300 Baud Modems and Autonomous Vehicles

By Josh Kopelman


I have a teenage daughter, and in a few years, she’s going to get her driver’s license. (Which terrifies me.) And while there have been incredible advances in autonomous vehicles over the last few years, I’m certain that she’ll still need to learn how to drive — the technology and regulatory environments just aren’t changing that fast. That said, after spending some time over the past few months looking at the autonomous vehicle landscape, I’m also equally certain that in 35+ years (when my daughter might have a teenage child), my grandchild will not know how to drive.

I recognize that 35 years is an awfully long time — and as a venture investor who invests funds (with 10–12 year lives) in companies (that typically must demonstrate traction in 18 to 36 months to raise subsequent financing), one of the big risks of investing in any nascent technology is being right but “way too early.”

I was fortunate to grow up in the 300 baud era. I got my first computer around 35 years ago, when dial-up web services offered a “preview” of the Internet yet to emerge. If one were to look back to the start of the consumer Internet, it’s clear that there were several “inflection points” that catalyzed the industry’s growth and accelerated its progress. Whether it be the advent of the free web browser, penetration of high-speed cable-modems, the accessibility of wifi and wireless, the regulatory landscape (around access, taxation, etc.) — there were certain milestone events that accelerated the penetration/availability of Internet technology.

So we’ve been thinking about how certain inflection points might emerge for autonomous vehicles. There clearly isn’t just one scenario — or even one group of scenarios… but we thought it would be an interesting thought exercise to try to anticipate what inflection points are “predicates” for accelerating penetration.

Today we wanted to share three of the milestones we believe foreshadow a reality where fully-autonomous vehicles rule the roads.

1.
Factory Installed Autonomy vs. Aftermarket Retrofit

While more consumers are excited about the semi-autonomous capabilities baked into today’s new car models — like lane correction — there are still over 200M older cars on the road in the U.S. that lack these new features. With the average lifespan of a car surpassing 11 years, it will take a lot of time for autonomous vehicles to enter the mainstream. And new technologies take a long time to reach scale. Even with Tesla’s success, they’re expected to sell just 55,000 cars this year (which is just 0.3% of the total new cars expected to be sold this year).

Clearly, one key “inflection point” for autonomous vehicles is the prevalence of autonomous features in the marketplace. While Tesla plans to upgrade their fleet with some autonomous capabilities this summer — and Elon Musk expects the technology for a fully-autonomous vehicle to be market-ready this decade — a bigger indicator will be the emergence of an aftermarket to retrofit existing vehicles for autonomous driving.

If aftermarket autonomous features pick up steam, we will see meaningful penetration of autonomous vehicles much faster than one might anticipate. A handful of companies like Cruise, have sprung up to outfit older vehicles with modern features, including advanced lane tracking and collision avoidance. As the components to enable these features inevitably become cheaper, smaller, and simpler to integrate into older vehicles, we will see progress toward a fully autonomous future gain momentum.

Specific investable areas we’re interested in:

Aftermarket components / systems that enable retrofitting of older vehicles for autonomous functions, and the companies that will make them.

2.
Persistent Internet Connectivity and Data in the Cloud

By 2020, an estimated 75% of cars shipped globally (~69M vehicles) will be built with the necessary hardware to connect to the Internet. In fact, a large number of car makers have already begun offering dedicated SIM cards and Internet plans in newer models. As cars become increasingly connected and enhanced with a complex array of sensors, the flow of data from vehicles will increase exponentially. According to one estimate from Hitachi, each connected car will send 25 gigabytes of data to the cloud every hour.

The result will be a need for far more robust data management and analysis platforms in the cloud, as well as a wave of new services tailored for passengers. Cars will have persistent connectivity (as offered by Vinli, a First Round company). Companies like Metromile (a First Round company) are able to wirelessly tap into telematics data (with consumer permission) to offer innovative consumer products — like automobile insurance that’s sold per-mile (rather than per month). Fleet operators will be able to offer real-time driver safety analytics (as envisioned by Zendrive, a First Round company). At the same time, there will need to be a robust app store native to vehicles with services that can be personalized based on past usage data and preferences.

The flow of data into cars may be even more critical as software push updates from automakers become the standard that enables cars to evolve in real-time. One of Tesla’s major advantages is that it can wirelessly push firmware fixes and new features to improve vehicle performance with no service stations visits. Not only does this make for a great consumer experience, but also huge cost savings in fixing problems that would otherwise trigger a recall. When applied to autonomous cars, push updates could be the bridge between today’s research and tomorrow’s efficient autonomous networks. After all, testing a technology with a handful of users is not the same as testing at scale, and push updates will be critical in overcoming the glitches that inevitably come to light with thousands of autonomous cars on the road.

Specific investable areas we’re interested in:

  • Expanded internet connectivity in vehicles
  • Distributed mesh networks
  • New digital services for passengers beyond internet radio
  • Remote vehicle diagnostics and repairs
  • Vehicle cybersecurity

3.
State vs. Federal Regulation

The regulatory environment will have a massive impact on the adoption of autonomous vehicles. So far, the federal government has been content to let individual states lead the charge in establishing policies regarding cars within their own borders.

The last official statement on autonomous vehicles from the National Highway Traffic Safety Administration (NHTSA) was released back in May 2013. While it set up a classification hierarchy for vehicle automation, ranging from Level 0 (no automation) to Level 4 (full self-driving automation), it fell far short of laying out a timeline or process for regulatory adoption. The statement even said, “In general, we believe that states are well suited to address issues such as licensing, driver training, and conditions for operation related to specific types of vehicles.”

Just like the states have led the way with the legalization of gay-marriage and marijuana, we’ve generally seen more flexibility from local lawmakers than the central government when it comes to progressive policy. I believe this could be the best case scenario for the emergence of autonomous cars. It allows the technology to get into the hands of early adopters and entrepreneurs much faster. Once the first state throws its weight behind full autonomy (i.e. allowing a car to navigate from start to finish without human oversight), we’ll see an explosion of utility for the car itself beyond personal mobility. For example, autonomous vehicles stand to reinvent distribution (including long haul trucking, overnight delivery and retail/local delivery) — and they will force cities and municipalities to rethink mass transit infrastructure.

California, Nevada and Florida have thus far led the charge in allowing autonomous vehicle testing on public roads, with an increasing number of states likely to follow suit in anticipation of the expected groundswell of economic opportunity. Projecting this trend into the future, it’s possible that the first fully-autonomous vehicle you experience will have to stay within state borders, or require you to swap license plates when crossing between them (as with Google’s current test car). Still, that’s a small price to pay for a technology with the potential to revolutionize not only the personal mobility market, but also the fundamental infrastructure underlying multiple related industries.

It’s been amazing to have had a front-row seat to the emergence and development of the consumer Internet. I look back at the 300 baud days with nostalgia and realize how far we’ve come. I’m pretty convinced that automotive technology today is still in it’s 300 baud phase — and I can’t wait to see what happens next.

If you’re working on something that fits one of the above themes, or if you have an opinion on the future of autonomous driving, I’d love to hear from you. I’m always tweeting at @joshk and you can share ideas you have in this area with me here.