it doesn't have as much computing power as a mobile phone, nor can it interact with information at the vehicle level or perform OTAs (online upgrades).

For software developers, it is not possible to develop an application for a specific car, because each ECU has its own "operating system".

Bosch (BOSCH) has divided the vehicle electrical and electronic architecture into three major development stages: distributed, cross-domain concentrated, vehicle concentrated, most of the current OEM plant is in the second phase of the backward, namely, distributed modularization/integration stage.

Tesla as the representative of the new forces of intelligent electric car manufacturing, in the beginning, is expected to break the distributed design thinking, consider the overall intelligence.

For example, Tesla has redefined the concept of "domain" on the Model 3, which is equipped with an on-board computer and directly crosses into the on-board computer and district domain-oriented architecture.

This changes the traditional way of "simple function accumulation" by integrating subordinate functions in the zone domain with a few large units for modularization and centralization, ultimately reducing the number of system modules by orders of magnitude, and thus the concept of "domain" is born.

It is possible to reduce dozens or even hundreds of ECU control units to 3-5 "domains" of control systems, i.e. body system, entertainment system, chassis and safety system, powertrain, and advanced driving assistance system.

While the different domains are safely isolated from each other, they can also communicate and interoperate with each other according to their needs, and become shared resources and computing power, with a unified "operating software" on top.

The change in electrical and electronic architecture is a true revolution at the bottom. At present, only Tesla, Li Auto, Xpeng, and Nio can achieve OTA for the whole car.

For example, if the Tesla Model 3's braking distance is too long, the solution to the problem would be a massive recall or replacement of parts through a 4S shop.

However, Tesla upgraded its system through OTA, just like a smartphone upgrading its system, and solved the problem within a few days.

China's new car makers Li Auto and Xpeng have also achieved multiple OTA upgrades.

Another change is the falling cost of building cars. Traditional cars rely on a huge variety of wire harnesses to transmit electrical signals to the ECUs, which are too long and "too soft" to be assembled manually.

Wiring harnesses are often the third heaviest and most expensive component in conventional cars, weighing up to 50kg and having a total length of up to 5km.

With the new electrical and electronic architecture, the wiring harness length is significantly reduced (3km for the Tesla Model S and only 1.5km for the Model 3) due to fewer ECUs and the breaking down of barriers.

The biggest benefit of this is increased productivity and reduced labor costs, as harness assembly can even be done by robotic arms, and in special cases the wiring can even be printed on the body of the vehicle, further increasing automation.

With the new "domain" architecture and central computing architecture, the value of an investment in application software and services is significantly increased.

At this stage, 85 percent of the value in a car comes from individual components, but in the future, when the hardware is abstracted away, the software will become a core strength.

Another "downfall" is the change in the business ecology of hardware and software, in addition to the electrical and electronic architecture.

The historical deposits of traditional car companies in engines and transmissions are disintegrating in the face of electric vehicles.

Easier-to-produce electric motors can easily achieve 100km acceleration comparable to that of a sports car, and the overall number of components has been reduced by a third.

Reinventing the supply chain opens up a plethora of new opportunities, with prospects for everything from batteries to car chips to self-driving software.

Of course, car-level products on the research and development, organization, production and other capabilities have extremely high requirements, for example, the smart car on the use of car-level chips than the consumer-grade chips used in mobile phones are more stringent standards, because the car is a matter of life.

How to find suitable investment opportunities in such a complex industry chain? In our experience, a company needs to satisfy three core elements: whether it is an incremental market (an added product or demand), whether it has strong enough technical barriers, and whether it is a sufficiently large market.

In terms of the smartphone industry chain, the Apple chain has driven the entire golden decade of consumer electronics.

Since the introduction of the iPhone 4 in 2010, Apple's soaring sales have brought a comprehensive opportunity for chips, cameras, glass bodies, etc., and even true wireless headphones AirPods, in which a large number of upstream component growth companies have emerged.

There are roughly two types of development models, one is to extend vertically from a single high-quality track, such as the acoustic track Goertek, optical track Sunny Group, Ofilm.

The other is to recognize new product trends and expand horizontally in multiple circuits, from single product categories to multiple product categories.

Along with the huge smartphone industry chain, the return on investment of these upstream component companies is as good as investing in a successful IPO of a mobile internet app company at the time.

Today, the smart electric vehicle industry chain is equivalent to the pre-smartphone, and the rise in valuation and performance of the industrial chain companies comes mainly from the high expectations of OEM sales.

In the future, the subsequent industrial chain opportunities will also come from the second wave of opportunities birthed by new product trends, equivalent to the important role of the camera for the smartphone, perhaps this new opportunity will come from the high computing power of the car chip, or automatic driving, or smart cabin.

At this stage, the electric vehicle industry chain is largely driven by sales of electric vehicles. Smart EVs are a combination of mass consumer goods and technology products, and therefore fit into the "S-curve" that technological change usually undergoes.

The general industry experience is that once the product penetration rate exceeds 15%, the growth rate will break through the tipping point and reach a new level.

Tianfeng Securities has done a comparative study, in the SUV and smartphone market penetration reached 15%, the annual penetration rate increase is slow.

SUV annual penetration rate of only 1% - 2%, smartphone penetration rate of 1% - 3% per year, this stage for the initial stage of the industry, the current smart electric vehicles correspond to this stage.

When the market penetration rate exceeded 15%, SUV and smartphone development accelerated, ushering in a 5-6 year golden period of development. Until the SUV penetration rate of more than 40%, smartphones more than 70%, the penetration rate slowed down or even stalled.

Looking back to the electric car market, although there are many new car-making forces to enter the game, the current electric car market penetration rate is still less than 5%, still in the early initial stage, so electric vehicle enterprises will maintain a high valuation state.

As the concentration of traditional car brands is much lower than that of smartphones (only 10% in the head), the fragmented state of brands will be the norm.

However, for the new car-making forces, after all, half a foot in the manufacturing industry, can't think according to pure Internet products, take the rapid iteration + trial and error mode, cars, and mobile phones are different, security determines everything.

Unlike mobile phones, cars and mobile phones are all about safety. Most of the smart electric car companies will go through "capacity hell" in the early stage of mass production, but they still need to build their own factories to ensure quality, which is a problem that cannot be bypassed.

The era of "software-defined car" will come when there is a breakthrough in the sales of smart electric cars.

It leads to the most direct result is the smart car's computing power will be greatly increased, no longer like the previous traditional car computing power is inferior to smartphones, this time the new demand for chips will explode.

At present, the average value of the traditional car semiconductor in 475 US dollars, a new energy vehicle value of up to 600 US dollars, or more.

From the development path, automotive semiconductors will mainly experience three major expansion cycles: the first stage is mainly based on automotive electronics, and the use of automotive semiconductors accelerates under the widespread application of automotive electronics.

The second stage is mainly driven by automotive electrification, where power semiconductors are beginning to be used in large numbers in electric vehicles. And in the future, as the intelligent process continues to accelerate, smart cars, car networking, smart cabin, and other application scenarios will continue to enhance the semiconductor single-car usage, automotive semiconductor will enter the third expansion cycle.

In the field of automotive chips, we invested in February 2019 in Jingliu, which specializes in gigabit high-speed automotive Ethernet chips.

In April 2019, we invested in Semi Drive, which focuses on three major automotive chip applications: smart cabin, smart driver, and central gateway; and in the same period, we invested in SENASIC, which mainly covers TPMS chips, universal interface chips, and battery management chips. All three companies correspond to different directions.

Because of the complexity of the chips, it is not easy to make a product in the first place, so we mainly invest in teams with successful experience, who have gone through the process of making a chip from 0 to 1, and who are able to generate enough sales from this product.

All three companies meet these criteria, and all are extremely mature teams that are building a product that is a continuation of their experience. The Semi Drive, for example, has a founding team from NXP Semiconductors and involves a product that generates close to $1 billion in sales a year.

Investment relationships with core OEMs are also very helpful in making investment decisions up and down the industry chain.

For example, in the investment in Kingslow, an important disagreement was how many megabits of in-vehicle Ethernet is enough? Is it a 100 Gigabit or a Gigabit?

At first, most people thought that the data transmission capacity of today's vehicles is still relatively low, so it is impossible to need Gigabit Ethernet.

However, through communication with Li Auto, gigabit technology solutions may exceed market expectations, so the company finally invested in Jingliu, which directly targets gigabit Ethernet.

Once the company's basic technical capabilities are established, as in the smartphone industry chain, some technologies will become more versatile, and horizontal expansion will become possible.

For example, SENASIC started out as a TPMS tire pressure monitoring chip, but the technology can later be applied to EV battery pack pressure management as well, and SENASIC has even developed a general-purpose chip that can be applied to air conditioning control and window control.

The new automotive electrical architecture requires not only higher performance chips, but also chips with higher product definition.

In the original distributed ECU level, only one function can be controlled, but now in the "domain" architecture, an in-vehicle computer will manage many segmented functions, which also puts forward new requirements for the chip itself.

For example, Semi Drive's newly released 9-series automotive chips are "domain" control level large SOC chips, a single chip can replace multiple traditional ECU.

Of course, in the field of chips are not easy to start a business or investment. For investors, the chip industry, the chip industry too long technology introduction period leads to investment judgment must be early enough.

For entrepreneurs, in addition to the chip products can be built and mass production, due to rapid technological change, but also need to catch up with the tide of development of the larger industry, each wave of change, and timely launch of new products.

If these challenges can be overcome, the ultimate payoff will be amazing.

When Tesla's market capitalization reached twice that of Toyota, it meant that the capital markets recognized the long-range vision of electric vehicles being smart first, then shared.

In the future, the smart EV leader will not just be an EV manufacturer, but also an operator of an automated fleet, similar to how Apple not only sells phones but also runs an operating system and app store.

For example, in Q1 this year, Tesla CFO explained to investors the composition of Tesla's software revenue, with three business segments: connectivity, online system upgrades (OTA), and fully automated driving (FSD).

This will be the biggest opportunity of the next 10 years - the huge new market born from the combination of smart electric vehicles and travel.

If automated driving is to fully replace human drivers, it must have the capabilities of human eyes, brains, and hands and feet, which correspond to the perceptual, decision-making and executive layers, respectively.

Lidar, millimeter-wave radar, cameras, ultrasonic radar, GNSS/IMU, etc. in the perception layer are responsible for positioning and sensing weather, road conditions, obstacles and moving objects, etc., and then transmitting the information to the decision-making layer.

The decision-making layer, on the other hand, is mainly processors, car chips, etc., which process the data received by the perception layer through algorithms, analyze the environment and plan routes, and finally issue instructions to the executive layer.

The executive layer is responsible for the final acceleration and deceleration, steering, braking and other operations.

The collaboration of multiple sensors in these processes, as well as the transmission and computation of large amounts of data in a short period of time, are two major challenges, but each of these processes represents a huge new market space and investment opportunity.

The vast majority of today's production vehicles are only capable of "assisted driving" at the L2 level, which according to the SAE definition means "supporting multiple actions at the steering wheel and in acceleration and deceleration through the driving environment, with the remainder being performed by humans.

This is still far from the imaginary "can read or sleep in the car", but automated driving will eventually be phased in by scenarios.

Before 2016, automated driving startups will face two paths: one is to operate their own fleets of vehicles, which will become commercially viable by 2020; the other is to provide automated driving solutions to OEMs. Many of the companies that took the first path failed.

Yihang.Ai chose the second path, the gross profit level is not low, because the hardware solution includes sensors, computing modules are also included, the opportunity for startups is to be able to provide hardware + software solutions.

The endgame for this market could become tier-1 similar to traditional car companies (like Bosch), which is a platform-based business model. We invested in Yihang.Ai (Series A) in February 2017.

"The core of investing in the smart EV industry chain is to go for areas that can bring high added value through technology, and not be too conservative." Wang Huadong, a partner at Matrix Partners China, believes.

He reflects on missing the best time to invest in the LIDAR field because he was too concerned about the matter of whether the mechanical technology path could pass the vehicle regulations in the first place, and the mechanical because the vibration of the vehicle's movement would lead to a decrease in accuracy, and wanted to wait for new pure solid-state solutions.

But that wave of startups had already generated very substantial revenues by selling products to automated driving companies for driving tests, and could have invested those funds in the development of new solid-state products to gain a foothold in the market, which proved to be too conservative an initial judgment.

If you want to talk about the next decade, where are the areas that will change people's lives dramatically? Smart electric vehicles are certainly one of them, and the ongoing innovation and smartness of automotive electrical and electronic architectures will dramatically change the shape of the automotive and related industries.

What's most exciting is that Chinese companies are already at the forefront of this innovation, with the new carmakers largely crossing the 10,000-unit delivery threshold.

Fresh off a Covide-19 third quarter, China's EVs have rebounded strongly, with wholesale sales nearly doubling in September from a year earlier, marking the largest increase since April 2019.

With rapidly declining battery costs and increased EV volume production scale, the tipping point is expected to come in 2022-2023, when full lifecycle parity between EVs and fuel-efficient vehicles will be achieved, and the subsequent maintenance costs of EVs are low.

When EV penetration reaches 10%, it will trigger the classic "S-curve" of the technology revolution, which means a wave of exponentially higher growth.

Just as the smartphone drove the golden decade of the consumer electronics industry chain, the smart electric car will lead an even larger new automotive industry.