[Insight column] Lee Hae-seung, "If SDV is the Next to Electrification..."
Peter Thiel, the founder of PayPal, and
an early investor in Facebook, SpaceX, and LinkedIn, describes the process of
creating something entirely new as going from 0 to 1, and then turning that
creation into a massive business as going from 1 to 100. If one succeeds in the
0 to 1 phase but doesn't consider the 1 to 100 phase, one may achieve the glory
of creating something new but miss out on the profits of scaling it into a
large business. The probability of success in the 0 to 1 phase is low.
Investing in such ventures is risky. The competition in the 1 to 100 phase is
fierce. How can one minimize risks, win in the competition, and achieve 0 to
100 in SDV?
Author | Lee Hae-seung, Managing
Director, FESCAROLee Hae-seung, managing directorIn 2021, celebrating 30 years of successful corporate life, he departed from the company he had been working for. Over 26 years, he has held various roles, including testing at an automobile manufacturer (formerly SsangYong Motor, now KG Mobility), system application at an automotive supplier (formerly Delphi, now Aptiv and Continental), and technical sales at an automotive software tool company (ETAS). He had the opportunity to be part of the development of the first integration of EBS (Electronic Brake System), EPB (Electric Parking Brake), and ACC (Adaptive Cruise Control) in domestic automobiles. Throughout his career, he has been involved in model-based software development, interactive calibration manual creation, AUTOSAR, and tools and consulting related to cybersecurity. Currently, he works as a part-time consultant for PICODE, a control system test equipment company, and FESCARO, a company specializing in automotive cybersecurity and ECU fabless.How far has the auto industry come?Cars have played a significant role in providing humans with the value of "freedom of movement." As more people have embraced this freedom, automotive technology and industry have evolved and will continue to do so, enabling even more individuals to enjoy greater mobility.However, most of the vehicles that have contributed to this freedom so far have been internal combustion engine cars, which have left a considerable carbon footprint in the atmosphere despite their positive contributions to humanity. (The road transport sector contributed to 13.9% of Korea's total greenhouse gas emissions in 2019 and [1] 16% globally. [2] This sector includes not only automobiles but also aircraft and ships.) The 6th report of the Intergovernmental Panel on Climate Change (IPCC) confirms that greenhouse gases emitted by humans are the clear cause of the current global warming. [3] This reality has resulted in global warming progressing from "climate change" to "climate catastrophe," posing a severe threat to human survival. Consequently, while automobiles have greatly enhanced the value of freedom of movement, they have also posed a significant threat to human survival, on par with their contributions.In response to these environmental concerns, the automobile industry has introduced electrification technology to promote eco-friendliness while maintaining the value of mobility. As a result, the widespread adoption of electric vehicles is now being active while still providing the essential value of mobility to people.Where is the auto industry heading?Electrification serves as a fresh foundation for the advancement of automotive technology and industry. This transformation allows for the convergence of semiconductor, communication, software, and automobile technologies on a larger and more diverse scale, known as CASE (Connected, Autonomous, Shared, and Electrified). While cars have long contributed to the value of freedom of movement, they now offer additional benefits through developments like vehicle-sharing systems. By optimizing car operation efficiency through increased car-sharing services, fewer cars are needed for the same amount of travel, leading to reduced resource consumption and environmental damage. CASE vehicles contribute to eco-friendliness both through electrification and sharing initiatives.Concerns arise about potential negative industrial growth due to reduced automobile production. While these concerns are acknowledged, it is essential to consider that an industry's size aligns with the value it generates, with people willing to pay a reasonable price for this value. Before CASE technologies, automobiles primarily contributed to freedom of movement, which relied on quantitative growth through increased car sales. However, CASE technologies provide a broader spectrum of values compared to conventional vehicles. A smart auto industry can focus on maximizing value contribution to outweigh the decrease in sales volume, transitioning from quantity-driven growth to a combination of quantity and quality through CASE technology.Beyond eco-friendliness, other values come into play. Road accidents claim 1.35 million lives annually, leading to an estimated economic loss of 2,390 trillion won/year (1.8 trillion USD), accounting for 63% of the automobile market.(The size of the global automotive market was estimated to be USD 2.86 trillion in 2021. [5]) CASE technologies, particularly the "autonomous driving traffic system," have the potential to eliminate traffic accident deaths entirely, reflecting a real value far surpassing its economic implications.Furthermore, CASE can further enhance the value of freedom of movement. Autonomous driving systems allow individuals unable to drive due to age or disabilities to regain their freedom of movement. By establishing a flexible "public transportation supply system" that adapts to real-time demand during commute times, passengers enjoy safer, more comfortable transfers, freeing up valuable time for various activities while on the move, including entertainment, gaming, shopping, healthcare, education, and finance.The automobile industry actively develops various systems based on CASE, with industry leaders envisioning these systems expanding into a new field centered around "movement." More and more people are embracing the narrative. To embrace the words of Chinese novelist Lu Xun: "Originally, there were no roads on the ground. If there are more people walking, it becomes a road." The automotive industry is moving towards a service-oriented approach, referred to as Mobility as a Service (MaaS). However, this service domain is not confined solely to mobility. Automakers aim to create services that assist passengers in various activities while they are on the move, ultimately transforming these services into novel revenue streams.Software-Defined VehicleAccording to the 2020 statistics released by the Korea Auto Dismantlement Recycling Association, the average lifespan of a car in Korea is 15.6 years, nearly doubling from 8.3 years in 2000 when statistics began. [6] In 2020, there were 25.5 million registered vehicles. [7] Let's conduct a thought experiment: What if the Korean automobile market was saturated in 2020? In other words, if one new car is sold only when one is scrapped? For ease of calculation, let's assume that the number of cars accumulated over 15.6 years is 25.5 million. Then 1.44 million cars are scrapped every year, and that number is sold as new. In fact, domestic sales of automobiles in 2020 were 1.91 million units, 1.73 million units in 2021, and 1.68 million units in 2022. [8] An increase in the average lifespan of a car, a reduction in accidents due to autonomous driving, and shared cars are factors contributing to the decline in car demand. How can the saturated auto industry grow further?Looking
at sales statistics, the smartphone market reached market saturation first. [9]
The automotive industry can learn from this situation. Manufacturers work to
increase revenue per smartphone. In terms of hardware, smartphones will be
equipped with better displays, faster processors, larger memory, and more
cameras to strengthen existing functions. To introduce new functions, new
sensors and foldable form factors are applied. This strategy compensates for
the drop in profit due to reduced sales volume with a high unit price.Figure 1 | Worldwide smartphone sales and revenue trendsThere is also a response on the
software side. The situation in the app market differs from that of hardware.
Software sales are not as affected by new sales as hardware because
already-sold smartphones form the foundation of the business. Cumulative
smartphone sales have not reached saturation yet. Unlike hardware, software
generates revenue every year, depending on the business model. Due to these reasons
and others, the app market, valued at 273 trillion won/year (206.85 billion
USD) in 2022, is expected to grow at an average annual rate of 13.8% from 2023
to 2030. [10] The key is to develop valuable services that customers are
willing to pay for.The automobile industry can take a
similar approach to the smartphone industry. On the hardware side, they have
already started implementing approaches like adding ADAS sensors, installing
multiple internal displays with larger and higher resolutions, and applying
massage chairs. Additionally, automakers want to create a car software market
that can generate steady revenue based on cars already sold, despite market
saturation, just like the smartphone software market.I intentionally used the word
"system" when explaining CASE and mobility services above. This is
because there is an intention to mention two aspects of mobility service,
asserting that "the system has several components, there is linkage between
elements, and the linkage is implemented through software."
The first aspect is that, just as
services using smartphones are implemented in software, mobility services using
cars are also implemented in software. (This is an inevitable parallel.)
Instead of calling these cars smart cars, the auto industry emphasizes software
and calls them Software-Defined Vehicles (SDVs). (It may have been because
smart cars have an image of self-driving cars.)Difficulties in developing SDVs
The second aspect is that, unlike a
smartphone as a single object, certain cars are composed of a network of
function-specific controllers. Automotive engineers are new to software.
Unfamiliarity comes with complexity. Adding complexity is the fact that they
have to deal with not a single controller, but a network of controllers. The
network consists of smaller networks and controllers. Various communication
protocols such as CAN, LIN, and Ethernet that are applied here add more
complexities. The complexity of network structures and protocols complicates
technology development as well as collaboration between development entities.SDVs must be able to distribute the
new software to the entire fleet, to be precise, to all the controllers in the
vehicle involved in the service. Distribution is a so-called Over-The-Air (OTA)
software update. It is the same as how a smartphone updates software through
wireless communication. SDVs should be able to update OTA software.
Implementing OTA software update of not one controller but several controllers
connected to the network is more complicated than software update of a single
object, a smartphone. In addition to this, most automakers are in a position to
allow others' non-standard methods for controller software updates. Complexity
increases with various software update methods. Due to this complexity, it is
more difficult to develop SDVs.Software development for SDVs cycles
"planning, developing, verifying, distributing, operating,
monitoring" repeatedly as shown in the left figure in Figure 2. It is a CD (Continuous Development) method, which is also called
CI (Continuous Integration), CT (Continuous Testing), CX (Continuous
Something), and the like. The traditional V-model in the automotive industry,
involving top-down and bottom-up approaches, generally concludes the
development process with the start of mass production, whereas the new trend in
the automotive industry requires continuous (and possibly intensified)
development to accommodate new services, even after mass production begins,
representing a fundamental distinction.
Figure 2 | DevOps (ⓒ FESCARO) and V-model https://commons.wikimedia.org/wiki/File:V-model.svgMobility services are implemented
with software that maximizes the potential of the hardware in the car (it may
be built-in, passengers may carry it on board, or it may be added externally by
wired/wireless connection). Hardware is difficult to change once mass
production begins. (Some hardware of SDVs may need to be developed in
consideration of improvement changes after mass production. Tesla upgraded the
FSD hardware. [11]) So hardware development is likely to follow the traditional
V-model.So, is the software easy to change?
The appropriate answer to this question is not "yes" or
"no," but rather "automobiles should be developed to facilitate
easy software changes." If the hardware development cycle according to the
V-model takes several years, the software development cycle for a CD could
follow a similar timeline, but it could also vary significantly, ranging from
several months to a few days. These differences add complexities to project
management in otherwise complex technology development.
The development of SDV starts at a
higher level of complexity than the development of existing cars. The
automotive industry knows it well through experience. Complexity increases
development time, raises development costs, and lowers quality. One of the
objectives that SDV development should pursue is "simplification", reducing the complexity not only of the vehicles themselves but
also of the development environment.How is service development carried out in the field?In the industry, there are various
attempts, but besides Tesla's achievement in autonomous driving, there seems to
be no definitive service that has reached a significant revenue position
through software updates yet. CNBC estimates that the revenue of the iPhone App
Store in 2021 is between 706 billion USD and 857 billion USD, based on Apple's
announcement. [12] On average, it is 103 trillion won. It is an amount that can
be divided among 1.03 million people, each receiving 100 million won.
Considering the costs involved in the business and Apple's profits, it is
unlikely that there would be 1.03 million job opportunities for developers with
an actual annual income of 100 million won each. Indeed, it is estimated that
there would have been tens of thousands of job opportunities with decent income
created as a result. It will be larger if we include the Android side. I don't
know how big the automotive software market will be. Its size depends on how
many service ideas consumers value. Imagine how service ideas and software
development will work in the field.○ Ideation:• Good service ideas often come to automotive researchers (which may
include anyone, even those not directly involved in the automotive industry.).
Some of the ideas will be centered around artificial intelligence technology.
Ideas like this are more likely to visit AI engineers than automotive
engineers. Automakers need to make it easy for these people to find them.• Several of them propose their ideas to the team leader.
• The team leader selects ideas that are worth verifying among ideas
and organizes TFT (Task Force Teams) for feasibility verification.○ idea verification:■ TFT downloads
data necessary for service development from the data cloud. • The data in the cloud is collected from cars in real operation. - There may be a business that collects and sells data [13], or
provides the result of processing data upon request. • The required data may not be available. - Software for controllers that can serve as sources and for
controllers involved in data transmission is being developed. → It is advantageous if the electrical/electronic/software
architecture of the SDV is developed in advance so that the scale of software
change for data collection is small. - Data collection software is distributed to automotives of a sample group. - The data uploaded by the sample group to the data cloud is
obtained.■ Process the data
to analyze the technical feasibility of implementing the service. • Simulation is done to save analysis cost and time. - For accurate analysis, it is simulated using data collected from
running cars. - Simulation can be done on a PC, but considering speed,
collaboration, and security, it would be more efficient to do it in the cloud. • It is confirmed whether or not the software of the output
controller needs to be changed to provide the service. - If necessary, develop prototype software for the service output
controller. → Prototype software can also be implemented with a virtual
controller or rapid prototyping to reduce development time and cost. → It is advantageous if a predetermined
electrical/electronic/software architecture for SDV is developed in advance, so that the
scale of software change for service provision is small.■ Evaluate the
utility of the service. • Simulation is done to save evaluation costs and time. • The simulation confirms the effectiveness of the service and its
potential side effects in various situations. • Based on the simulation results, it is evaluated whether the
service is as effective as expected.○ Escalation &
Ok-to-develop: • If the service utility evaluation is positive, request development
approval from management. • The management level makes a comprehensive judgment and approves
the development. - There are many other things to consider besides the expected
customer response. For example, ideas of other TFTs, expected return on
investment, development capacity, etc.○ Development: • Controller developers shall; - modify the software of controllers related to providing services - and related to data collection. • Back-end developers develop server software for services. • When necessary, service app developers develop smartphone software. • Development, analysis, and evaluation are repeated until it is
judged that the utility of the service is sufficient through simulation and
real-vehicle evaluation. • Also ensure that new features do not affect existing features. • The process simply mentioned above consists of various
micro-tasks. Without simulation and automation (DevOps), the above
development-evaluation cycle would not be practical.○ Production: • Promote and sell the services. • The software via OTA is distributed to the customer's
car. • For various reasons, the usage environment considered during
development and the actual usage environment of the customer may differ. • The continuous improvement (CI) through the "operate,
monitor, plan, develop, verify, evaluate, and deploy" cycle is sustained. • Some of these developed and marketed services will be commercially
successful. Some of these successful services will have great success. • Successful services are further improved, which brings additional
profits.The above imagination
supersimplifies the process of developing a service idea into the software.
Many important subjects are missing. Among the missing topics, one fundamental
aspect crucial for SDV is cybersecurity. However, there is no feeling that the number
of topics listed above is insufficient. Rather, there are so many that makes it feel the burden of where and how to find the necessary development manpower in all these fields. Each topic, including missing topics, adds to the complexity of
development. It makes the already scarce development manpower scarcer. SDVs are
a significant burden for automakers. McKinsey predicted in 2019 that automakers
would need to invest 92 trillion won (70 billion USD) over the next 10 years to
be in a strong position. [14] Not all automakers will be able to make this
level of investment. Even for companies capable of making such significant
investments, selecting the wrong investment target can lead to severe
consequences and heavy losses.In all subjects, complexity should
be eliminated as much as possible, and simplicity should be pursued.Limited resources necessitate the selection of a focused topic
to concentrate on.
They need to make partners who will focus on the remaining
topics.How to reduce complexity?Indeed, given the limitations of space, it seems appropriate to transition to a discussion on how to reduce complexity, and we can conclude the conversation at this point. On the other hand, it is not possible to cover various topics with limited space.I pay attention to the role of data in the process of developing service ideas into software. Data is needed to objectively analyze the feasibility of a service idea. Data is also needed to rationally evaluate service effectiveness through simulation. Service implementation requires data communication between the vehicle and back-end, and between controllers. (Software update is also the communication of data called software between them.) Vehicles should be developed to facilitate data collection, transmission, processing, and utilization. Cars need an architecture that takes data into account, both electrical and electronic architecture and software architecture.I propose electrical and software architecture design as a way to reduce complexity, with data (ease of collection, transmission, processing, and utilization) being one of the criteria for making design decisions.Complexity begets complexity. If the architecture on which the service software is to be built is complex, the software itself will inevitably become complex. Software updates also become complicated. Complexity costs resources, time, money, and quality. So, the software architecture needs to be simplified. The automotive industry has developed an open software architecture called AUTOSAR early on. As SDV developed, AUTOSAR also developed. AUTOSAR is divided into classic AUTOSAR for dedicated controllers specialized in real-time control and adaptive AUTOSAR for upper-level controllers that manage flexible function changes and controllers through software updates.Classic AUTOSAR provided standards in three areas: architecture, methodology (software development methods), and APIs (Application Programming Interfaces). Architecture and methodology are inherent in AUTOSAR software development tools. That is why it is evaluated that it has taken its place. The API, on the other hand, is not. The automotive industry lacks a unified standard API and it is uncertain whether the differences in commercial relationships between car manufacturers and their collaborators, or lack of technical knowledge in each other's domains, or other factors were the sole causes. An industry-wide standard API is a far-reaching goal at this stage. Among the automakers that collaborate with multiple partners, how many automakers would there be that have a standard API which spans the partners? In reality, many cars are equipped with controllers to which AUTOSAR is not applied. Simplification is required by applying standard software architecture along with the application of standard API. If complexity is not reduced at this stage, subsequent development stages will have to deal with exponentially increasing complexity.In addition, the application of a standard software architecture reduces the complexity of software portability between controllers or controller integration, which will inevitably be pursued in the future. It simplifies not only the subsequent stages of current development, but also future ones. The automotive industry has learned a lesson from the classic AUTOSAR API failure. Adaptive AUTOSAR serves as middleware providing standard APIs to service software. Adaptive AUTOSAR has to be made to go down a different path than classic AUTOSAR's API.In addition, the application of a standard software architecture is a prerequisite for collaboration between automakers and partners, and between partners and partners. Collaboration allows everyone to focus limited resources on their own needs. Automakers should focus on enhancing their competitiveness in the core field of automobiles. Suppliers should focus on competitiveness in the core field of the system. The rest of the fields are secondary. Sub-fields are those that target the core business. They should “create” partners who focus on those ancillary fields as their core business targets in order to build competitiveness through them.Automakers that cannot make sufficient investments in the development of CASE technology for SDV should collaborate and join forces. Indeed, potential partners for collaboration can be found not only within the automotive industry but also outside of it. By joining forces, there is an opportunity To secure a broad territory in the emerging realm of SDV (Self-Driving Vehicles).CollaborationPeter Thiel, the founder of PayPal and an early investor in Facebook, SpaceX, and LinkedIn, explains the concept of going "from 0 to 1" by creating something entirely new and "from 1 to 100" by scaling it into a large business. If one succeeds in the 0 to 1 phase but doesn't consider the 1 to 100 phase, one may achieve the glory of creating something new but miss out on the profits of scaling it into a large business. The probability of success in the 0 to 1 phase is low, making investing in such ventures risky. Additionally, the competition in the 1 to 100 phase is fierce. So, how can one minimize risks, win in the competition, and achieve 0 to 100 in the SDV industry? It seems that there is not only one way, and I believe that collaboration will be involved in some form in any approach. Indeed, SDV is not simple enough to handle everything on its own. What collaborations would be possible? In the fields I have covered, there are the following collaborations.Cybersecurity for automotive control units relies on encryption management and cryptographic operations. Increased connectivity has increased the demand for cybersecurity. Chips are increasingly incorporating cryptographic accelerators (referred to as HSM - Hardware Security Module or HSE - Hardware Security Extension) to support encryption management and operations. Hardware without software is useless. It is very inefficient for someone who develops the software for a service to have to deal with the software that runs the cryptographic accelerator as well. The similarities and differences between the two fields can be compared to Korean and English. Therefore, companies specializing in cybersecurity software have been created for automakers, suppliers, and chip makers to collaborate with them.The concept mentioned above is not entirely new and not solely due to cybersecurity. It can be extended to other fields. Long before the emergence of cybersecurity, AUTOSAR was designed with a multi-layered architecture consisting of the ASW (Application Software) layer, where functionalities are implemented, and the BSW (Basic Software) layer, which supports the ASW. Indeed, specialized companies have emerged, focusing on developing BSW as a product and making it their business target. As AUTOSAR expands, automotive software companies that are targeting BSW as a service product are emerging. Automotive companies and their partners cooperate with companies specializing in automotive software.This collaboration spreads the 0 to 1 investment risk, as each one's contribution enhances the technical completeness, raising the likelihood of success. However, this can only happen when everyone is doing their best. There will be many ways to ensure that everyone is performing at their best. One common and effective approach among them is setting a shared goal of dividing the benefits of achieving the 1 to 100 milestone. Achieving the 1 phase provides a solid foundation to secure the necessary investments for expanding to 100. This business approach is completely unfamiliar to the automotive industry. SDV is already challenging with just a change in technology. Will the automotive industry be able to embrace business-side changes and make SDV a success? I hope so. And I hope that success will happen around me. If you are looking for a software solution partner for customer-centric engineering in the automotive industry, FESCARO could be a viable and promising option.[1] Korea Institute of Energy Research-Technology Policy Platform Energy & Data-KIER TPP (https://www.kier.re.kr/tpp/energy/B/view/184?contentsName=sub2_4&menuId=MENU00962)[2] Bill Gates. How to avoid a climate disaster. Chapter 7 How to get around.[3] ipcc 6th report. Climate Change 2023 Synthesis Report. Summary for Policymakers https://www.ipcc.ch/report/ar6/syr/downloads/report/IPCC_AR6_SYR_SPM.pdf[4] US Agency for Disease Control, website https://www.cdc.gov/injury/features/global-road-safety/index.html[5] Global automotive manufacturing market size 2022 | Statista (https://www.statista.com/statistics/574151/global-automotive-industry-revenue)[6] 2020 vehicle scrap statistics, Korea Automobile Dismantling and Recycling Association (kadra.or.kr) (http://kadra.or.kr/kadra/contents/sub02/02_01.html?idx_num=2536&pidx_num=&skin=4&mode=view&GotoPage =1&column=&keyword= ֱ &BbsId_01=&BbsId_02=&BbsId_03=&MenuNum=&BbsId=2010316112316&Tbn=&srhData1=)[7] Vehicle Registration Indicator Service | e-Country Index (in-dex.go.kr)( https://www.index.go.kr/unity/potal/main/EachDtlPageDetail.do;jsessionid=T-Oeb3-jdaLfPaKWUMeotl7cNreep-2ONoGkUPwq.node11?idx_cd= 1257)[8] Vehicle production, Korea Automobile Mobility Industry Association (KAMA) ( https://www.kama.or.kr/NewsController?cmd=V&boardmaster_id=industry&board_id=12274&menunum=0004&searchGubun=&searchValue=&pagenum=1)[9] Global smartphone sales and revenue Smartphone sales worldwide 2007-2021 | Statista Global smartphone revenue 2011-2022 | Statista(https://www.statista.com/statistics/263437/global-smartphone-sales-to-end-users-since-2007/)[10] Mobile Application Market Size, Share & Trends Report, 2030 (grandviewresearch.com) (https://www.grandviewresearch.com/industry-analysis/mobile-application-market)[11] Full self-driving computer installation | Tesla, South Korea (https://www.tesla.com/en_kr/support/full-self-driving-computer)[12] Apple implies it generated record revenue from App Store during 2021 (cnbc.com) (https://www.cnbc.com/2022/01/10/apple-implies-it-generated-record-revenue-from-app-store-during-2021-.html)[13] Monetizing connectivity services and car data. Value creation and performance in the auto industry | McKinsey (https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/down-but-not-out-how-automakers-can-create-value-in-an-uncertain-future)[14] Annual CASE investment by automakers-How automakers can master new mobility | McKinsey (https://www.mckinsey.com/industries/automotive-and-assembly/our-insights/down-but-not-out-how-automakers-can-create-value-in-an-uncertain-future)[15] Peter Thiel/ Zero to One: Notes on Startups, or How to Build the Futuresource: AEM (https://www.autoelectronics.co.kr)
23.08.21