Q&A with Tony Cannestra, Director of Corporate Ventures at DENSO
Updated: Dec 17, 2018
This is an excerpt from ‘Corporate Venturing and the Future of Mobility, Automotive and Travel’, a new report from Global Corporate Venturing and LEIF, which is sponsored by BP Ventures, Denso International America, and JetBlue Technology Ventures.
If you’re an investor or a start-up in the mobility, automotive and travel industry, and have news and views of interest to the corporate venturing world, please contact me firstname.lastname@example.org. But please, no PR people. I mean no offense. It’s just that I want to hear from the engineers, not the oily rags – so to speak. (And I’ve done some oily rag work myself, by the way).
Denso International America and the fast-closing window on auto-tech venturing
Tom Whitehouse spoke to Tony Cannestra, Director of Corporate Ventures at Denso Corporation, a global automotive parts and components manufacturer based in Japan. Tony leads its corporate venture practice out of Denso’s office in San Jose, California. Tony is also among the speakers at this week’s Global Corporate Venturing and Innovation Summit in Monterey, where i will be moderating sessions.
Please describe the auto-tech venturing opportunity as you see it today. How have things changed over the last year?
Right now, it’s a gold rush and we’re very close to commercialising technologies that will radically advance automotive transportation. For example, all the key technologies that will be needed for level 4 and level 5 autonomy will be created within the next 2-3 years. I’m not saying that the government policies and customer acceptance will necessarily be in place, but the necessary hardware and software technologies will be available and policy-makers will have no “operational integrity” excuses for not allowing autonomous vehicles on the roads. The auto computer vision challenges (LiDAR, advanced Radar, etc.) will also be solved over the next 2-3 years. So, if you’re venturing in the automotive and transportation area, now is the time. Quite soon, the opportunities to disrupt at the levels we’re currently seeing will go away. There will be very limited opportunities for outsized returns. I think we have a maximum of ten more years of serious innovation in auto-tech before the venturing opportunity becomes seriously limited.
So, retirement beckons?
Yes. I’ll be moving to England to open a pub. It’s always been a dream of mine.
No. Typical venture capitalists are like monkeys and they will soon be jumping to other “investment” branches once the autotech opportunity is done. However, corporate venture capitalists don’t have that luxury. My group will continue to search for new start-ups working on automotive technology, but we will also probably expand our search to include new areas of business expansion that DENSO will target. Robotics might be a possible area of new investment for us.
Good. We have plenty of pubs in England, not enough corporate venture capitalists. Can you expand on the notion that autotech investing will become “seriously limited”? GCV data shows investment rising year-on-year in auto-tech and related sectors. Are you predicting a peak?
I believe that we are approaching the peak. In fact, it’s already getting very hard to fund a start-up in some automotive sub-sectors. For example, at last count, I have identified about forty-two LiDAR start-ups, eight of which are very well funded. Probably only three or four are going to survive the next 5 years, leaving a lot of failed businesses and unhappy investors. This means that we’re not going to see another wave of LiDAR companies getting funded. There will just be too many investors who got burned.
So that window has in fact already closed.
I believe so. It will be the successful start-ups that will do the R&D for the next iterations of LiDAR. It won’t be new venture-backed start-ups because those start-ups won’t get the venture backing.
Where are the windows still open? What are you looking for that you don’t currently have?
I am increasingly interested in how to combine the personalisation of transport with its increasing autonomy and automation. The car is going to remain, for a lot of people, a personal domain, an extension of yourself. Generally, most consumers are okay with non-intrusive automation, like ABS braking systems and intermittent windshield wipers for example, because the driver remains in control. But some of the autonomous and automatic features that are now being introduced are intrusive. Studies have shown that people don’t like those systems, and often turn them off. They don’t want to fight with their car about taking some sort of action, like changing lanes. If I, as the owner/driver of the car thinks it is safe to change lanes, but the system in the car does not think it is safe, that becomes a problem. And the auto industry needs to address this situation because we don’t want the consumers turning off systems that are meant to help them drive more safely. Also, the lingering effect on the consumers is that they then don’t like those new “autonomous” systems. The car needs to be able to understand the driving “personality” of each driver, and then adjust its systems to fit that driver. So, I’m certainly interested in start-ups that are addressing the problem of getting us to adopt and use new cars’ safety capabilities, for example, while still giving us the feeling of control over our own domain.
Can you give examples of where this fight is being avoided, where drivers still feel in control of their domain?
A great example is the birds eye camera system that Nissan has pioneered to give the driver a view of the car from above to help with parking, particularly with parallel parking on streets. This is being embraced and loved. For a lot of us, it’s been the end of cursing in the car while trying to park in a tight space.
So, you want more such innovations that free the driver without withdrawing his or her sense of control?
Yes! A lot of start-ups tend to think of personalisation inside the cabin – the radio stations, the seating position. But they now need to move to the car itself and how it responds to the driver, especially up to level 4 autonomy. With level 5, it won’t matter, but at level 4, we need to overcome increasingly negative attitudes to autonomy from drivers. For example, few of us like having a camera pointed at us, which seems to be the current solution for driver status monitoring systems, and many of us are worried about data security. So, start-ups that have alternative monitoring systems that don’t point a camera at your face and that can make the data anonymous, or that can prevent the release of private data would be of interest to me.
What else is of interest?
How to make a car see – computer vision. This excites a lot of VCs. By synthesizing hearing, sight and mental intuition, humans become good drivers. Cars need to do this through sensor fusion, they need to fuse all the information they’re getting from different sensors, understand it and make immediate decisions. And behind this problem is the “compute” problem. Basically, each car would need a super computer in order to achieve Level 4 autonomous driving. But you can’t stick a mainframe on the top of the car and no one wants one in their trunk. Don’t forget that silicon chip investing dried up several years ago. We’re going to have to find new ways to deliver the compute functionality that next generation vehicles are going to need.
How are your portfolio companies approaching these problems?
I can give you two good examples. ThinCI is a California-based company we invested in two years ago that is developing a unique computer chip architecture. It’s able to put deep learning algorithms on the chip itself. We’re confident we’ll get somewhere around 5-10 times the compute performance offered by the next best solution. Trilumina is an Albuquerque-based company we invested in back in 2016. We believe that the company’s VCSEL technology is going to be a fundamental illumination technology for both driver status monitoring systems and some LiDAR solutions.
What’s the strategic angle here for Denso?
The technology being commercialised by our portfolio companies is what our customers – the automotive OEMs – will need in the near future. Our strategic benefit is being able to balance internal development efforts against external development efforts; it is a type of buy versus build comparison. For example, ThinCI provides a step change in processing capability that will enable computer vision and complex decision matrices while also reducing power requirements. It would be difficult, if not impossible, for DENSO to internally develop that same sort of technology. Trilumina provides unique VCSEL technology for illumination applications, but at an affordable price. This is the reason why we venture; we want to enable continued success in our core business of supplying auto parts and we can’t always rely on our internal R&D to provide those types of new technology.