Source: ADS Think Tank
Autonomous driving is essentially a collision and integration of emerging technologies and traditional transportation industries, and its commercialization should be considered from the perspective of urban design and operation. After decades of technology and capital rushing, the commercialization of the global autonomous driving industry is still a problem.
Robotaxi (self-driving taxi), which borrows from the shared ride-hailing model, is the most common business path, but since 2021, Google Waymo, the representative company of the Robotaxi model, has suffered a series of declining valuations, CEO departures and test cars.
GM's Cruise has instead released the first driverless minibus without a steering wheel, and there are no large-scale landing plans so far. Many people firmly believe that the future of autonomous driving is inevitable, but the current commercialization path is worth discussing.
The difficulty of commercializing autonomous driving lies in the system.
As a revolutionary and subversive new technology, autonomous driving has the huge energy to affect a country's automobile, transportation, energy, manufacturing and even comprehensive national strength, and is a system engineering related to science and technology, consumption, industrial chain, policies and regulations, etc. All elements interact and are indispensable.
In terms of technology, according to the iterative logic of data-driven algorithms under the framework of deep learning, enterprises need to invest huge amounts of R&D funds all year round and collect data to optimize their autonomous driving systems in order to truly enter the stage of large-scale commercialization.
According to the Rand Think Tank in the United States, this mileage data is about 17 billion kilometers. Even if a team has 100 road test vehicles and 24-hour non-stop testing, the time required is based on a hundred years. This also explains why Waymo's 20 million miles of road testing, which it has accumulated over decades, still cannot be implemented on a large scale.
In terms of commercialization, the intensity of market demand essentially depends on the actual ROI of autonomous driving to replace labor costs.
At present, the cost of retrofitting an L4 level autonomous bicycle is often millions, compared with the negligible cost of drivers that can be saved. Many industry experts believe that robotaxi may open the market faster only in places where taxi drivers have higher incomes such as Japan and Northern Europe.
Finally, the popularity of any new technology is inseparable from regulatory approval. The policy challenges encountered by autonomous driving are even greater than when cars were first born 135 years ago, and they are not only related to life safety, but also closely related to today's data security and national security.
Safety can not see the end in the short term, policy permits can not be released, and the ROI of transforming an autonomous vehicle cannot be counted. This seems like a vicious circle that is almost impossible to break.
The breaking point occurred in Hengyang, Hunan, a central transportation hub city in China. This is currently the world's largest city-level autonomous driving commercial project, with a total investment of 500 million yuan and a total project mileage of up to 200 kilometers.
With the world's leading "Vehicle-Road-Cloud Integration" system, Mushroom Car-Union signed a strategic cooperation agreement with the Hengyang Municipal Government of Hunan Province in March this year to intelligently upgrade urban roads and provide a full range of autonomous public service vehicles covering the entire city.
Recently, Zhu Lei, founder and CEO of Mushroom Car Union, and Zhou Wei, head of Genesis Partner Capital, made a guest appearance on the technology podcast "Silicon Valley 101" to talk about the problems and ways out facing the commercialization of autonomous driving in depth.
Zhu Lei believes that the 2.0 era of autonomous driving has arrived, but how to commercialize is a problem that every technology entrepreneur should not escape. Autonomous driving is essentially a collision and integration of emerging technologies and traditional transportation industries, and its commercialization should be considered from the perspective of urban design and operation.
Here are some of the interviews:
Silicon Valley 101: How did you get involved or focus on autonomous driving?
Zhu Lei: I am a person from a technical background, the earliest after graduation in Baidu for about 7 years, the search engine related system architecture has been done, after leaving Baidu to start a business; in Didi for a few years, in terms of technology and business operations, these two companies have given me a lot of experience.
In the whole industry development process, one is technology-driven, one is commercial-driven, and the field of automatic driving is also very interesting, and it is very related to technology and business, so it is still very natural to enter automatic driving.
We first started with the road-to-car collaboration. We feel that whether it is from the Internet of Vehicles or autonomous driving, it is really necessary to land, or to be closely integrated with traffic.
Zhou Wei: I am an entrepreneur who is at least a generation before Zhu Lei, who has been a business for 11 years and a venture capitalist for 14 years.
I was involved in China's fintech entrepreneurship around the beginning of 1993 when I was in college, and together with the team designed China's first generation of independent intellectual property rights of financial payment pos machines, pos machines for stolen credit cards, and the development of a series of related products.
I joined KPCB (Kleiner Perkins), one of Silicon Valley's most legendary venture capital funds, in early 2007 and was a managing partner for ten years. KPCB led to an overall wave of green technology investment in mid-2000, involving many investments in energy, batteries and vehicles.
After I left KPCB with my team in 2017, I created Genesis Partner Capital CCV, and we've always been focused on this area.
Overall, we believe that future autonomous driving is inevitable. The last wave of green technology investments went through so many years, and many of them were unsuccessful because the business path could not be achieved at that time. So we are also very concerned about the topic we are talking about today: what kind of path is most appropriate in the commercial progress of autonomous driving.
"Silicon Valley 101": Mr. Zhu, in this industry, how will you position yourself?
Zhu Lei: Our positioning is to do along two core elements: one is the technical service provider of the full-stack technology of automatic driving, including the core algorithm and hardware of automatic driving; the other is the operation service provider of automatic driving.
Operations include autonomous buses, sweepers, patrol cars, robotaxi, embedding technology into actual operational scenarios, such as autonomous bus services may require cooperation with bus groups, and bus groups purchase or lease autonomous vehicles, we provide operation and maintenance solutions.
"Silicon Valley 101": Mr. Zhou, when you invest in energy projects, many of you are too far away from the commercialization scene. It is also difficult to commercialize fully autonomous driving in the early stage, will it be easier to commercialize like Mushroom Car Alliance to do autonomous driving technology services and operation services at the same time?
Zhou Wei: This is a good topic. In Silicon Valley, whether it is the Internet or AI, it takes the technology to reach a high level of maturity before it can put the product on the market.
But in China, the way startups have been using in the past is that when the technology is not mature, use a flexible or mixed-race way to provide services to the market in advance, and improve the product in a rapid iterative way, which is very different from the entrepreneurial cultures of the two countries.
In the early days, this approach must have been dominated by Silicon Valley, but Chinese companies will have a head start in commercialization ahead of time. In the field of autonomous driving in China, high-speed autonomous driving, and L5 are not easy to achieve on a large scale in a short period of time, but in some specific areas, such as semi-closed environments, fixed routes, etc., fully automatic driving has opportunities.
At a conference in Silicon Valley three or four years ago, I mentioned that the world's first city-wide fully autonomous driving should appear in China. The combination of administrative strength and institutional advantages and the ability to rapidly iterate provide a world-leading practice environment for China's autonomous driving.
"Silicon Valley 101": Now in the American autonomous driving circle, everyone will also feel that the landing of China's automatic driving may be faster than that of the United States, because China's infrastructure construction is faster, in this sense, automatic driving and the planning of the entire city, government support has a lot of relevance.
01 The world's largest autonomous driving project landed in Hengyang, Hunan
"Silicon Valley 101": In 2021, Mushroom Car Union signed a cooperation project of about 500 million yuan with the Hengyang Municipal Government in Hunan Province, which is a very large investment amount. Can you tell us a little bit about this project?
Zhu Lei: Indeed, China's autonomous driving is very advantageous in large-scale city-level landing. Not only the Hengyang project, including the earliest projects in Shunyi and Suzhou, are typical cases of large-scale application of autonomous driving at the city level. In the operation of automatic driving, taking the city as an independent unit is a very reasonable landing method.
First of all, the regional management of the entire city is unified; second, the infrastructure is consistent; third, the route environment and traffic conditions are relatively unified when the city lands.
The hengyang project is very distinctive, we provide a complete set of "bicycle intelligence + vehicle-road collaboration" program. Vehicle-road collaboration is a very good supplement to bicycle intelligence. For the safety issues that everyone is most concerned about, our program has been greatly upgraded to the previous bicycle intelligent solution.
By analogy, when we were young, we often encountered power outages, but now there are very few, not because a single circuit is more stable, but because the current circuit system is multiple sets of backups to ensure system reliability.
In terms of bicycle intelligence and vehicle-road collaboration, it is equipped with multiple redundant systems, which will greatly improve the safety of operation and landing of automatic driving. In theory, it is possible to achieve 100% safety, which is also the biggest advantage and necessity of vehicle-road collaborative autonomous driving.
We tend to divide the automatic driving scheme into the 1.0 era and the 2.0 era, 1.0 is the autonomous driving technology with bicycle intelligence as the core, and the 2.0 era is the complete set of automatic driving solutions with "bicycle intelligence + vehicle-road collaboration". We are now moving forward very resolutely along this 2.0 scenario.
"Silicon Valley 101": Disassemble the "car-road collaboration", "car" everyone understands, what does "road" mean?
Zhu Lei: We will put several devices on the roadside: First, roadside sensors similar to those on the car, including lidar, cameras, and millimeter-wave radar.
The other part of the equipment is the roadside communication equipment, including the roadside communication unit RSU, 5G civil network and other communication network equipment; there is also a part of the computing equipment, such as edge computing servers, can be understood as a complete traffic perception system.
It uses the data perceived by the sensor to do real-time calculations on the roadside edge server, and the results are fed back to all vehicles on the road for the first time, which can support autonomous vehicles, and can also provide more traffic information for ordinary vehicles, and digitize the entire traffic in real time.
Silicon Valley 101: Is this planned together with the laying of 5G?
Zhu Lei: Yes, 5G is a roadside communication mode, one is a 5G civil network mode, and one is a C-V2X vehicle-road collaboration of the vpc, these two sets of communication schemes will be considered in parallel according to the actual situation or use one of them.
"Silicon Valley 101": What does your cooperation with Hunan Hengyang probably include, and what do you want to build in the end?
Zhu Lei: The Hengyang project is a continuation of our projects in Shunyi and Suzhou. The Hengyang project is very characteristic that it should be the first city in China or even the world to realize large-scale operation of autonomous driving.
We mainly do a few things in our actual operations:
First, help the entire city to do infrastructure upgrades, or ride-hailing road collaboration digital road upgrades. This part is mainly to install a lot of sensors and computing equipment on the roadside to digitize traffic. This is mainly related to the government's new infrastructure.
Second, we will provide city-level autonomous public service vehicle upgrades and fleet operation services, including municipal public service vehicles such as buses, taxis, patrol cars, and sweepers.
In addition, after the completion of infrastructure construction, we will provide more services for ordinary users, including digital twin systems, lane-level navigation, and lane-level digital information presentation, so that ordinary users can get more safe driving services when driving. After the implementation of the whole plan, the overall digital transportation construction of the city is a huge improvement.
"Silicon Valley 101": When will it be landed, and what stage is it progressing to now?
Zhu Lei: It is under rapid construction, and it is estimated that it will be fully operational in Q3 and Q4 this year.
Silicon Valley 101: Then you guys are fast.
Zhu Lei: Yes, we are building infrastructure very quickly, about half a year or so can build the city's main roads, and at the same time we will invest about thousands of autonomous vehicles for public services in the entire main urban area to help the automatic operation of public services.
When you visit, many public service vehicles are already in the state of automatic driving, which should also be the most typical case in China on the large-scale landing of city-level autonomous driving.
We hope to use this case to design future urban smart transportation, and use this model to copy the entire solution to more cities and more scenes.
"Silicon Valley 101": Hengyang Road is also equipped with sensors, can it prevent accidents such as vehicle collisions?
Zhu Lei: You are particularly right, there are two main core factors in driving safety: one is whether you see the dangerous situation around you; the other is whether there is enough time to deal with the dangerous situation after seeing it. One is information comprehensiveness, and the other is decision-making timeliness. These two problems will face theoretical bottlenecks in bicycle intelligence, and some special situations will be encountered in practice.
For example, there are no traffic lights at intersections in many cities in China, if two cars in the vertical direction are coming at high speed, only the bicycle sensor cannot perceive the 90-degree car, and it is easy to collide. However, if there is roadside equipment, the global information of the intersection will be synchronized to each car in real time, which completely solves the problem in terms of information comprehensiveness.
Second, global information can be prejudged, that is, all traffic information 1 km or 10 km away can be predicted in advance. In terms of comprehensiveness of information and timeliness of decision-making, the infrastructure on the side of the road will greatly improve the safety of autonomous vehicles or ordinary cars.
Silicon Valley 101: Is this also very demanding on signal transmission delay?
Zhu Lei: People's reaction time to an event is about 500 milliseconds, which can generally be done in 300 milliseconds on this device, and now the whole system architecture is perceived from the roadside to the decision-making on the car, and the time can be controlled within 100 milliseconds.
This time is definitely leading in the industry, about 3 to 4 times faster than the general speed in the industry, and the overall processing delay has been greatly improved.
"Silicon Valley 101": The whole processing process will also produce very big data, Waymo's early processing of data or rely on engineers to go to the car at night to get the hard disk directly copied. How do you transfer and store data?
Zhu Lei: First, on the basis of the transmission of communications, 5G and C-V2X, these new networks, the transmission speed and the amount of transmitted data will be greatly improved.
The second is in computing. In the past, in order to ensure computing power, the computing power of bicycle calculations may generally be 300-400TOPs. Now the computing power of some new energy vehicles wants to rise to such a level as 1000 TOPs.
We can actually process the existing data by maintaining the bicycle at 300-400TOPs. On the entire device on the roadside, because there are very many edge processing servers, which is equivalent to dispersing the previous proprietary computers to the computing devices at both ends of the road, the computing power of edge computing has also made a huge leap forward.
Whether it is from the perspective of computing power, or data transmission efficiency and transmission volume, the whole set of solutions has been greatly upgraded. This is also why I said that bicycle intelligent plus vehicle road collaboration may be the beginning of the 2.0 era of automatic driving.
"Silicon Valley 101": To deal with a lot of edge data, the technical threshold is still quite high. Mr. Zhou, what do you think of this project?
Zhou Wei: I also talked about this project with Mr. Zhu, and I personally think that the vehicle-road collaboration is an intermediate stage, and I still believe in Elon Musk's point of view, and in the end it should not be used in this way.
In the process of commercialization of technology, commercialization can be achieved in advance through some so-called hybrid methods. This process can accumulate a lot of knowledge, data and related resources, so that it can do greater things in the future.
In the current Environment in China, it is feasible to adopt the vehicle-road synergy approach and Hengyang's operating model. Now the Chinese government is also considering experimenting with vehicle-to-road coordination in some places, including highways. This approach involves a new set of infrastructure construction, which must be cooperated with the government, or a considerable workload, and it is too difficult for a single enterprise to do it.
Although this is a transitional window, it is also valuable for the future, but it can't stay here, to know what is accumulated in the process, the ultimate goal is to enter the next battlefield of fully autonomous driving.
"Silicon Valley 101": Is the data of vehicle-road collaborative training meaningful for full automatic driving?
Zhou Wei: The road conditions are definitely changing all the time, but I have invested for 14 years, and now I summarize some of the opportunities I have missed, because the problem is too ultimate. I am a hardcore science fiction enthusiast, and the result of watching too much science fiction is that I always think about what the final mode is, and I will find that some things are transitional things, not the final model and product, and may not be ready to participate.
Some things have very long intermediate stages, and how useful this data is depends on the point in time at which the problem is viewed, and at a point in time in an intermediate stage, this data must be very useful.
I have also seen many companies achieve large-scale marketization and capital market success through technology solutions or transition solutions. For example, China's balance car company No. 9 Robot has obtained the Market for To C by making some relatively rudimentary balance car products, about a few hundred or a thousand pieces of products. But there is a huge gap between patents and technology, and Segway in the United States.
Segway has always taken the professional route, it is very expensive, its market has never been opened, and finally the more successful Chinese small company has acquired all of Segway's patents and products, becoming more powerful.
A company that chooses a transition plan can become a great company, provided that it knows what its goals are, rather than getting caught up in its own transition plan or technical solution. Eventually, when the new generation of technology matures, it is likely to be an instantaneous change, and it must be dealt with.
02 The landing of automatic driving should have the perspective of urban designers
"Silicon Valley 101": Before the Hengyang model, everyone talked about automatic driving and talked about Beijing Shunyi. You also built the country's first 5G open commercial vehicle road synergy road in Shunyi North Xiaoying Town, many autonomous driving in the test, you already have Shunyi experience, Hunan Hengyang project and Shunyi what is the difference? What new gains have you gained from these two collaborations?
Zhu Lei: The first is the expansion of scale. We have full coverage of the main urban area of Hengyang Hi, and there should be no such a large range of autonomous driving in the world.
The second is the advancement of the entire program. From Shunyi to Hengyang, the two sets of solutions of bicycle intelligent plus road coordination have been very mature. Many views in the industry believe that bicycle intelligence and vehicle-road collaboration are two technical routes, and I completely disagree.
I have always believed that bicycle intelligent plus vehicle road collaboration is an upgraded version of the bicycle intelligence 2.0 era, and bicycle intelligence should be done to the extreme in any case, which is the inevitability of technological development.
From a business point of view, it also involves a more macro perspective, that is, the whole social structure, including the issue of social development. From the perspective of urban management, if there are many autonomous vehicles in operation in the city, city managers will think that it is very unrealistic for each bicycle to operate independently in the city.
Therefore, all autonomous vehicles must be connected to the unified management service of the city to do unified scheduling and information sharing, in order to truly solve the efficiency problem of the city, and the systematic efficiency will far exceed the efficiency of a single point.
Another issue to consider is resource consumption. The most core resource competition in the future commercial competition is the urban public resources, when the autonomous vehicles in the city to do large-scale landing, the ultimate test is not only the technology, but the road occupied by the vehicle, parking resources, personnel consumption problems.
When the entire autonomous driving public service system is landed in the city, it must be an orderly, controllable, controllable resource that is included in the government's routine management.
We are relatively far ahead of the company, but it is estimated that in the next year or so, this topic will float to the participants of each industry: in the end, when the city-level large-scale commercialization is landed, what is the most important factor to consider?
In my opinion, the biggest difference between Hengyang and Shunyi is that the dimension of thinking has switched to the dimension of the entire city rather than a separate autonomous driving company, which may be some of the key factors we consider in the process of landing in more cities in the future.
"Silicon Valley 101": Will you later copy the Hengyang model to other cities on a large scale, and how do you decide which cities to go to first?
Zhu Lei: It's a question of business strategy. On the one hand, our current environment is very good, and with the support of 5G, new infrastructure, intelligent manufacturing, scientific and technological innovation and other strategic directions, there are many cities moving towards digital transportation and the future of intelligent driving. On the other hand, we may also have to fully consider the local development situation, and we will give priority to selecting some developed regions to operate first.
"Silicon Valley 101": Now that autonomous driving is in the scenario-based commercial stage, Robotaxi is indeed a big market after the technology is fully mature, but it still needs a lot of capital to pass the intermediate stage, so everyone is grabbing customers.
Zhu Lei: Autonomous driving has been developed for so many years, and from the second half of 2019, commercialization has entered the fast running stage. This point of the industry is actually gradually converging, everyone realizes that in some specific scenarios or public service systems, automatic driving should be prioritized.
03 Commercialization scenarios, transitions and endgames
Silicon Valley 101: Mushroom Car Union is a ready-made commercial template for autonomous driving. Mr. Zhou, what other realistic commercialization paths do you see?
Zhou Wei: Zhu always made a relatively complete operation plan, which is a relatively grand solution. At present, in semi-enclosed spaces or controlled environments with relatively fixed routes, the commercialization of autonomous driving is progressing very rapidly, including vehicles like urban sweeping vehicles.
Some companies are now doing well in the local area, and there are also some mines, port areas, park warehouses, and intelligent logistics. For example, the "fast warehouse" we invested in, although it is called "fast warehouse robot", is actually operating automatic driving technology to achieve intelligent warehousing unmanned.
What individuals are most looking forward to is fully autonomous driving, ideally automatically lined up like a train, and it is conceivable that if it is realized, the traffic situation will be much better.
It will be like watching a science fiction movie, all cars are lined up, the same speed, do not have to worry about the risk of collision, and automatically leave the formation when you exit. This is still a bit far away, the current large-scale commercialization in a few years, or the changes in these scenarios just described.
Silicon Valley 101: What do you value more from an investment perspective? Is it the ability to achieve fully autonomous driving or to solve problems in business scenarios?
Zhou Wei: We are an institution that focuses on early-stage investment, and most of the investment in the past few years has been the first round of the project. In this case, I personally hope to achieve the ultimate goal in the future, although it will take many years, but as a team out of silicon valley genes, we believe that the long-term vision can bring great rewards. Although the time will be long, we have such patience.
As far as CCV is concerned, the dilemma we face is that we only raised funds at the end of 2017 and began to invest, when all the companies that focus on high-speed autonomous driving are already valued at billions of dollars, which is far beyond our range direction. That's why we chose the direction of low-speed autonomous driving later, because at that time we were still early enough to enter.
In terms of my tendencies and hobbies, I certainly want to invest in companies of the real future. On the one hand, we want to make money for investors, and on the other hand, we also have to contribute to the process of realizing this dream, so we will also participate in some specific scenarios and transitional content.
Like automatic forklift trucks, in addition to automatic driving technology, forklift jacks, angles in all aspects must be very precise operation, the need for a lot of vertical field accumulation. Such vertical scenes will always exist, and this is also the direction of our special attention.
Silicon Valley 101: Investing is also a matter of timing. Although everyone is the same in concept and direction, but different periods of entry, the price difference is very large.
Wei Zhou: When I was at KPCB, I had a long time to focus on green technology investment, but at that time some technologies and directions were not perfect enough to generate huge business value.
But recently, because of carbon peaking and carbon neutrality, there has been a new wave of green technology investment. The concept is the same, the original solar energy, wind energy and other new energy things, in that era because of the cost problem, if there is no subsidy, it is impossible to commercialize, but today many are not allowed to need subsidies. Timing is very important.
Silicon Valley 101: With the development of 5G, will you make an investment layout in this regard?
Zhou Wei: We've seen two rounds of edge cloud computing, and we've been watching intensively lately. Edge computing is huge to some extent, whether to combine edge computing and 5G is a direction that many companies doing edge cloud computing are exploring, and Telecom and Huawei have many actions in this regard. The answer is indeed correct, we are currently very focused on these directions and are ready to make a lot of investments in them.
"Silicon Valley 101": Mr. Zhu, you are doing a lot of things now, including urban infrastructure, autonomous vehicles, vehicle transformation, industry solutions, do you think the battle line is too long?
Zhu Lei: This may be from a third-party perspective. For us, what we do can be summed up in two ways.
First, at the technical level, we adhere to the "bicycle intelligence + vehicle-road collaboration" of the whole set of programs to land, we internally called "vehicle-road cloud integration" system, in the whole system does not separate the car, road, cloud data separate, is the vehicle-road cloud integration system, is a set of large-scale complete Internet system architecture.
The difference between this system architecture and the architecture done in Baidu and Didi is that this may be the largest, most complex, and most data processing real-time system architecture in the entire Internet system, which is also our most core advantage.
Historically, no product has had such high requirements for real-time and data volume scale. Whether it is autopilot at the end of the car, roadside infrastructure construction or overall data collaboration in the cloud, the biggest challenge is the construction of a complete set of real-time systems.
Second, we mainly focus on commercial operations. Whether it is a public service vehicle or a driverless taxi, it has been fully standardized in our engineering operation system and can be quickly deployed and operated.
We build all the components with these two as the core, but because this thing is too new and the time to appear is too short, everyone will feel that there is a lot of content. In fact, in the abstract, it is actually a set of technical systems plus an operating system.
Silicon Valley 101: Can you reveal how many cars are running on the road now?
Zhu Lei: At present, including running on the road and in the process of modification, it is expected that there will be about 800 or 900 units in this batch, and there will be thousands by the end of the year.
Silicon Valley 101: Can you elaborate on your current revenue structure? What is the expected future growth direction?
Zhu Lei: At present, the whole set of commercialization paths of Mushroom Car Union can be divided into three different stages: the first stage is to appear as a technical service provider, provide a complete system of technology, and have a certain business and cooperation model.
In the medium term, that is, at this stage, the typical feature of the commercial operation model of autonomous public service vehicles is that the business scenario is very solid and the operation model is very healthy, but the biggest problem is that in a certain period, the gross profit margin will be maintained at a relatively reasonable or low level.
Everyone cuts into the public service system market, more to maintain a certain gross profit margin, rather than to pursue a higher gross profit margin, which is determined by the commercial characteristics of the scene.
In the third phase, we want to build the whole set of cloud services, which is the most core piece of our entire commercial construction or future scenario. We have sensors on and on the road that provide a lot of traffic information and services in real time. The biggest feature of the business value of the cloud is that it is very large, and the business performance will be very good from a longer period.
04 The gap between China and the United States in technical talents is narrowing
Silicon Valley 101: Whether it is new energy, autonomous driving or smart roads, the automotive industry is undergoing great changes this year. What directions do you think the automotive industry needs the most talents now?
Zhou Wei: China's manufacturing industry has actually developed and accumulated a large number of talents, but the problem is that the old automobile architecture fixed thinking is relatively strong. I remember Wang Xing (founder of Meituan) turned around a post: the older generation of car designers and manufacturers' view is to add computing units and intelligent parts to the car to make the car more and more intelligent.
The new generation of smart cars is designed to think of it as a computer. In today's market, I think there is no shortage of talents in various industries and modules, what is missing is that people with vision can design cars in the direction of the future at the beginning.
In China, we've been following Tesla now, and Tesla, like Apple, redefined the car. We and Tesla will get closer and closer in terms of the underlying technology, but at the same time, we can go faster in terms of human-computer interaction interface and redefinition of the environment inside the car.
Silicon Valley 101: In the early years, only a dozen auto companies set up laboratories in Silicon Valley, and now there are hundreds. I've talked to people at some traditional car companies about what they want to develop most when they set up a lab in Silicon Valley, and they give it a word: interaction. Mr. Zhu, you have done a lot of things in the direction, what kind of talents do you think you need the most?
Zhu Lei: In the process of iterating the car forward, intelligence is the soul. I think the technology and talents of China and the United States are gradually leveling, and from the perspective of industrial development, they will go through several stages, the first stage is the stage of technological change, mainly in the competition or training of technical talents, which we feel very obviously.
In 2016-2017, the special equipment was in the field of AI and vision, and algorithm talents were very scarce, most of which were still more in the United States. However, after several years of iteration, domestic algorithm talents have been greatly improved, at least in terms of scale, including quality.
"Silicon Valley 101": Is some talents in Silicon Valley returning to China or is the mechanism for cultivating talents in China established?
Zhu Lei: Talent flow is part of the reason. The core reason is China's huge engineer dividend in recent decades. In 2009, there was still a big gap between our search technology and the United States, but by 2013 and 14, it was basically in a flat state. The cycle and efficiency of technology iterations are very high, and the speed of technology flattening across the industry and globally is very fast.
After the first stage of technology, the core of the second stage is product drive. When the technology reaches a certain maturity level, user perception will be more important. The domestic user base, user scale and product volume are maintained at a huge order of magnitude, and optimizing the user experience will be very advantageous.
The third stage is the competition for commercial talents. More talents who need to be localized and grounded, familiar with China, a certain region, a city, and a provincial market, can put the entire commercialization scene into it. So at different stages of development, the needs of companies or industries will be different.
"Silicon Valley 101": How do you see the difference between this round of automobile revolution and the previous development of mobile Internet?
Zhu Lei: China's innovation in the past mobile Internet era has focused more on the level of business models, and in terms of underlying technology or product innovation, it has followed Silicon Valley in the United States more. This is an inevitable process of development, we have a huge market scale and user volume, the overall business innovation and business scenario path must be the first.
In the era of intelligent cars with autonomous driving as the core, the biggest difference is from the era of business model innovation to the era of truly technology-driven innovation.
In the past, everyone paid more attention to product application functions, but in the era of intelligent cars, everyone paid more attention to the system covering the underlying hardware, operating system, communication, algorithm, and service application, which is a complete set of flattened technology-driven systems. This may be the greatest opportunity of the times in generations.
From the perspective of GDP, real estate ranks first, and the second is the automobile industry. In industries that account for such a large proportion of the national economy, driven by technology and hard science and technology will bring about large market opportunities and give birth to a number of great companies.
Sometimes I myself get excited to think, in the next ten, twenty or even fifty years, what changes will this industry bring to technology and life? Very, very much worth looking forward to.
Zhou Wei: It's worth looking forward to. I totally agree, and that's why I've been in a lot of discussions lately, because we still want to continue to invest upstream and downstream of the car industry chain.
Silicon Valley 101: This field is now in China's dividend period. The development of Silicon Valley to new energy and smart cars is not as enthusiastic as China, but the development of unmanned driving is still relatively fast.
Zhu Lei: Yes. In the era of smart cars, whether it is driven by technology or changed in the energy structure, I think China does occupy a very good position and is in a dividend period.