#Autonomous Driving
Aug 07, 2023
SoftBank Corp.
Blogs
Autonomous driving is anticipated to address various issues such as rural depopulation, the increase in individuals experiencing mobility difficulties due to voluntary surrender of their driver's licenses, and labor shortages in industries including logistics. Furthermore, it is expected to contribute to the reduction of traffic accidents and environmental impact. However, the actual implementation of autonomous driving technology in society presents a multitude of challenges. SoftBank’s Research Institute of Advanced Technology believes that ensuring the continuity of the business model and its economic sustainability is crucial amidst these challenges.
From the perspective of what is necessary to rapidly provide a sustainable autonomous driving service to society, we are advancing research and development as well as demonstration experiments. We believe it is extremely important to understand why it is difficult to ensure the business viability and economic sustainability of autonomous driving when aiming for service sustainability.
In parallel with advancing research and development, we are collaborating with group companies such as MONET Technologies Inc. and other enterprises, facing the challenges of business establishment daily and exploring solutions.
In this article, we will introduce our considerations on the difficulties in ensuring business and economic viability in autonomous driving and its relationship with technological development, as revealed through these activities.
The concept of a business model is one of the necessary perspectives in considering business viability, and it is a framework for thinking about how a business selects value and how it converts that value into transactions. Even in the field of technology development, when developing new technology, by defining how the relevant technology responds to market needs and how it provides value, the direction of technology development can be aligned with market needs, and the adaptability to the market can be improved. We perceive a strong correlation between technology development and business models.
Looking at the technology of autonomous driving and related business model examples, we speculate that there is still no one in the world who has a clear answer or firm belief in the business model of autonomous driving. In other words, no one has yet emerged who has found a way to establish the business viability of autonomous driving. When we hear the term autonomous driving, many business models and businesses come to mind, but we would like to consider why their realization is difficult from the perspective of a typical business.
One example is the possibility of autonomous driving replacing certain aspects of existing transportation and logistics businesses. For example, when you try to introduce autonomous driving into a taxi service, if you consider only the business viability, you need to build a business model that generates more profit than the current taxi service (there are many merits such as reducing the burden on drivers and reducing the accident rate, but for simplicity, we focus on the basic business model of carrying passengers and earning fare income). In this case, the value of autonomous driving can be considered to be "labor saving." In other words, it is the act of automating the driving process and entrusting the work that humans were doing to machines.
For instance, looking at the taxi industry, it is said that the taxi transportation revenue before the COVID-19 pandemic amounted to approximately 1.5 trillion yen (source: Ministry of Land, Infrastructure, Transport and Tourism, "Fiscal Year 2023 Transportation Policy White Paper"). Additionally, there is data indicating that labor costs, including drivers, account for about 73% of the total cost structure (source: Ministry of Land, Infrastructure, Transport and Tourism, "自動車運送事業経営指標").
In order to enter the taxi industry using autonomous driving, the total operating cost must be lower than the current driver's personnel cost. To enable autonomous driving, various initial costs such as the creation of HD3D maps for localization, development of autonomous driving software, and introduction of sensor equipment are incurred. When considering these in total, it is required that the cost is lower than driving by humans. What comes out here is the concept of multiple vehicle monitoring. This refers to an operation form in which one person monitors 10 autonomous driving taxis, for example. If one person has to keep an eye on one unmanned taxi all the time, it would be better for a human driver to drive. What if one person is monitoring 100 autonomous driving taxis (in reality, 10 people are monitoring 1000 taxis)? If we just talk about driver's personnel costs, the cost is 1/100. It becomes possible to operate a taxi with 0.01 driver per vehicle, which may create economic benefits.
On the other hand, how about from the perspective of income? The market size (Total Addressable Market: TAM) basically does not change whether it is a normal taxi or an autonomous driving taxi. Therefore, to increase income, the only option is to take away the share of the existing market, and to do so, it is necessary to gain some competitive advantage by introducing autonomous driving (for example, reducing service prices). There are various regulations, and the value provided is essentially no different from existing taxis. It is difficult to increase the income side, and a significant cost reduction is needed to ensure economic rationality. This shows the difficulty of the autonomous driving business as an "alternative for existing transportation/logistics businesses." Whether the target transportation mode is a route bus, a community bus, whether the object to be transported is a person or a thing, whether the provision area is rural or urban, the essence does not change. Whether the model is to receive fare income from consumers or to depend on taxes and subsidies, the payer is different, but essentially it is the same structure. It is necessary to solve the issues from the viewpoint of economic rationality, and it is not easy to ensure a sustainable business.
This is where the concept of MaaS comes in.
"MaaS (Mobility as a Service)" is broadly understood as the provision of existing models of transportation and their alternatives "as a service." The scope of this concept is wide, encompassing mobile medical services, convenience stores, and municipal procedure centers. These can also be considered as "alternatives to certain existing services," just as they are "alternatives to existing transportation/logistics businesses."
When we simply observe things from a business perspective, we again face the need to improve economic efficiency compared to existing services. In this context, the concept of "mobile services" suggests the possibility of expanding the market size. Specifically, by moving the service itself to the customer, it becomes possible to reach customer segments that have not been reached so far. This is one of the good points of MaaS business.
However, what we need to be careful about here is whether the factor that realizes market expansion is "autonomous driving." If it is simply a “service on wheels," a human driver can provide the same value.
Then, what is the significance of using autonomous driving in MaaS business? One is cost reduction through labor saving mentioned in the previous chapter, and the other is securing labor force. The service moves to the customer and provides the service there. However, there is a shortage of labor to make that move possible. The idea is that if it is autonomous driving, it can guarantee the labor force that is lacking.
We believe that MaaS business has the potential to create new markets. On the other hand, the basis for considering "various services on wheels" is the balance of supply and demand, just like other general services. What kind of challenges can be seen from a business perspective on this point? We would like to delve a little deeper.
When you think of "a certain service on wheels," suppose there is an attractive market. For example, shopping needs of the elderly in depopulated areas. (1) There is a "sufficient" need for "mobile supermarkets or convenience stores for the elderly" (demand), (2) It is a challenge that labor cannot be sufficiently secured (a challenge in meeting demand), (3) By using autonomous driving technology, provide an unmanned mobile supermarket (solution), (4) Create a new market for shopping needs of the elderly in depopulated areas (formation of a new market), and so on.
It looks like a win-win business as a scenario, but when you actually research and consider the business model, you can see the challenges. What kind of challenges can you see?
That is, in the case of the example mentioned above, the point is that "(1) demand" does not directly become a market. Specifically, there needs to be a situation where there is a need for a mobile supermarket for the elderly, and there is no driver. Even if there is a sufficient need for a mobile supermarket, there are two ways to access that need: "human driver" and "unmanned autonomous driving." Of these, only a part of the market where "unmanned autonomous driving" is superior can be acquired as a business.
This is the concept of TAM (Total Addressable Market), SAM (Serviceable Available Market), and SOM (Serviceable Obtainable Market) in business. TAM is the "total market size" that a certain business can potentially acquire, SAM is the "maximum market size" that a certain business can "acquire", and SOM is the market size of customers that a certain business can "actually approach."
In the case of the previous example, whether the market that "unmanned autonomous driving" can approach is "SAM" or "SOM" changes depending on the way of thinking, but at least, the market size that the company can acquire is "The need for a mobile supermarket for the elderly" will be smaller.
Of course, it is necessary to have a premise that unmanned autonomous driving has a competitive advantage over human drivers (such as lower service provision prices and costs).
In other words, in the previous example, it is necessary to have the assumption that, in addition to demand, unmanned autonomous driving holds a competitive advantage over human drivers, and that a certain number of users can be acquired within the market while ensuring economic feasibility.
Autonomous driving is a technology related to "cars", so from the perspective of the market, there are many use cases and it is very complex. In our case, we are working on the challenge of how to ensure economic rationality for autonomous driving, using the way of thinking about the communication business that we have cultivated so far and the knowledge of marketing and business models gained from various new business considerations.
In the previous chapter, we examined the MaaS business from the perspective of "mobile services," but if we think about acquiring as many customers as possible using autonomous driving, the MaaS business can also be considered to have the potential to acquire users from other modes of transportation options. This often includes private cars.
Since the value provided by autonomous driving is fundamentally rooted in “mobility”, which means that if autonomous driving is superior to other transportation options, it has the potential to attract more users. For example, when creating a business plan, if you stack up numbers on the premise of acquiring customers from various transportation options, you may be able to create a market size that looks rational at first glance.
However, the problem that must be considered here is that when considering the market from the perspective of "mobility", it ultimately becomes a competition over the pie of the existing market. When comparing your own service with "each transportation option", it is necessary that "each" is superior from the customer's perspective. Transportation and logistics are somewhat mature services. Generally, competitive advantage in a mature market is often the price of service provision. When considering a business related to autonomous driving, no matter how you capture the market, you essentially need to solve the "cost of moving people and things."
The "cost" mentioned here is not just the cost to establish autonomous driving by autonomous driving. Various costs are incurred in providing services, such as operation and maintenance. We are working on research and development from the perspective of what kind of cost is necessary when viewed from the perspective of the market, and what kind of technology development is necessary to reduce it.
From the next chapter, we will introduce a consideration about the concept of "platform."
The "platform business" is fundamentally different from the business models introduced so far. So, what's the difference?
What is common to the examples of business models introduced so far is that they are all "linear business models." A "linear business model" refers to a flow in which Company A manufactures a product, sells it to Company B, and Company B adds value and sells it to Company C or customers. Whether the product is software or a service, the basic concept does not change.
For example, in the case of unmanned taxis, our company procures ADS (Autonomous Driving System) from Company A, then integrates ADS, vehicles, and other necessary functions into our own service, and finally provides that service to Company B or end users. The same is true for MaaS. You procure/manufacture something and sell it to customers.
In contrast, platforms primarily create value by "facilitating transactions." For example, while a typical linear business model creates value by "creating products or services", a platform creates value by creating connections between something and something else and "creating transactions." General Motors "creates" cars, while Uber "creates" transactions between drivers and passengers. Under such a concept, it is not easy to form a platform business model in autonomous driving. Why is that?
Generally, successful platform business models create value outside the platform. The difficulty of the platform business model in autonomous driving is whether you can "successfully" create "value creation outside."
There are various ways of thinking about platforms, but we think there are three points. The first is whether value creation can be done outside the platform, as mentioned earlier. The second is whether the external transactions are large-scale. The third is whether it is a business model with low marginal costs. Since the word marginal cost has come up, let's first explain marginal cost.
Platforms are often a business model with low marginal costs because they are a place for transactions. Marginal cost refers to the "additional cost required to produce one more product" (the difference between marginal cost and marginal expense, the concept of management accounting, etc. is left to specialized books. Here, it is simplified for easy understanding).
Let's take the example of car manufacturing. When making a car in a car factory, the marginal cost refers to the "cost required to make one more car."
Let's say a factory has already produced 100 cars. There is customer demand, so you want to increase the production number to 200. In that case, you need to know how much cost is required for each additional unit.
The cost required to manufacture that additional unit is the "marginal cost." In this case, the marginal cost includes additional parts, labor hours, electricity, etc.
However, the cost required to manufacture the 101st car is not necessarily the same as the cost required to manufacture the 100th car. For example, the production line may already be operating at maximum capacity, and you may need to hire additional workers to manufacture one more car. Or, a large number of parts may be required, and you may be able to receive additional discounts from parts suppliers. All these elements affect the marginal cost. Taking all these into account, how much additional money is needed to make the 101st car where 100 cars are being made, that is the marginal cost.
Generally, models with high marginal costs do not increase the profit "rate" even if the scale increases (here, for simplicity, words such as operating profit rate are not used and simply explained as "profit rate"). As the number of products produced increases, the cost also increases in proportion. On the other hand, business models with low marginal costs increase the profit rate as the scale increases. Sales increase, but costs do not increase much.
Let's get back to the platform business.
The concept of marginal cost is also important in platform business. This is because, as a platform, the cost required for the platform itself does not increase significantly even if the amount of "transactions" outside increases (it is important to design it accurately). Of course, there are costs that are affected by the content of the platform, such as maintenance and operation costs and electricity bills for running servers, but the marginal cost is lower compared to the so-called "linear business model." There is a certain limit, but the more the scale increases, the more the profit rate increases.
Let's think about it with a specific example.
Earlier, we explained that the important things in platform business are (1) whether value can be created outside (2) whether the external transactions are large-scale (3) whether it is a business model with low marginal costs. For example, what about the App Store, which can be said to be a representative of the platform?
The App Store is a platform that connects users and app developers. Users download various apps, and developers sell apps. Apple receives part of the sales as a handling fee, which is the source of revenue for the App Store.
(1) Can value be created outside?
The value created by the App Store is also formed outside the App Store itself, not inside. The App Store is a place that enables "transactions" between users who want to solve daily problems and app developers who provide solutions. The value of problem solving has been created outside the platform. The "potential demand and potential supply" that had been dormant until now has been created by the App Store, creating a "new market" where smartphone apps solve the problems of individuals around the world.
(2) Are the external transactions large-scale?
As of 2022, the number of app developers registered in the App Store is about 37 million, and it is said that an average of 3,732,113,96 user accounts are looking for some app in the App Store every week (source "2022 App Store Transparency Report"). It is very large because transactions are conducted on a global basis.
(3) Is it a business model with low marginal costs?
The people who make transactions are individual users and individual app developers (or companies) around the world, but even if the number of apps lined up in the App Store increases from 100 to 200, the cost of the App Store does not double. The more the number of "transactions" (of course, transactions are conducted through the App Store in terms of commercial flow) conducted outside the App Store increases, the more the sales of handling fees that the App Store receives, not only the sales but also the profit "rate" increases.
We think it's an easy-to-understand example of a platform business. The structure is the same for GitHub and EC malls.
The platform in autonomous driving also needs to create value outside, have a low marginal cost model, and scale out, referring to past platform businesses.
For example, what kind of value can the platform for autonomous driving create outside?
There are two possible approaches. The first is that the number of things provided explodes. If the service using autonomous driving was "movement", the existence of that platform could dramatically increase the total number of "movements" in Japan. The number of "movements" of the people every day and the "movements" of tourists from home and abroad increase due to the existence of the autonomous driving platform. On the other hand, if the total number of "movements" in Japan does not change due to the emergence of the autonomous driving platform, it will become a competition for the pie with existing means of transportation, and there will be a limit to the scale that the platform can pursue at an early stage (refer to the concept of TAM/SAM/SOM explained in Chapter 4).
The second is to create a completely new value outside of "movement." For example, data business using sensing data falls into this category. By connecting to the platform, there are benefits for both people who need data and people who provide data, and some new transactions are conducted on a large scale outside the platform.
There are various challenges such as data ownership and monetization methods, but each company is believed to be considering various approaches.
The difficulty in ensuring economic rationality in the autonomous driving platform lies in "forming a specific model that creates specific value externally and generates transactions by connecting that value." Due to the nature of autonomous driving technology, it is difficult to clearly define specific value creation, and we believe that building a business model to solve this value creation challenge is a major challenge.
In order to incorporate autonomous driving technology into a business model, it is necessary to consider many elements, not just the technology itself, but also market needs, the balance of costs and revenues, etc. These elements are complexly related to each other, and a business is established only when they are well combined.
In the midst of this complex situation, we carefully analyze and structure each element. We believe it is important to clarify "which part of the market our technology specifically contributes to."
As mentioned in the article, in our case, we are tackling the challenge of how to make autonomous driving economically successful by utilizing our thinking about the communication business that we have cultivated so far and the knowledge of marketing and business models gained from considering various new businesses.
In the future, as autonomous driving is widely implemented, answers will be sought to the questions of what each company should provide, what range of responsibility it should have, and what will become the core competence.
In this context, we are advancing both technology development and business model verification in parallel, taking into consideration what our strengths and capabilities are and what challenges Japan as a whole should overcome.
We believe this will ultimately lead to implementation of autonomous driving in Japanese society as a whole.
Reference:Our Initiatives to Achieve an Autonomous Driving Society Through the Utilization of an “Autonomous Driving Operation Platform”