Electric Vehicles, the Automotive Disruption that We Have Been Waiting For?
Cars haven’t changed much in the past few decades, but we might see a lot of changes to cars in the next few decades. Electricity is one and software is another. Those two will be the primary factors in changing what a car is, how it gets made and who might own one. They might also change the competitive landscape and shake the supply chain from top to bottom. The question then turns to: are these changes disruptive enough to turn the entire industry upside-down? Could the innovation be innovative enough to launch an iPhone-killing-feature-phone type of attack on the automotive industry? Right now, Tesla is leading the charge. But where would Tesla lead us to? Is Tesla the iPod maker that tried to bridge the gap between CD player and smartphone or the iPhone maker that kicked Nokia off of pole?
A lot of people are dismissing Tesla. Elon Musk himself has been calling out Wall Street Journal among others. When automotive veterans/insiders talk about Tesla, they see a not-that-well-made car with some cool features that are also on their roadmaps (but maybe several years away, if not dozens). The volume that Tesla is producing is merely a small fraction compared to major automakers…and they are building them in tents!
That’s exactly what happened when Nokia people saw the first iPhone. They saw a not-that-well-made phone with some cool features that are also on their roadmaps (touch panel? checked. smarter OS? delivered!) and the volume of 1st-generation iPhone are so puny that it eventually became a collector item worth auction 10 years later. Even worse, iPhone wasn’t even on 3G! (Remember the famous joke? iPhone 3G doesn’t have 3G).
We know how that story ended. Would the history repeat itself for the automotive mainstay? Let’s get a few things straight first.
Way back in 2009, the advancement of powertrain technology, especially diesel, had reached a new level in efficiency that dissuaded automotive OEMs from continuing their hybrid electric-vehicles (HEV) projects and threatened to kill off those electric vehicles (EV) dream. At that time, no one has a clear idea how to expand the range of travel of battery-powered vehicles to compete against internal combustion engine (ICE) powertrains. Another funny thing is, after the financial crisis, analysts widely believed that thinner wallet and stricter regulations for emission control and fuel efficiency will seriously hurt the demand of SUVs. Eventually, automakers faked diesel performance to sell more SUVs than ever before eventually got busted. At this point, HEV probably won’t be able to mount a comeback as EV has positioned itself as the direction of future on most automakers’ roadmaps. So the EV trend probably isn’t going anywhere, and diesel isn’t the savior that ICE has been expecting. In time, EV will be widely commercialized (in addition to Tesla, Chevy Bolt and Nissan Leaf).
With more EVs in the pipeline, the integration of the electric-driven train (opposite to ICE powertrain) components will become increasingly critical. There’s an old car industry joke: you can see the organization chart of a car company in the dashboard. The steering wheel doesn’t mesh with the gear stick. The ABS has nothing to do with the blind spot detection. Systems are built separately and independently by different suppliers (mostly Tier 1 suppliers). Being a fragmented industry, it’s unlikely that automakers can start producing everything in-house anytime soon. However, accelerating digitalization in automotive manufacturing could be a theme to stay. The advances in AI, analytics, IoT and 3D printing could further transform the manufacturing process.
Most analysts agree that autonomous driving Level 3 will be available in the next 10 years, and the vehicles with Level 3 autonomous driving hardware capability, both computing and sensing, will be widely produced starting from 2020. New connectivity technologies (including 5G) will further accelerate the adoption of autonomous driving and advanced ADAS. The annual production could reach 6 million units by 2022, creating sizable demand for semiconductor chips including processors, sensors and connectivity chips. To integrate all these chips, most cars might need a more powerful CPU and a corresponding OS. Right now, car industry calls its millions-of-line-of-code firmware as software, but those so-called software has a very long way to becoming an OS. The present paradox of complex cars with simple software will have to evolve into simple cars with complex software to accommodate the higher level of autonomous safety and driving functions. Since the hardware might be implemented earlier than the actual deployment of autonomous driving Level 3 for data collection and certification purposes, OTA (over the air) update to enable certain functions might be crucial for next-generation vehicles.
From my observation, there are 4 key trends in this iteration of disruption within automotive industry: EV/battery for stricter law and better earth, digitalized manufacturing for better integration, more software and semiconductor to enable higher autonomy and safety features, and the better user experience (UX) by meshing everything mentioned above together.
So how does Tesla fare in those 4 aspects? Could it turn its innovation into disruption
EV and Batteries:
Tesla is well positioned in this regard. With its own battery factories and the partnership with Panasonic, Tesla could own 15% of worldwide battery capacity within a decade. If Tesla restricts its battery supply to itself (which is fairly common in auto industry), Tesla would have a 15% base for EV market…which would fit perfectly in today’s fragmented auto market (no one really exceeds 25% of market share). Tesla might lead the industry by several years in implementation and engineering, but battery is similar to semiconductor industry. No one can really dominate the market with totally in-hosue IPs and patent-protected equipment. For example, Apple might be the pioneer in touch panel and probably still lead every phone maker in that regard. but the lead won’t prevent others from giving up using touch panel. In other words, unless you can dominate the entire supply chain (like 3D camera in iPhone, at least currently), there’s no “winner-take-all” effect in this slim lead. Consumers won’t ditch all other options just because your touch/battery tech is slightly better than others.
One of the opportunities for Tesla to completely pull away in this regard would be if Tesla can achieve (and perhaps..own) the Lithium-Air batteries technology. Lithium-air batteries is made by combining a lithium cathode with an oxygen anode, which in theory would have 10 times the energy density of today’s lithium ion batteries. The lack of breakthrough since the introduction of lithium ion battery has been the bottleneck of all electronic development (including smartphone). However, lithium-air battery is still at open-cell stage to allow the inflow and outflow of oxygen, resulting in safety concerns. The physical limitation of the flow of air/oxygen also cap the efficiency of these batteries. Most of the recent lab-scale breakthroughs have been coming from universities in the United States, which could play as a Tesla’s advantage as trade war looms. However, it could easily take more than 10 years to success.
The lead in EV train and battery could help Tesla to maintain a higher price tag like Apple with iPhone’s components, but it would unlikely to give Tesla a total dominance like Intel in CPU or Microsoft in Windows. Not to mention the lead in battery could very well take Tesla to Sony’s position: the clear leader in camera sensor but clearly not a legitimate competitor in smartphone arena. Judging from the fact that Tesla has only begun to make profit from its lineup, short of achieving Intel or Microsoft status could be a devastating blow to its cash-burning pace.
An X factor for EV is the regulations. Vehicles running on new energy, such as electric vehicles (EVs), historically have not sold well at times when oil is priced low. But now, they sell well even in a situation of low oil prices, mainly due to the recent government regulations and incentives. For the past 2 years, oil price has even climbed a bit to fuel more EV demand. Most analysts believe that oil prices will start to impact EV adoption when they constantly surpass $80 per barrel, and if oil prices were to increase significantly (to $100 per barrel or more), adoption rate would increase significantly.
In 2020, three new emission standards will come into force. The “Euro 6d” standard will start in the European Union, the “Bharat Stage VI” standard will start in India, and the “China 6a” standard will start in China. The impact from the introduction of these standards will be significant, because the number of cars sold in these areas accounts for a large proportion of global automobile sales. More electronic powertrain technologies, such as idle start-stop and the 48-volt mild hybrid electric driving system, will widely be used to reduce the total emissions for each automaker.
There are also regulations that will impact the choice of powertrain type for automakers. Zero-emission-vehicle regulations in California expand to middle-size automakers from 2018, encouraging automakers to produce more EVs and PHEVs. Regulations on carbon dioxide emissions in Europe will also accelerate the growth of the PHEV market from 2020. The Chinese government targets the sales of new-energy vehicles at 10% from 2019, encouraging automakers to produce more EVs.
With all these regulations kicking into gears, the EV adoption pace might be accelerated, and thus leaving only a handful (if not one or two) of automakers to consumers. With its current lead, Tesla would be primed to excel…if they can solve the next problem:
Digitalized manufacturing isn’t easy. Just ask Musk. Tesla’s Model 3 began production in July 2017, but the advanced assembly line was never fully ready. As a result, Tesla’s factory workers had to piece together parts of the cars in a special area (tent!).
Analysts at UBS found that the first few batches of Model 3 performed poorly in terms of finishing quality, sporting missing bolts and uneven welding, compared with GM’s Chevy Bolt or BMW’s i-Series. “As of April, the standard deviation of all gaps and offsets across the entire car had already improved by nearly 40% on average, with particular gap improvements visible in the area of the trunk, rear lamps and rear quarter panel,” a Tesla spokesperson said in a statement.
All in all, it’s safe to say Tesla certainly isn’t enjoying much of an advantage from its high-tech assembly line. Without evidence to prove that traditional assembly line cannot churn out EVs with high quality, Tesla’s plan remains an experiment. As of now, the poor building quality is one of the biggest concerns of Tesla’s potential consumers.
And make no mistake, other car makers’ assembly lines is far from primitive. Lean production, complex planning, high degree of automation and detailed tracking have all been characteristics of modern auto plants. The only major hindrance that prevent more upgrade is the complicated political factors shadowing the auto industry. More than any other heavy industry, auto industry has long been coined as the single greatest engine of economic growth in the world. From the steel and aluminum in the backbone to the service of sales and repair, there are more than 20 million people working directly in making the vehicles and parts that go into vehicles. That is over 5% of the world’s total manufacturing employment. It is also estimated that each direct auto job, which pays decently with high degree of stability, supports 5 indirect jobs in the community. Each vehicle sold will also create a nearly 20% ratio of jobs related to off-market manufacturing, maintenance and sales. With so many decent/well-paid jobs at stake, many counties have laws forcing auto makers to build plant locally (and more extreme: China and its JV requirement). Once the plants have been established, governments usually use environment requirements and energy restriction to slow the pace of implementing automation. More automation usually means fewer jobs. Worse yet, while automation could create some jobs with great pays, the jobs they replaced are generally people with fewer skills and thus less likely to find a job with similar pay. Leaving large amount of low-skill workers has always been a political bomb, and therefore, for Tesla to expand beyond the States, finding a correct balance between creating all sorts of jobs and maintaining its efficiency would be the deciding factor.
Another obstacles could be the supply of semiconductor components. Regulatory authorities globally require semiconductor manufacturers to achieve certification, which proves they fully meet strict safety and quality standards. Safety-critical automotive chips require compliance to ISO 26262, which defines automotive safety integrity levels ranging from ASIL A, with the fewest safety requirements, to ASIL D, with the highest safety requirements. ISO 26262 has been constantly upgraded with stricter requirements since it was first established. New features in autonomous driving, drivetrains and safety systems have put additional reliance on semiconductor chips, which now need safety verification documents for applications in automotive systems. Automotive OEMs are demanding semiconductor chip vendors comply with the most stringent regulations for chips not even meant for safety applications, which has created a bottleneck in supply chain.
Semiconductor chip makers have long been well aware of ISO 26262 and have deployed automotive qualified processes. They have taken additional steps in ensuring their design and process technology can deliver products in accordance with the required and relevant regulations. But compared to smartphones or even PCs, the quantity of vehicles is relatively small and the price that automakers are willing to pay cannot justify the astronomical investment needed to comply with all these automotive requirements. Improving the reliability for semiconductor chips up to 20-year functionality isn’t easy. Chip makers will need to adopt new approach of data analytics during chip manufacturing to ensure those levels. Semiconductor manufacturing and design constitute complex processes that generate huge amounts of data from each step in the process, and the data has a field of view of several months. The vast amount of complicated data collected during chip manufacturing and design will require sophisticated data processing, well beyond most other processes.
Software and User Experience:
So far, the most discussed Tesla’s disruption on “user experience” has been its software. The key to Tesla’s ability to run software, instead of firmware, is the OTA function. Larger, more complex software are prone to errors, and these errors would need to be patched. Ability to patch relies on OTA. Without OTA, patching software could be extremely costly.
OTA (Over the Air) software updates refer to methods of using wireless communications technologies to distribute new software or firmware updates and/or configuration settings to automobiles’ various computing systems, including infotainment systems and electronic control units. OTA software updates typically leverage dedicated server locations to send and manage the updates to all automobiles in a given region, or by model and year.
Before Tesla, OTA updates have been uncommon in auto industry. Automakers are planning to adopt the technology, but have differing approaches, including using cellular networks and home Wi-Fi. Unfortunately, none of those approaches is reliable enough to ensure nearly everyone has their OTA updates. Tesla deploys a continuous deployment method, including regular firmware updates to operational controls and safety-critical functions, by necessitating wireless connectivity with some crucial features (route planning with charger information, etc.) and bypassing dealership to offer direct service (which we will talk more later). OTA benefited from the increasingly wider adoption of 4G connectivity, and 5G connectivity or even satellite should further bolster it.
With more chips and modules than ever, OTA offers the ability to repair and upgrade vehicles at a much lower cost and much higher completion rate. From the past experience of parts recall, completion rate has long been an obstacle in fixing defects. OTA also creates an aftermarket opportunity to maintain vehicles over a long period of time. Currently, Tesla holds the advantage of the ability to have customers enable connections to vehicles as well as establishing a trustworthy data security and software validation process. Most automakers are not set up to do regular deployment of new software on a large number of different models due to resistance from dealers as OTA updates represent a potential loss of their service business.
This leads us to the unnoticed breakthrough from Tesla: direct sales. With direct sales, Tesla sells its cars directly to its customers on a fixed price instead of going through dealers. This may not sound much, but it could be the biggest barrier for other automakers to follow its path. Automakers often have contracts around who can install new software and most car dealers make most of their profits from repairs. Something around half of repair spending is on things directly links to the ICE. No ICE means no oil leaks or broken fan belts. The original motivation for Tesla to bypass dealers could simply have come from no dealer was willing to cooperate with Tesla. However, bypassing dealers has offered a often overseen benefit: dealers usually play an important role in setting pricing and incentives, and driving demand to specific models, and most of these information is far from transparent. Customers have hated the back and forth of price negotiation. Cutting out the middleman enables OTA, lower repair cost, and allow the buyers to know that price. This user experience would be hard for the incumbent industry to adapt to.
For other automakers to counter this, they probably have to roll the data connection price into the vehicle purchase price to ensure fleetwide connectivity. In addition, they must invest in a new system for continuous deployment of software that puts an emphasis on continuous deployment rather than very long deployment cycles. Automakers probably have to sit down and co-develop a standard on cybersecurity. They also likely have to go further and allow full software operation independent of a network connection (as they didn’t earn that first-mover advantage and have to compensate for their legacies).
The use of OTA software updates in the automotive industry can reduce service and maintenance, product and warranty costs, including reducing the need for costly vehicle recalls. It also improves product differentiation and consumer satisfaction during the vehicle ownership phase, and provides new revenue opportunities — for example, adding new applications and enhancing vehicle capabilities and performance. However, this feature probably won’t be the determining factor. It’s something that help increase customer loyalty/stickiness, but consumers won’t be buying a Tesla just because there’s OTA. Tesla’s share or business model likely won’t collapse either if some other traditional automakers suddenly start offering OTA. The direct sales part could be potentially ground-shaking as it offers some long-term disruption. But unless Tesla can solved the “production hell” mentioned above, it won’t be able to build the scale to offer its cars to the mass public. After all, most of the entry barrier that Tesla has built on user experience is only held by the lack of willingness to burn more capital by other automakers. Once the motivation is big enough, I’m sure at least some of them will turn and throw in more capital to compete against Tesla.
Mobile device’s integration into vehicles also posts a threat to Tesla’s advantage. So far, these connections are fragmented and take several forms. SmartDeviceLink, a standard developed by Ford, has been made into an open standard, which now has been adopted by Toyota. MirrorLink, offered by the Car Connectivity Consortium, is used by a number of automakers, and projects applications from a mobile phone onto a screen. However, the two fastest growing systems are Apple’s CarPlay and Google’s Android Auto. They both have significant advantages, because the systems display a familiar interface to a mobile phone and include integration of Google’s and Apple’s respective virtual voice assistants. They enable the native Google or Apple maps to be displayed on a vehicle’s screen. These maps can use the cloud-based voice assistant to enter destinations or do searches, a significant improvement over embedded navigation systems.
However, CarPlay, Android Auto and MirrorLink generally do not have complete access to vehicle controls, and the mirroring technology is imperfect and can be frustrating. So far, automotive companies have mostly resisted seeking in-depth collaboration with consumer device manufacturers and technology companies. Unless automakers start investigating operating systems such as Automotive Grade Linux and Android and better integrate them with automotive system, Tesla should be able to keep an edge here.
Autonomous Driving and Share Economy
Electrifying vehicles is compelling but will probably be a commodity eventually. Tesla’s improvements on top of electrifying may not be commodities but are not necessarily decisive. If realized, autonomous driving and shared vehicles would change the world in fundamentally profound ways. Tesla is doing this…but it has not mastered it.
Autonomous driving and shared economy have been linked together. While most automakers have rejected these ideas or slow in development, conglomerates at Silicon Valley, and honestly, all the technology companies around the globe, have been racing to this goal. To reach either goal, data is the most crucial element. In this competition to more data, Tesla does hold an advantage in that it can collect data from its cars. However, the reason why autonomous driving went from pipe dream to a distant possibility is the emergence of machine learning (ML) and the subsequent application in artificial intelligence (AI). All those breakthroughs in ML and AI mostly happened in the past 5 years or so, and Tesla hasn’t been a major contributor, to put it kindly. Another question to ask is, we all know developing ML and AI requires a lot of data, and with its relatively tiny scale, could Tesla generate enough data from its cars in time to reach the goal earlier than others?
In getting enough data, Tesla’s approach has been to put as many sensors as possible into its cars, and collect as much data as possible from these sensors. It can achieve this because its cars are already built on a software platform and sensors can be added as the whole system is much better integrated. Furthermore, with OTA function, Tesla can push out more service related to autonomous driving (or closer to) through OTA to award its customers for providing the data. Unfortunately, Tesla would need a lot more new components in mass production to drive down the cost and deploy it to all vehicles. With wire harnesses now weigh 20 kilograms or more and are eroding the space for passengers and luggage, shift to high-speed Ethernet within the car is imminent. Other wired and wireless connectivity solutions, such as Clock Extension Peripheral Interface, Bluetooth 5 and SmartMesh, etc. would need to come through before 2025, too. Above all, computer vision is still not capable of fully autonomous driving. All the data that Tesla acquired might turn moot if it’s not useful to next-generation technology. Meanwhile, even if you do have the hardware that can create an accurate 3D model of the world around the car and power a capable computer vision for autonomous driving, the rest of the autonomous puzzle isn’t working yet for anyone, nor does anyone in the field think that this is close. Bits of it work quite well (for example, cruise control on highways), but the whole does not. Hence, it comes back to whether Tesla is a capable enough software company. A lot of these autonomous problems are software-related. Moreover, without investing heavily on LiDAR like many others (for example, Tesla’s biggest competitor in autonomous driving, Waymo), Tesla would actually fall behind if its camera-based solution failed to work.
Another disruption comes with shared economy. While car sharing services will likely not disruptively change consumers’ car purchasing behavior before 2030, the car sharing service market is definitely growing strongly. So far, these services have not drastically changed the consumers’ car purchasing behavior, but internet-based car sharing services, such as Uber, are considered as an alternative to existing taxis. While people still think they should buy and possess cars privately rather than share them as is a symbol of providing private space for them and their families, the increasing cost of EV and autonomous driving could force car owners to share their cars with others. In all likelihood, the logistics won’t become practical until Level 4 and Level 5 autonomous vehicles become reality. 2030 is more than a decade from now, and many things could happen within a decade. Banking on a decade-away advantage could be too risky for a private company to bet on.
In my opinion, combining autonomous vehicles, electrification and shared mobility, Tesla would hold a hard-to-beat advantage. The raising expectations for cars to have a much higher average daily use when shared economy expands with autonomous driving would also turn the more advanced battery tech into another leverage.
But the fascinating thing about Tesla is that there are so many different things going on at the same. It’s safe to say that Tesla is the only company with so many different kinds of innovations in production. The history of technology industry for the past 20 years have taught us that building lovely product or being the first to some cool feature isn’t enough. For a company to thrive, create and dominate, it has to think about how all these innovations and breakthroughs fit into a broader system. Tesla has drawn up a blueprint of this bright, broader system, but the road to realizing this roadmap remains unclear. That’s probably why the rumors of Apple buying Tesla or Waymo working with Tesla are so juicy.