Autonomous driving market overview

Introduction to autonomous driving

Autonomous driving is one of the most sought-after market in tech right now. Along other major changes in the automotive industry such as electric vehicles, connected cars, or ridesharing, autonomous driving is at the heart of what is considered to bethe second inflection point of mobility with a promise of a greener, safer, more convenient, and cheaper transportation.

Indeed, just like we turned from horses to cars about a 100 years ago, mobility is slowly turning from mechanical transportation machines to supercomputers on wheels; creating a new land of opportunities for outsiders to come in and for balances of power to shift drastically in a trillion dollar automotive industry.

“Autonomous driving is at the heart of what is considered the second inflection point of mobility.”

Since autonomous driving activities kicked off with the DARPA challenge in 2004, the ecosystem became a lot larger and fiercely competitive with OEMs and tier 1 suppliers now joined by internet companies, TELCOs, electronics manufacturers, and a large crowd of startups. A spurge of innovation and enthusiasm notably took the market by storm from 2013 to 2017 with expectations that autonomous cars would be widely adopted by 2020.

Billions of funding later, where is autonomous driving standing right now?

The promise of a better future

What’s a self-driving car? In simple words, a self-driving car is a car with the ability to perceive the outside environment and to make driving decisions upon it, and thus, drive by itself. It usually takes the form of a car retrofitted with a bunch of sensors (e.g. cameras, lidars, and radars) and powered by an embedded supercomputer trained to be a super driver (sort of substituting for the eyes and the brain of the driver).

What sounded like science fiction 20 years ago has never seemed so real today with autonomous cars quickly becoming the greatest hope to get rid of bad human drivers (e.g. responsible of 90% of accidents on the road), offer mobility to people with health problems and disabilities, reduce congestion in cities (e.g. traffic jams and parking), increase productivity for soon to be ex-drivers, have robots deliver packages to our doors, or even to bring many mobility companies in profitable territory (e.g. taxis and ride hailing companies).

“Humans spend on average thousands of hours in cars (including hundreds of hours stuck in traffic or looking for parking), while also being held responsible for 90% of road accidents.”

These perspectives for innovation drove tech companies to take the market by storm which in turn, thanks to majorbreakthroughs in artificial intelligence and machine learning, led to unprecedent progress and massive amounts of funding. The boom of the industry in the mid-2010s further drew over-optimistic predictions with a wide range of actors such as General Motors, Ford, Google’s Waymo, Toyota, Honda, and Tesla all promising us autonomous cars around 2020.

Yet, 2020 is here and self-driving cars aren’t.

Overcoming the data challenge

Autonomous driving is at the crossroad of many challenges which are likely to unfold as followed: technology, and then, in a lesser extent, regulation and adoption.

First and foremost, autonomous driving companies need to make sure that the tech is solid and that self-driving cars are as close as possible to being 100% safe, making a case at surpassing their human counterparts. In fact, autonomous cars being 99% safe would result in killing 1 every 100 pedestrians crossed and would amount to millions of deaths very quickly. Making self-driving cars work is extremely difficult and involves a set of very complex technologies such as computer vision, artificial intelligence, or machine learning for cars to learn how to drive from data accumulated on the road and that will be further transformed in hard-coded computer rules.

Once the tech is ready, governments and public institutions will need to lay the ground to introduce and regulate autonomous cars in our day to day transportation. That includes granting authorizations to autonomous driving companies to test and to drive in public areas as well as agreeing on a legislation defining rules and responsibilities (e.g. liability in accidents, proper data collection, sharing of the road for self-driving and regular cars).

Lastly, it will be important to provide user friendly commercial applications and to educate people about self-driving cars in order to enhance public adoption. Today, more than 50% of people still don’t feel comfortable about riding autonomous cars. Moreover, even though younger generations are more inclined to shared mobility services and new technologies, cars have long been meaningful possessions advocating for freedom and social status.

Each of these challenges will take some time to be addressed but what has truly been delaying commercial roll plans is the technical complexity of building the technological stack. Over the last 10 years, the industry has laid the ground to capture, store, and process billions of hours of driving data to teach cars how to behave in increasingly complicated scenarios, but self-driving cars now seem to be confronted to the limitations of big data.

Indeed, in order to learn from new scenarios, self-driving cars need to run into infrequent situations, or “edge cases”, which require an ever-larger number of miles to be driven and data to be accumulated. The trick in improving autonomous driving is that the closer we get to the 100% safety mark, the more infrequent the edge cases are, the more data we need to find them, and the exponentially more difficult progress becomes.

In 2017, Intel notably claimed that self-driving cars generated between 1TO and 5TO of data per hour per test vehicle, which for companies running fleets of tests vehicles all day long represents the equivalent of an ocean of data. Working through these massive amounts of data proved to be highly complex, time consuming, expensive, and unsustainable; forcing many companies to reconsider their go-to-market.

“Self-driving cars generate between 1TO and 5TO of data per hour per vehicle.”

Who’s leading the autonomous driving race?

Despite the bumps in the road, many companies are stepping up their efforts and betting that future gains will far outweigh the burdens of bringing the technology to market.

To help us track the race towards fully automated vehicles, the Society of Automotive Engineers (SAE) notably described autonomous driving in 6 levels of automation ranging from no automation to full automation.

At its core, the tech behind autonomous driving relies on capturing, processing, and deploying data that will power a self-driving software. The driving software is the holy grail of the industry and many tech companies leveraged their unfair advantage in software and data management.

 “Not surprisingly, the leader in autonomous driving today isn’t a car manufacturer but Waymo: a Google spinoff.”

Autonomous driving performance has been measured in a few (maybe debatable) metrics that became standards in the industry, one of which being the miles driven per disengagement reported at the Department of Motor Vehicles (DMV) in California (level 3 and above).

In 2019, surprisingly and for the first time since 2015, Waymo came second in miles per disengagement behind Baidu rising to the top with over 18 000 miles per disengagement. However, Waymo and GM Cruise largely dominated other metrics with 1.4 million miles driven by 148 test vehicles and 0.8 million miles driven by 228 test vehicles respectively (e.g. Baidu drove 0.1 million miles with 4 test vehicles).

It’s worth noting that if startups seem to be leading the way in autonomous driving performance, large corporations tend to be more secretive about testing while also filing more patents to protect intellectual property.

In terms of commercialization and state of the art, autonomous driving is still very limited with Tesla proposing the most sophisticated self-driving car on the market with level 2/3 automation and a few actors offering level 4 automation in specific use cases (e.g. parcel and goods delivery, residential transportation).

Hence, besides all the excitement about autonomous driving and all the big announcements about commercial plans rolling out anytime soon, we’re still a long way from level 5 and full automation in our day to day transportation.

“It could take up to a few decades before autonomous cars are widely adopted.”

Market structure & trends

By the end of 2019, we found a total of 937 companies and 2292 investors partially or closely related to autonomous driving.

To get a better idea of the balances of power geographically, we aggregated data of companies, investors, and funding for seven strategic zones: Northern America, Asia, Europe, Middle East, Oceania, South America, and Africa.Northern America, Asia, and Europe formed a distinct leading trio, with Northern America in a clear dominant position.

That dominance is easily explained by the United States representing almost half of the market (399 companies) with natural clusters forming in the Silicon Valley as well as in other states trying to lure the hottest startups with friendly regulations such as Pennsylvania, Arizona, and Nevada. China (90), Germany (82), Israel (64), the UK (49), France (38), and Japan (35) also proved to be competitive and innovative locations.

“The United States represents almost half of the whole market of autonomous driving.”

Considering the global landscape of innovation, having the US taking the lead on a burgeoning technology like autonomous driving isn’t surpising. What’s striking however, is how little funding ($6.5B) has been injected in Europe, which ranked second as a zone with 258 companies (28%) and is home to powerful automakers such as Volkswagen, PSA, Daimler, and BMW.

To put it in perspective, the funding throughout Europe is barely over twice as much as the funding in Israel alone ($2.9B).

Funding & notable exits

Funding in autonomous driving exploded in the past few years to reach about $150 billion today; and that’s without taking most R&D budgets of large corporations into consideration.

For instance, it has been estimated that Waymo spent an aggregated $1 billion before 2015 and is now spending more than $1 billion per year. Furthermore, with several corporations already publicly committing nearly another $150 billion for the next 5 years (Volkswagen $50B, Hyundai $35B, Samsung $22B), funding in autonomous driving could soon stack up to a staggering $300 billion.

“About $150 billion have been injected in autonomous driving from 2004 to 2020, a figure expected to double in the next five years.”

Among the most funded companies, we find some of the largest mobility companies like Uber, Didi Chuxing, or Tesla, but when it comes to companies focused solely on autonomous driving, Cruise Automation and Waymo reportedly were both funded above $5 billion while Argo AI ($3.6B), Sensetime ($2.4B), and Mobileye ($1.4B) are trailing behind.

In terms of investment volumes, incubators such as Plug and Play (37) and Y Combinator (18) funded the most deals while other top investors well represented the wide range of actors in the space with venture arms of large corporations (Intel, Hyundai, BMW, Daimler, Toyota, Tencent, Bosch), mobility VC funds (Maniv Mobility, Fontinalis Partners, Trucks Venture Capital), or even regular vanilla VC (Sequoia Capital).

With all the funding and the size of the actors at stake, autonomous driving companies are naturally set up for big exits. Even though it’s still early, some of the most notable ones were Mobileye purchased by Intel in 2017 for $15.3 billion, Cruise Automation by General Motors in 2016 for $1 billion, Otto by Uber in 2016 for $680 million, nuTonomy by Aptiv in 2017 for $450 million, Drive.ai by Apple in 2019 for $100 million, or VisLab by Ambarella in 2015 for $30 million.

Interpretations & projections

In 2019, we recorded 13 new companies (max: 133 in 2017), $36.7 billion in funding (max: $40.6B in 2017), 26 exits (max), and a Google trend score of 570 (max: 681 in 2018). We’re also estimating between 4000 and 5000 autonomous vehicles in the world (up from about 3200 in 2018).

It might seem tricky to interpret these numbers without a macro perspective to understand what’s going on in the space today, however.

Compiling data (base 100) of funding, companies created, autonomous vehicles, media attention, and exits over time, we can see how autonomous driving trends fit in the Gartner Hype Cycle.

1 - Interest around autonomous driving started with the Darpa Challenge in 2004 and took off with the creation of Waymo in 2009: everything is slowly picking up (technology trigger).

2 - The frenzy around autonomous driving came in with multiple actors succinctly joining the space from 2013 to 2017 and forming all the major autonomous driving companies today such as Cruise (2013), Zoox (2014), Argo AI (2016), or Aurora (2017) with the promise of bringing self-driving cars everywhere by 2020: companies created, funding, media, and exits explode (inflated expectations).

3 - All the excitement is brought to a halt when technological challenges came to light (e.g. in 2017 Intel revealed that an autonomous car generates up to 5TB of data per hour), bad press started accumulating (e.g. Uber accident in 2018); forcing commercial roll out plans to be pushed further back (e.g. GM back pedaling on 2019 launch) and the ecosystem to reconsider the state of the art (e.g. Waymo’s valuation cut by 40% by Morgan Stanley in 2019): funding and exits stagnates, media attention switches to negative, and companies created crumbles (trough of disillusionment).

4 & 5 Going forward - Despite the difficult challenges of autonomous driving such as the technology, the regulation, and the adoption; the space remained committed to innovation with announcements about injecting about $150 billion in the next 5 years.

However, the current COVID-19 pandemic is going to impact these projections, bringing activity and funding down in 2020 before it slowly recovers in the next 3-4 years.

What is likely to happen is that leading autonomous driving companies will turn back to R&D and focus on products instead of rushed commercial solutions, heavily accumulating R&D vehicles to take advantages of economies of scale. This will create an opportunity for innovative enablers to come in and improve the value chain, until the ecosystem reaches a plateau of productivity and can switch to commercial applications.

As a result, the main companies will absorb most of the funding to become bigger yet fewer, companies created will startincreasing slowly due to enablers penetrating the market, exits will stagnate in volume but will increase in size (with enablers representing most of them), autonomous vehicles will increase tenfold (multiplying by 3 or 5 in the next 5 years), and media will start positively picking up in volume again.

Summary & key findings

What really came out of this market study was:

• Autonomous driving is considered the second inflection point of mobility but has been delayed by the challenges of big data.
• The ecosystem is really large and competitive (actors + funding).
• In terms of performance, autonomous driving is led by newcomers and tech companies, but corporates are heavily committing to the industry.
• Overall, the United States is dominating the market and Europe is strongly underrepresented in terms of funding.
• The autonomous driving space has been steadily growing since 2004 before being brought to a halt in 2017 and has been in consolidation over the last 3 years.

Autonomous driving went through the typical ups and downs of the hype cycle with players staying committed to the space and to the perspectives for innovation.

The COVID-19 pandemic will further challenge the economic momentum of the industry, but it has also showed the relevance of autonomous transportation and deliveries. Furthermore, even though the automotive industry has been particularly affected by the pandemic, autonomous driving is also supported by cash rich tech companies (e.g. Google, Apple, Amazon) as well as the VC ecosystem and public institutions that will keep investing in innovation and mobility.

Hence, the market remains very attractive and relatively crowded even though already settled major players will benefit from a considerable advantage in navigating trouble waters, likely capturing most of the funding in the future.

There is a unique opportunity to contribute to the consolidation phase of the market however, by stepping in as an enabler to those large players who also represent great potential acquirers for exits. Moreover, there are sizeable opportunities in underrepresented areas such as Europe to capture some of the funding that is likely to pick up in the next few years.

Over the last 10 years, autonomous driving has experienced unprecedent progress with a first generation of innovation creating the leaders of the industry today. We believe Heex Technologies, just like many other enablers, will make up for the second generation of innovators that will allow production at scale of autonomous driving solutions, before a third generation rises and focuses on commercial applications.