In recent years, there has been growing excitement around autonomous vehicles. Some of the world’s most visible, and valuable, technology companies have been making very public efforts to establish themselves as key players in this field, positioning themselves as the frontrunners bringing the potential of this technology to the public.
However, is the buzz around these companies justified? Earlier this year we turned our attention to Silicon Valley -where we met with over 25 companies, ranging from producers of highly accurate, computer-readable maps, developers of 360 degree sensor technologies and the regulators of this technology. Our insights may surprise you.
Contrary to popular perception, we believe that that the expansion and mass adoption of vehicular automation technology will not be driven by companies such as Tesla, Uber and Google, or any of the famous car brands, but from the innovators more comfortable among consumer electronics.
Why tech companies won’t lead the revolution
Google, with its comprehensive mapping tools was initially considered to have an advantage over the established industry.
However, the view within Silicon Valley is that long-term it won’t be mapping that steers autonomous cars through traffic, supported by basic sensors to avoid obstructions; rather, it will be much more sophisticated 360-degree sensors that will guide vehicles, with mapping information relegated to a back-up. This will free cars from having to be ‘on the grid’, allowing them to function offline and ensuring they won’t be led astray by out-of-date or inaccurate maps.
The technology for low-level automation already exists - it is inexpensive and will bring about most of the safety benefits needed for fully autonomous driving. Computing power, data storage and cloud computing are no longer the impediments they once were. Sensor technology can be integrated relatively easily into existing car designs using established suppliers such as Delphi. It’s easier to build a chip into a car than build the car around electronics, making it challenging for newcomers like Google, Apple or Uber to gain a dominant position.
Sensor technology is rapidly evolving & costs are lowering
For lower-levels of automation, vision sensors that use cameras can do most of the work. However, high level automation will require additional, complementary sensors. We met a number of new private companies who have developed cutting-edge laser-based sensors. While this technology is still in its infancy, it will be interesting to see which of these companies benefit in the automation race.
Meanwhile, the cost of existing sensor technologies are lowering. As an illustration, Delphi’s first adaptive cruise control system using radar (instead of laser), developed with Jaguar and launched in 1999, sold for $5000, compared to less than $100 now for a similar system today. This means that ‘off the shelf’ automation solutions will see companies such as Delphi able to scale their offerings and push automated solutions forward
An arms race in mapping
Although high-density maps may not be the driver of the autonomous roll out, they will be required for higher-level automation. The mapping companies we spoke to conveyed a real sense of urgency in their crowd-sourcing and data-gathering arms race to map all major geographies in minute detail. This data will be very valuable - a big prize for whoever gets there first. We anticipate that alliances with car manufacturers and suppliers (like Delphi) will be plentiful in the coming years.
So where should investors be looking?
There are a number of key areas that will be crucial to the advancement of autonomous vehicles, and our in-depth research shows there are some serious players who are making inroads through rapid R&D, and building of relationships with existing car manufacturers, particularly for lower-level automation.
Our view is that the best opportunities for investors won’t necessarily be the manufacturers of the electric cars themselves, but rather in those companies who can supply the semi-autonomous driving systems and all the individual components required to make it work - from sensors, mapping and system integrators - even down to the semi-conductors powering this automation.
