Decoding the Autonomous Driving Landscape
Software will indeed eat the (automotive) world
Update: In response to many requests we are providing all charts in high-resolution for download. Feel free to use them for commercial and non-commercial purposes as long as you keep the reference to Firstmile as the author.
This July, the two motoring giants Ford and Volkswagen announced they would expand their global alliance on electric and self-driving vehicles in the US and Europe. Marking the beginning of a new era, past competitors now join forces and work together on highly strategic projects. The partnership has only been one of the many recent announcements in the field of autonomous vehicles, with OEMs scrambling to catch-up with new competitors such as Google’s self-driving vehicle unit Waymo and the now publicly listed Uber. Technology platforms are, it seems, becoming increasingly valuable and scientifically advanced, much to the discomfort of traditional vehicle manufacturers.
Meanwhile, the global automotive industry continues to pour billions of dollars into research and technology startups, with the urgency to find new people with the appropriate skills growing every day. According to research from BCG and the Michigan Mobility Institute, the US automotive industry alone will require around 45,000 mobility engineers and 70,000 skilled trades workers for the testing and rollout of autonomous vehicles in the next decade.
One thing we do know for sure today: The imminent wave of change in mobility will make the ownership of vehicles obsolete. It will disrupt all current business models: car insurance, financing, and fleet management and once ready, will be commercially available much quicker than we anticipate today (despite the recent too optimistic push for Level-5).
In order to grasp the magnitude and understand the blurring boundaries between vehicle manufacturers, traditional automotive suppliers, and technology companies, we focused on seven different clusters within the autonomous driving space. All 232 firms in the landscape are either currently or formerly VC / PE-backed. The data was compiled using Crunchbase, Pitchbook, Tracxn, and Firstmile internal sources.
Autonomous driving systems offer full-stack autonomous driving software and technologies for their customers. These may include advanced driver assistance systems (ADAS) which aim to augment certain driver capabilities, to technologies approaching complete autonomous driving systems.
Data and simulation firms offer data analysis, image annotation and clustering, training data and advanced simulation tools for customers looking to train their autonomous driving systems. These have to provide more than 99.99999% accuracy and reliability, hence accurate simulation data and annotation is more than crucial.
Light detection, ranging and sensing includes hardware and software to enable the analysis and comprehension of vehicle surroundings in real time. Included are radar, light detection and ranging (LIDAR) systems, ultrasound, cameras, chip and software technology startups.
Autonomous vehicle manufacturers build vehicles with autonomous driving capabilities as OEMs. While fully autonomous vehicles with the primary purpose of transporting passengers may still be a few years away, current products are becoming more entrenched in industrial practices. These are special purpose vehicles that range from street cleaning vehicles to last mile delivery pods.
Map and location-based services in their simplest form allow modern vehicles to locate themselves with pinpoint accuracy and even provide further data vital for autonomous mobility in the future (i.e. curb-side availability, live traffic and parking data). These services, relying on highly precise geolocating, enable vehicles to understand their surroundings, interpret, and deduce actions from the data received from hardware sources.
Apps and mobility infrastructure companies offer mobility services, provide software or physical infrastructure for autonomous vehicle testing purposes, and next generation mobility-as-a-service platforms (MaaS).
Connected car and V2X communication companies allow communication between devices, vehicles and their environment to enable a fully autonomous mobility ecosystem. In the future, 5G communication technology will play an imperative role.
Autonomous Vehicle Landscape
Top 20 Autonomous Vehicle Startups (by total funding volume)
50% of the top 20 autonomous vehicle startups are from America, while only 40% are from China. The remaining two are from Israel (Innoviz Technologies) and Austria (TTTech). Interestingly, 55% of the most funded startups offer either advanced driver assistance systems (ADAS) or develop full-stack autonomous driving software, highlighting the value of owning the entire autonomous software value chain. 30% of the startups are manufacturers of autonomous vehicles and the remaining 15% are active in the LIDAR segment, focussing on hardware components.
Notable Autonomous Driving Acquisitions
Intel’s acquisition of Mobileye in 2017 is perceived by many as an inflection point in the autonomous driving industry, as it proved to traditional automotive OEMs that the market was no longer solely based on physical vehicle components: Incumbents now have to stay alert. Other notable acquisitions include Ford’s acquisition of Argo AI and Journey Holding, GM’s investment into Cruise, Aptiv’s (formerly Delphi Technologies) purchase of nuTonomy and ottomatika, and Apple’s recent acqui-hire of Drive.ai, as it struggles to gain a foothold in the depths of autonomous driving.
Notable Autonomous Driving Partnerships
At first glance the sheer volume of partnerships and alliances can be quite overwhelming. When looking more closely however, we deducted three key trends:
1. A new wave of next generation, software focussed automotive supply firms is emerging
In the year 1977 the GM Oldsmobile Toronado was the first ever vehicle to feature an electronic control unit to manage spark timing. By 1981, GM had deployed around 50,000 lines of code across their passenger car line and today, the Chevrolet Volt, for example, runs on more than 10 million lines of code. Hardware suppliers and automotive OEMs have painstakingly developed their software skills over time, but remain hardware manufacturers at heart.
Now, the number of players with purely software and technology-based backgrounds in the automotive industry is growing rapidly. It is no longer sufficient to produce the best drivetrain or motor vehicle. It is becoming more and more evident that software will be king in the future — and has to be developed simultaneously and not in siloed business units.
2. The magnitude of technical challenges creates new alliances among large OEMs and technology firms, while only a few select firms choose to develop technology in-house
After experiencing some of the challenges imposed by developing largely complicated autonomous driving technology, large OEMs have shifted their siloed approach to a more open one: Forming partnerships with established technology firms to boost their expertise and gain access to state-of-the-art knowledge. Currently a growing domain, recruiting experts from own ranks is proving to be more difficult than anticipated: OEMs need to partner with and acquire outside knowledge. While BMW still relies on in-house development for example, Volkswagen has jumped ahead by partnering up with Argo AI and Ford.
Technology giants (ie. Microsoft, NVIDIA, and IBM) are also joining the battle, which may leave OEMs behind. Tesla and Apple on the other hand seem to be dodging partnerships actively as they head into the storm without a broad network of automotive or technology allies: They firmly believe in in-house development.
3. New players are emerging in the US and China, Europe lagging behind
Along with the recent announcement that five automakers are joining the self-driving technology joint venture backed by Toyota and SoftBank (“MONET Technologies”), Asian startups are leveraging their technological expertise to make rapid advancements in self-driving vehicle technology. Baidu’s autonomous driving platform Apollo is the perfect example of how Asian technology companies have grasped the value of wide partnerships before anyone else: A glimpse at their partnership page has the potential to make German automotive executives severely dizzy.
However, even though German automotive firms have remained rather silent regarding their autonomous technology, a closer look behind the curtain reveals an extensive network of partnerships, alliances and joint ventures looking to shape the future of autonomous mobility. The most important ones are detailed below.
While the drive towards electrification plays a dominant role in Volkswagen’s new mobility strategy, it has a broad positioning in the field of autonomous driving with a diverse set of partners. Two key alliances form the foundation of its recently stronger foothold: The Partners for Automated Vehicles Education (“PAVE”) alliance with other global OEMs and the Networking for Autonomous Vehicles (“NAV”) alliance for the interconnectivity of vehicles.
This July, VW also decided to cut the cords with Aurora, a leading technology player in developing self-driving vehicle technology, and joined Ford by investing $2.6B into Argo AI as part of a broader alliance for autonomous and electric vehicles. This enables Argo AI to step up their game in the global race for developing pioneering autonomous technology. Specifically, their goal is to develop a Society of Automotive Engineers (“SAE”) Level-4 capable self-driving system and dramatically boost its staff by 40%.
Finally, VW displays clear strategic intent through partnerships with Asian players to strengthen their international presence, partnerships with tech companies to give them wider access to new technology innovations and the partnerships with Microsoft and Baidu to boost their automotive cloud.
BMW and Daimler
The recently announced partnership between two German premium OEMs came as a surprise to many, with internal technology development clearly a priority beforehand. Even combined both players seem to have much less autonomous driving partnerships than VW.
What does stand out however, are BMW’s partnerships with cloud and technology specialists (i.e. with Microsoft, Tencent, and IBM) and Daimler’s efforts to expand its network to Asia with Geely, as well as integrate self-driving truck technology through TORC Robotics and Transdev. Particularly notable is the partnership between Daimler, Bosch and NVIDIA to develop autonomous Level-4 and Level-5 driving technology, starting with an automated valet parking test run.
While the level of knowledge-sharing in the unnamed autonomous driving platform driven by BMW compared to PAVE or the NAV alliance may be difficult to assess externally, the depth and level of the platform seems to be more shallow. It does however, include a variety of traditional automotive suppliers, technology players and autonomous vehicle startups to drive the development process forwards, including Intel, Aptiv, Innoviz Technologies, TTTech and Baidu’s Apollo platform.
MONET Technologies, which was formed just weeks ago, lays the groundwork for the sturdy positioning Toyota has built up through various co-investments into AV technology and with strategic partnerships across the globe. Together with SoftBank, Suzuki, Isuzu, Subaru, Mazda, and Daihatsu the joint venture combines data and know-how from all members to create a dominant player in the Asian market.
The clear geographic focus allows Toyota to explore the realms of autonomous driving technology with less regulatory boundaries, which many of the competitors in the US and Europe face. Meanwhile, they are in close proximity to leading cloud infrastructure and technology firms with similar interests of leading the race to developing fully autonomous vehicle technology.
Other notable players are Tesla, Apple, Waymo (Alphabet) and Uber. All four have strong backgrounds in technology with a clear strategic objective to develop autonomous driving technology in-house, with no direct need for a broad network of partnerships.
Apple sends mixed signals with its efforts in autonomous driving, hiring former Tesla Engineering VP Steve MacManus in July after two other chief engineers, while terminating 190 employees in February. Recent commentary has highlighted the unfavourable working conditions and high attrition of employees, stating the difficulties in advancing autonomous driving technology at Apple as the primary culprit. The acqui-hiring of Drive.ai in July for around $77M now offers Apple the opportunity to make use of seasoned engineers to refine their self-driving car efforts.
As usual, Tesla is especially radical in its public appearance and opinion on autonomous driving technology. First, LIDAR technology was brutally shunned by Tesla founder and CEO Elon Musk. Second, partnerships do not seem to be a high priority, as displayed by few publicly known partnerships. Tesla maintains an almost siloed position within the automotive industry: a pioneer, but closed-off.
Waymo and Uber
Waymo and Uber both develop their technology and software in-house, partnering only with firms able and willing to provide them with the required resources. They visibly have a limited number of technology partners with their view and ambition to develop the full-stack autonomous driving technology internally. The primary purpose of partnerships are to test and improve the self-driving vehicle technology on cars, while fine tuning the customer value-proposition. A recent call for support was exerted by Lyft, opening their Level 5 data set (containing 55,000 3D frames) to the public.
Surprisingly, despite the close proximity of Tesla, Waymo and Uber headquarters, there are no public partnerships or alliances between these key industry players. Waymo does, however work closely with DeepMind, a fellow Alphabet company, to develop new AI training models.
Autonomous Driving Company Locations
European automotive players are well known for their manufacturing ability and have built up a strong reputation as both OEMs and suppliers. Notable companies include Continental, Bosch, and Siemens. This focus and perhaps lagging ability to adapt to the fast moving industry is evident when examining the few large pure play technology players in the European market.
America seems much better suited for the impending wave of technological change in the automotive industry: Amazon, Apple, Microsoft are large technology players slowly dipping into the deep waters of autonomous driving technology, while Aurora, Aptiv, Mobileye, and Waymo already have a strong standing. Ford, GM, and Tesla are similarly at the top of the autonomous driving value chain.
Asia offers a strong nurturing ground for technology firms, hence home to many leading mobility firms and market challengers (i.e. Didi and Baidu’s Apollo Platform). The MONET Technologies joint venture underlines the high level of ambition present.
Autonomous Driving Patent Applications
Painting a different picture than the spread of autonomous players across the globe, the top patent applications show registered patents through the World Intellectual Property Organisation (WIPO). In Europe, the number of patent applications in autonomous driving has grown 20 times faster than other technologies in recent years, increasing by 330% compared to 16% across all technologies in the same period. While all firms in the autonomous vehicle landscape clearly share a similar vision and see the clear need to drive AV tech, the statistics for top filers for autonomous driving patents should, however, be interpreted with caution.
Overall, German firms have a strong presence when examining autonomous driving patents. This may be for two reasons. First, software components are only registered with the patent office under very specific conditions, some of which may not directly apply for future-oriented autonomous driving technologies. Second, Bosch, Audi, and Continental have a history of registering many different variants of hardware patents. Meanwhile pure software players Waymo and Argo AI are not even among the top 10. Hereby the accuracy of interpreting the limited data available and extrapolating the true competitive positioning of automotive firms with registered patents may be limited.
Autonomous Miles Driven per Disengagement
Miles per disengagement is an indicator of technological advancement for autonomous driving companies. The California Department of Motor Vehicles defines a disengagement as a “deactivation of the autonomous mode when a failure of the autonomous technology is detected or when the test driver disengages the autonomous mode.”
There is a lot of disagreement on the level of accuracy of the metric as it may be too vague, is valid only for California, allows companies to avoid reporting certain road events and may not reflect the difficulty of the chosen driving environment. Nevertheless, it is a metric which can be used to show that there is a large gap advantage for American technology companies with Waymo, GM (Cruise) and ZOOX able to cover 11,000, 5,200 and 1,900 autonomous miles driven before a disengagement.
For context however, Waymo (first) has driven 40x as many miles as ZOOX (third) in 2018. Not only is it therefore difficult to further interpret the level of progress made by each of the companies based on the miles driven, but also as some disengagements may be deliberate to test the reaction of the software to new situations. Without the appropriate context and exact routes driven relating to the data, the numbers show the significant progress and are an indication of the technological advancements made within the industry, but are not sufficient for a full comparison between the different players.
Implications for the Future
As we have shown all players, from automotive OEMs to technology challengers, are now on track to chase fully autonomous driving technology. Today, we can deduct no foreseeable frontrunner or laggard, as well as no precise date by when fully autonomous driving will be commercially available. In our opinion, however, it is not important when this exactly will take place, but how incumbent firms and new startups emerge, develop and evolve in the coming years. It is undoubtedly one of the most exciting periods in the history of mobility.
Fully autonomous driving will come and once ready, will arrive in the blink of an eye, ultimately turning mobility into a commodity. As in the traditional automotive landscape today, there will be no winner-takes-it-all, as seen with the social media giant Facebook. Regional differences in culture and regulation will allow different players to dominate local markets. Software however, will indeed eat this (automotive) world.
Firstmile is specialised in mobility & automotive tech investments. Our mission is to back the next generation of European mobility tech companies with capital, deep domain expertise and a strong industry network. We are part of the Atlantic Labs Group, one of the leading early stage investment firms in Germany.
This research was conducted and written by Markus Ferres and Jens-Philipp Klein. If you have any feedback regarding this article or if you would like to exchange ideas around mobility tech in general, we would love to hear from you.