The Road to Autonomous

Introduction

Uber is one of our investments at Lead Edge. We believe it represents one of the most transformative companies of our time and we continue to be bullish on its prospects. In the past, we have written about the investment merits of the company, but in light of the constant news around autonomous cars, we thought our LPs would like to understand our view on how we think the industry might unfold, and how it could impact Uber’s business. Please note that these are our views contained within and we are trying to guess the future, which we will inevitably get wrong, however we just wanted to share our current thinking.

But first, some relevant business updates on Uber in Q3:

• Uber sold their China business unit to Didi Chuxing in August. Though much of the press cited this as defeat, we viewed it as a MAJOR win for Uber. As part of the merger, Uber now has 18% ownership in Didi Chuxing, the market leader in China, which should become one of the largest ride-sharing markets in the world. We believe long term this could add $10B to $25B of market cap to Uber.
• Uber made an acquisition of Otto, an autonomous vehicle technology company. Otto was founded by Anthony Levandowski, who was an early executive on Google’s self-driving car team. The Company has been focused on technology that could be retrofitted into trucks, and a few weeks ago, Otto completed the first driverless commercial delivery—50,000 cans of Budweiser across 120 miles from Loveland, CO to Colorado Springs, CO. We believe trucking may be the right initial application, given the relative simplicity of highway vs. city driving.
• Karhoo, a startup that enabled consumers to compare taxi and black car services, suddenly shut down in October after raising around than $250M from investors. The company was attempting to compete with Uber by creating a metasearch of car transport options but was unable to secure additional funding after burning through its cash in 18 months. Investors are no longer tolerating excessive amounts of cash burn and growth without solid unit economics is going out fashion. All of these indicate a healthy normalization from the high valuations we have seen over the last two years.

A double-edged sword
The prospect of fully autonomous cars is a double-edged sword for Uber. In one way, having them would make Uber a mass alternative to car ownership. With no driver costs, Uber could price their rides to the point where it makes economic sense for most people to get rid of their cars. Rather than having an empty car sit in your driveway, you would use ride-sharing only when you need it, and Uber could still capture strong margins for providing this service. On the other hand, if Uber had fully autonomous cars, its competitors would most likely have them too, and it would be a lot easier to deploy a competing fleet of robo-cars—driving profit margins lower for all ride-sharing companies. Yet, the market size is so large when you consider the entire world using self-driving cars that we would expect there to be multiple large winners.

However, we currently believe fully autonomous cars will have a limited to modest impact on Uber’s business during our investment holding period over the next five years. Longer term, anything is possible but there are a huge number of moving parts that need to come together before widespread fleets of self-driving cars can compete with Uber’s human fleet. There are several reasons for this view.

First, we believe while fully self-driving cars are coming they aren’t coming all at once. It’s not like we are going to wake up January 1, 2020, and the whole world is going to be moving around in millions of self-driving cars. We believe the full global roll out of self-driving cars will take many, many years if not decades to complete. In 2014, per the International Organization of Motor Vehicle Manufacturers, there were 900 million cars in use around the world and 65 million new cars sold, so it would take nearly 14 years to replace the existing fleet of cars if all new car sales were autonomous cars. Of course, not as many cars may need to be on the road for self-driving since these cars can have significantly higher utilization rates. Let’s assume total car ownership can go down by 50% as people share cars more efficiently so there only needs to be 450 million self-driving cars on the road. If we assume that 20% of the cars sold each year are fully autonomous that means 13 million self-driving cars are sold each year (note in the US in 2014, 7.7 million new cars were sold). That would imply it takes over 34 years to replace all the cars on the road to self-driving. Of course it won’t take this long since at some point more than 20% of cars sold each year will be self-driving yet in the beginning it won’t be anywhere near 20%. Let’s also not forget that in order for self-driving cars to be ubiquitous there needs to be a global infrastructure to support the cars, which will also take many years to build.

Second, no one knows how self-driving cars will start to be rolled out although many people pontificate on it. Some people say there will be specific self-driving car lanes on the highway (similar to the car pool lane now), while others say there will be specific zones in cities where self-driving cars operate. Frankly, we have no idea, all we believe is that a combination of technological, regulatory and practical challenges will make the roll out take longer than people think. I personally believe that my four children aged 4, 2, 1 and 1 will all have driver’s licenses yet their children won’t due to self-driving cars becoming ubiquitous. Many people think I am crazy for believing this I just think it will take longer than most people think. I like to draw an analogy to high speed internet access on smartphones. The first 3G networks were introduced in 1998, fourth generation 4G networks in 2008 and the first iPhone was released in 2007. 10 years ago people were saying that fast ubiquitous, global cell phone service was just around the corner. Well you know what, it’s now nearly 10 years later and in a crowded football stadium in New York City I still can’t upload pictures from my phone to Facebook; from my houses in Westport, CT and Santa Barbara, CA I cannot make a cell phone call without it dropping let alone try to download any data; and I recently read an article from a San Francisco newspaper (i.e. the center of global technology innovation) that “Despite glossy maps from telephone companies showing blanket coverage across the Bay Area, The Chronicle found that large swaths — including the heart of San Francisco — are riddled with more holes than a bagel shop.” I realize these are corner cases but I expect that most every person reading this letter has experienced a dropped cell phone call over the last 6 months. In the case of self-driving cars, a dropped cell phone call could be the difference between life and death. I expect over the next 10 years that most if not all of these cell phone corner cases will be solved. Don’t forget as well that for all of us who love iPhones/smartphones there are still millions of flip phones on the market (i.e. the ones you used in 1999). The point in all of this is that changing the habits of billions of people around the world who have grown up driving a car won’t change overnight. Sometimes we forget that the world is a whole lot bigger than Silicon Valley.

The labor cost equation
The largest cost of a ride with Uber is the payout to drivers. When someone is deciding whether to drive themselves (where they still have to pay for gas/depreciation) or use a ride-sharing service, they are making a trade-off between the convenience of Uber and paying the cost of the driver.

In developed countries, labor is expensive. Minimum wage creates a floor on the price of a ride and consequently creates a barrier to mass adoption. Not everyone can afford to pay other people to chauffer them around, and many choose to drive themselves. However, in the developing world, where labor costs are substantially lower, Uber can price rides more attractively for riders. This increases adoption, and wide swaths of middle class consumers can use these services very often.

In some ways, China represents the greatest market opportunity in the world because of its income inequality. There are people who are willing to drive for very little income, and there are lots of middle-class consumers with money to spend. When you have a driver willing to work for a couple dollars an hour, you already get most of the benefit that self-driving cars bring to the equation. In the US and Europe, people definitely have more money to spend, but labor costs are higher. Therefore, the race to have self-driving cars is more relevant in the US and Europe, as it can open huge pockets of demand by dramatically decreasing the price of rides.

When does self-driving technology impact Uber’s business?
Most of the commercially available autonomous technology out there today requires a driver behind the steering wheel. Yes, drivers can take their hands off the wheel at times but drivers are instructed to be alert in case any issues arise. For example, Uber is offering rides in Pittsburgh in its self-driving car, but there is a “safety driver” up front in the driver seat to ensure the ride goes smoothly. Tesla’s Autopilot feature has similar safety guidelines requiring the driver to be alert since the system is not perfect, as noted by the deadly crash in Florida that killed the driver when a tractor trailer truck turned in front of the Tesla and the car didn’t stop.

Until true driverless cars are cruising our streets, picking up and dropping people off without a human in the driver seat … there is little impact to Uber’s business model. Going back to the labor cost equation, if a driver needs to be behind the steering wheel even for corner cases and exception handling, Uber or any other ride-sharing service will need to compensate that driver for their time. Maybe one could read a book while being an Uber driver because the car is doing most of the work, and so you can pay the person $10 vs. $15 per hour. So long as human drivers are a necessary part of the ride-sharing experience, Uber’s massive and growing network of drivers will remain a barrier to entry for companies looking to compete with the business.

There are still major improvements that need to happen with autonomous technology, before Uber sees any impact to its business model by self-driving car competition. Cars need to be able to handle all driving circumstances and weather conditions. The below chart depicts the various levels of automation, from Level 0 (no automation) to Level 5 (full automation), which requires no driver in the front seat. Only at Level 5 automation is Uber’s business model potentially impacted. For reference, Tesla’s Autopilot, defined as an “advanced driver assistance system” is currently at Level 2.

What does the government care about?
The government’s key priority in supporting the development of autonomous technology is to improve the safety profile of the cars on our streets. More than 33,000 people die of car accidents per year in the United States, and this technology could help reduce that number of deaths.

The reality is that many of those deaths could be prevented by increasing the number of cars on the road that have Level 1 to 4 autonomous technologies. Better cameras and sensors with varying degrees of assisted-driving technology will help people avoid deadly mistakes while driving. That said, many of the regulations proposed to date mandate a driver be behind the wheel, in the case that manual intervention is required by a human. These interventions could be weather related or otherwise, but they highlight the idea that the government does not need to sanction Level 5 driverless cars to accomplish its safety goals. This ultimately helps Uber’s competitive position because it creates inertia around the deployment of driverless ride-sharing fleets, but at the same time, makes the addressable market smaller because driver costs continue to be part of the equation.

Mileage math
Much of the press around ride-sharing services and autonomous car technology cites the eventual doom of car manufacturers. We think quite the opposite. We believe these services could help smooth revenues for car makers, turning them into more predictable, recurring revenue businesses.

Though the lifespan of a car can be measured in years, in reality, it is measured in miles. If you put 250K miles on a car in a year, it’s probably not going to last much longer than that, hence why taxis have a lifespan of just a few years.

Here is a very simplistic example of how an auto maker can smooth its revenue on the back of ride-sharing adoption:

• Let’s assume the useful life of a car is 100,000 miles and that a car costs $25,000
• A typical person drives 10,000 miles per year, so a car will last for 10 years

If 10 people buy cars, they collectively spend $250,000 today but don’t need another car for 10 years. The car maker gets $250,000 today but doesn’t get another car sale for 10 years.

Assume 10 people ride-share instead; they have to refresh their shared fleet every year, because they are putting 100,000 miles on each car as a group. Now, the car maker gets $25,000 every year over the 10-year period, still earning $250,000 in total, just smoothed out over time.

This example isn’t exactly practical because it would be hard for 10 people to share 1 car, but it illustrates how a car manufacturer doesn’t generate less revenue when people use services like Uber. This dynamic can positively impact a car maker, who would much rather prefer a steady stream of predictable car orders rather than infrequent orders. Even if people downsize on car ownership, they are just shifting miles driven from their personal cars to Uber cars, and the end number of miles being traveled by the passenger isn’t changing. That is why car makers like GM and Toyota are cozying up to Uber—because they want to be OEM partners for Uber drivers!

If people were to actually ride-share
The risk to carmakers is not the taxi or black car product that Uber is well-known for. It is Uber Pool. For those of you who aren’t familiar with the product, Uber Pool is a short-distance carpooling product. An Uber driver might pick up and drop off multiple people along a certain route, similar to a bus, but with the flexibility of going anywhere the demand takes it. Today, Uber Pool is still in its early days, but it has the ability to reduce the number of miles driven by cars on the road. For example, if 2 people individually drive from San Francisco to Palo Alto every day (30 miles each way) that is 120 miles of driving between two people. Instead, if were to carpool every day via Uber, we could cut the number of miles driven collectively by half to 60 miles.

Automakers should fear the behavioral shift to Uber Pooling more than anything else. Car manufacturers make money when we put miles on our cars, and if we are clocking fewer miles by carpooling, they stand to lose in a meaningful way. We do not believe people will entirely get rid of owning cars. For families that need to transport children in car seats, or for golf players that need to store clubs in their trunks, owning a car still has its many benefits that ride-sharing services cannot provide. Over time, however, it is possible people will reduce the number of cars they own, opting for ride-sharing instead.

Uber Pool and the practice of sharing rides with strangers, has the potential to dramatically reduce traffic congestion and carbon emissions. As much hype as self-driving technology gets, our belief is that Uber Pool is the most powerful experiment Uber is working on today. If we can get more people into fewer cars, our highways will be more efficient and our carbon footprint lower. We encourage our LPs to try Uber Pool and let us know what they think. The Company loves when we send them feedback. In our opinion, we believe that Uber Scheduled Rides is the first step in a broader plan in regards to Uber Pool. Imagine right now that between 6 AM – 615 AM tomorrow 3 people in Westport, CT all have scheduled rides to Manhattan and they are each paying $80 dollars taking 3 separate cars. Now imagine that Uber connects these people together and that instead of paying $85 each they pay $27 dollars each. This not only lowers congestion on the road but it massively increases the set of people who can take a car over other forms of transportation (note the train currently costs $16.75 per trip).

Recent Industry Press – Pieces of Interest:
We wanted to share a few articles and reports we read in the past quarter that we believe are interesting and relevant to our portfolio and investment strategy.

Joe Tsai: Alibaba Finding Its Groove
This article features a brief overview on the history of Alibaba followed by a quick Q&A with Joe Tsai, Executive Vice Chairman of Alibaba. The article shows how Alibaba has evolved from an e-commerce company to a platform company with many fast-growing business units. Similar to Amazon, Alibaba has either built out or invested strategically in new areas of opportunity – for instance, Alipay, Alibaba’s online payment platform, controls close to half of China’s online payment market. There is also Alibaba Cloud, a cloud computing services and solutions provider, that sells to SMBs through enterprise-class customers at incredibly competitive prices (analogous to Amazon’s AWS). Understanding why these tech giants invest strategically in certain assets should absolutely be factored into any prospective investment opportunity. For instance, this past week, Amazon started to hire “home assistants” or “cleaning technicians,” presumably to offer free housecleaning services to its Prime members. There are massive implications for businesses that offer housecleaning services. TaskRabbit… Handy… these are businesses that have raised massive amounts of capital, slugged through difficult marketplace dynamics, only to be potentially squashed out of nowhere by Amazon. For that reason, it is important to understand not only who your competition is today, but also who it might be in the future. In evaluating investment opportunities, VCs must be careful in considering how easily a larger business could enter its market.

Overdosing on VC: Lessons from 71 IPOs
This article was written by Eric Paley, managing partner at Founder Collective, and it speaks to the difference in capital efficiency of start-up businesses that have gone on to IPO. It outlines the different paths many unicorns have taken to IPO and how these businesses subsequently performed in the public markets. The article questions how to think about growing efficiently, enduring unsustainable burn rates, and trading on the public market. It is important to consider how funding history can affect terminal value / exit opportunity of a certain business. For instance, a business that has raised hundreds of millions in venture capital has likely commanded a high valuation. If performance slows down, it creates a misalignment between shareholders and management. Shareholders might have strict structure and plans for getting their money out, whereas management would likely prefer to raise more money, but avoid a down-round to not negatively affect company moral. This is why the Associates at Lead Edge are trained to consider capital efficiency in evaluating an investment opportunity. Because certain industries require heavy upfront capital investment prior to generating meaningful revenue, Lead Edge has typically shied away from these businesses. Often times infrastructure software businesses or Internet of Thing businesses will raise $20-50M prior to generating revenue. It is difficult to underwrite future growth into a high valuation just based on innovative technology.

The article continues to talk about performance on public markets of heavily funded companies versus more efficient, less funded companies. The takeaway here is that heavily funded companies often times do not generate a better multiple return on capital than the lesser funded companies – it is actually the opposite. Therefore, it is very important to understand that fundraising is a strategic choice that needs to be carefully considered by any entrepreneur. Often times entrepreneurs will make funding decisions opportunistically, or potentially worse out of a sense of pride or false validation. Just because a company has a high valuation on paper does not reflect future liquid value of that business long-term!

The Tech Bubble Didn’t Burst This Year. Just Wait
This article covers how the funding environment for entrepreneurs and venture capitalists has changed significantly since 2012. In 2012, an entrepreneur with a pitch focused around mobile-first solutions for ‘x’ would likely walk away from an investor meeting with a check in hand. This was due to mass adoption of smartphones leading to the success of businesses like Instagram and Pinterest. Today, there is no parallel technology that has excited mass appeal warranting heavy investment. It could be wearables, Internet of Things, AI – however it seems unclear still. This means fewer opportunities for entrepreneurs in the near-term, while investors are flush with capital to deploy. For that reason, we have seen the highest quality startups raise nearly unlimited funds at very high valuations. Uber, Airbnb, and Snapchat have commanded top dollar valuations and now have a war-chest of capital to fight off potential entrants into their respective markets. However, on the contrary, there are a number of businesses that have raised money at very high valuations that are struggling. Perhaps the biggest unicorn disappointment (excl. Theranos) came when One Kings Lane sold recently to Bed Bath & Beyond – a company it was meant to disrupt – for a mere $12M, after once being valued above $1B. While it is unlikely that unicorn businesses die off quickly, perhaps we see one per quarter. Recently, Uber rival Karhoo shut down after reportedly burning through $250M of funding. This will likely be a slow correction as opposed to a sudden crash, so important to keep an eye on once highly touted businesses going under. Companies in the middle-of-the-road scenario that are performing decently will likely have to take on additional capital at flat or down rounds. Many of these companies are good, but have raised money as if they were truly great. An example here might be Instacart, a grocery delivery company, selling shares to Whole Foods at a $2B valuation, the same price it had raised money two years prior.

Links to articles:
http://www.alizila.com/joe-tsai-alibaba-finding-its-groove/
https://techcrunch.com/2016/10/15/overdosing-on-vc-lessons-from-71-ipos/
http://www.bloomberg.com/news/articles/2016-10-20/the-tech-bubble-didn-t-burst-this-year-just-wait