Beware of the Legislative Slow Hand as Self Drive Act Inches Forward

Yesterday, the U.S. House of Representatives passed proposed legislation that would make it easier for companies to test self-driving cars on the roads. The new rules would add as many as 100,000 autonomous test vehicles to U.S. roads every year, compared to less than 500 today. Our guess is that the Senate will also approve the “Self Drive Act” and and it will become law by year end.  That’s good news, but it doesn’t change the likelihood that lawmakers will be the biggest roadblock to autonomous adoption.

The big picture. Taking a step back, the impact of autonomy will be widespread including trucking, ride sharing, insurance, entertainment, energy companies, auto part suppliers, parking lots and even fast food (fewer impulse tacos once you’re no longer in control of your ride).

The gap between testing and adoption. Make no mistake, the passing of the Self Drive Act is essential to advancing a paradigm shift in transportation, but the transition to a world of autonomy will take longer than the five years that many tech observers expect. And the biggest risk to the timing of self-driving adoption isn’t tech, but lawmakers’ aversion to risk, and the inevitable slow hand in making autonomous cars street legal. You can see it now, federal and state lawmakers feverishly debating AI mortality around a car’s crash path, overlooking undeniable evidence that human error causes more than 90% of accidents and machines can reduce that risk.

They’ll come around. While lawmakers will slow the adoption of self-driving vehicles, they’ll eventually come around because of all the reasons why the Self Drive Act moved forward today, because the world will be a better, safer place with self-driving cars.

Disclaimer: We actively write about the themes in which we invest: artificial intelligence, robotics, virtual reality, and augmented reality. From time to time, we will write about companies that are in our portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making investment decisions. We hold no obligation to update any of our projections. We express no warranties about any estimates or opinions we make.

Auto Outlook 2040: The Rise of Fully Autonomous Vehicles

Special thanks to Austin Bohlig for his work on this note. 

Today, we are introducing our 2040 Automotive Model, available here, detailing our projections for electric vehicles, autonomous vehicles, and fleet services through 2040.

The global automotive industry is quickly approaching a transformation that should fully take shape by 2040. While 20 years doesn’t seem very far away, keep in mind that technology is advancing at an accelerating pace — the next 20 years of innovation will see changes equivalent to what we’ve seen over the last 50 years. We expect to see three major automotive themes emerge: 1) the transition to electric, 2) fully autonomous vehicles, and 3) a higher percentage of people relying on ride sharing services as their primary source of transportation. We believe these three themes will create enormous market opportunities. While some of the traditional auto players will capitalize on these emerging themes, the competitive landscape will change dramatically as more technology companies enter the space to bring these revolutionary technologies to market.

Theme #1 – Transition to Electric

According to Bloomberg New Energy Finance, 84.0M new passenger cars and light commercial vehicles were sold globally in 2016, up ~5% y/y. Of all vehicles sold, 81.5M were internal-combustion engine (ICE) vehicles, 2.0M were hybrid, and 440K were electric. While electric vehicles only accounted for <1% of new vehicles shipped in 2016, this segment of the market has seen tremendous growth over the past 4 years, and we believe we are nearing an infection point for demand of electric vehicles. By 2033, we believe electric vehicles will surpass 50% of total market share. By 2040, we believe 86% (87.9M) of new cars sold will be electric vehicles; from 2020 – 2040 the electric category will experience an ~18% unit CAGR, while ICE vehicles will decline ~13% over that same time period.

Over the next 20 years, electric vehicles will become more affordable, but due to the advanced sensors, onboarding computing processors, and other components that will enable fully autonomous driving capabilities, we expect electric car ASPs to increase modestly. That said, in order for ICE and hybrid vehicles to compete, these categories will see prices steadily decline. In 2040, we believe the global passenger and light vehicle automobile market will represent a $3.8T annual market opportunity, up from $2.9T in 2016. The bulk of the growth will be driven by electric vehicle demand, which we anticipate to increase from $20B in 2016 to $3.4T in 2040, representing a ~18% CAGR.

While affordability will be a meaningful catalyst to electric car adoption, there will be multiple additional catalysts driving the shift to electric vehicles:

  • OEMs Focus Towards Electric – Today, and for the next 20 years, we believe the leading electric car OEM will be Tesla, but we anticipate almost all other traditional car manufacturers will eventually switch their focus to electric vehicles. Volvo was one of the first to do so, and recently announced all new cars they manufacture will be electric or hybrid starting in 2019. We anticipate other OEMs to make similar announcements in the coming years, providing additional tailwinds to the industry.
  • Government Intervention – We believe we will see legislation over the next 5 – 10 years enticing consumers to buy electric vehicles through subsidies; gas powered vehicles may even from the road. France and Britain plan to ban the sale of gas and diesel vehicles beginning in 2040. Scotland recently announced similar plans but with an implementation date of 2032. While this may seem extreme, it’s not unprecedented, even in the US: In the early 1900s, when the Model T began shipping in volume, the government banned horses from operating on the same public roads as automobiles.
  • Transition To Fully Autonomous – Although autonomous cars can take the form of ICE or hybrid vehicles, the majority of autonomous vehicles deployed will be electric cars because there are many synergies between the technology implemented in electric vehicles and what will be incorporated in fully autonomous systems. In addition, given our thesis that most car OEMs will switch their focus to electric, it only makes sense autonomous cars will follow suit.

Theme #2 – The Rise of Self-Driving Vehicles

Today, 99.9% of all passenger and light commercial vehicles on the road have little to no automation capabilities. However, Tesla and a few additional OEMs have made great strides in introducing what the industry classifies as Level 2 (Partial Automation). By 2040, we expect that over 90% of all vehicles sold will be “Highly” and “Fully” autonomous systems, classified as Level 4 and 5 automation, respectively. Here’s a brief definition of the different forms of automation according the National Highway Traffic Safety Administration (NHTSA):

  • Level 0: No Automation – A human controls all the critical driving functions.
  • Level 1: Driver Assistance – The vehicle can perform some driving function, often with a single feature such as cruise control, but the driver maintains control of the vehicle.
  • Level 2: Partial Automation – The car can perform one or more driving tasks at the same time, including steering and accelerating, but still requires the driver remain alert and in control.
  • Level 3: Conditional Automation – The car drives itself under certain conditions, but requires the human to intervene upon request with sufficient time to respond, but the driver isn’t expected to constantly remain alert.
  • Level 4: High Automation – The car performs all critical driving tasks, monitors roadway conditions the entire trip, and doesn’t require the human to intervene. But self-driving is limited to certain driving locations and environments.
  • Level 5: Full Automation – The car drives itself from departure to destination, and the human is completely removed from the process.

This will not be a gradual transition from one level to the next; we expect most players to skip Level 3, going straight from Partial Automation to High or Full Automation. We also view Level 4 and 5 as very similar levels of automation; Level 4 has a steering wheel but Level 5 does not. So, in our forecast we combine Level 4 and Level 5 into one category labeled “Fully Autonomous.”

Self-Driving Car Rollout Begins in 2020, Inflects in 2028

We estimate that ~130K Level 2 vehicles will be sold in 2017; over the next few years, the industry will see a significant acceleration of Level 2 vehicles delivered, occupying a growing percentage of new vehicles sold through 2033. However, we believe 98K Fully Autonomous vehicles (Level 4 and 5) will enter the market in 2020, which is when the transition to self-driving will start to take shape. While some Level 1 and 2 systems will still be sold in 2040, the two groups combined will account for <6% of all new vehicles delivered, and >94% of systems will take the form of fully automated vehicles. It will take time for fully automated vehicles to gain meaningful traction, largely due to legislative hurdles, but beginning in 2028 we believe the industry will see an influx in demand for Level 4 and 5 automobiles. We expect the industry will go from shipping 98K Fully Autonomous vehicles in 2020 to 96.3M in 2040, representing a 41.2% CAGR over that time frame.

Leaders in Autonomy

While there will be many companies that will benefit from the transition to fully autonomous vehicles, a few companies are already positioning themselves to be early key players:

  • Tesla – Tesla has already established their dominance in the electric vehicle market, and we expect their commanding market position to prevail through 2040. We estimate that Tesla currently controls ~20% of the global electric vehicle market, and although we anticipate competition to increase in the years to come, we believe Tesla can maintain low-to-mid teens market share through 2040. Almost all Teslas today incorporate Level 2 driving automation, and while Tesla is hoping to get fully autonomous cars on the road by 2019, we believe their near-term focus will be ramping production of the Model 3 and less on getting fully autonomous cars on the road. That said, we view Tesla’s leadership around autonomous driving technology and AI is a step up from almost everyone else, and expect larger deployments to begin in 2020.
  • Waymo – It is still not completely clear what Waymo’s go-to-market strategy will be with regards to autonomous cars, but the company will have a meaningful presence. Waymo’s biggest competitive advantage thus far is the millions of miles their self-driving cars have driven and the terabytes of data gathered, which they can use to train their self-driving car algos. Waymo has already launched a ride sharing service in Phoenix, AZ, and we wouldn’t be surprised if they sell a Waymo branded self-driving car or develop a self-driving car OS that they license to third party car OEMs.
  • Traditional OEMs – It will be a significant challenge for traditional car OEMs to compete as we transition to electric and full autonomy, but there will be some legacy car brands that effectively transition by leveraging decades of car manufacturing expertise to compete with Tesla and Waymo. Ford is one traditional car company that has begun the transition, including promoting a CEO with deep autonomous experience and acquiring leading startups in the space (Argo). While some of these traditional car companies will be able to develop self-driving systems internally, we believe the more effective way will be to enter the space via acquisition.
  • Start-Ups & Others – The self-driving car industry is still in the very early innings, and other tech giants such as Apple, Uber, Lyft, and relatively unknown startups will deliver meaningful innovation. There are many technological gaps that still need to be solved before self-driving cars are fully deployed on public roads.
  • Apple – We continue to expect Apple to play in the self-driving car market, possibly bringing a self-driving car to market, or, more likely, developing an autonomous system for self-driving cars. We’ll cover this with more detail in a future note.

Theme #3 – Transition to Ride Sharing Services

In addition to the world transitioning to electric and autonomous vehicles over the next 20 years, we expect an increasing number of consumers will forgo owning a car and rely fully on ride sharing services for transportation. While we anticipate the number of cars sold will continue to increase through 2040, many of these new cars will go directly towards ride sharing services.

We estimate there were 1.3B passenger and light vehicle cars in use in 2016, of which 5% were dedicated to ride sharing services. In the coming decades, as ride sharing becomes more cost effective and reliable, the percentage of cars dedicated to ride sharing services will increase steadily. By 2040, we estimate that 68% of all vehicles in use will be dedicated to fleet services. As a result, the number of cars personally owned by individuals will decrease at a -2.0% CAGR from 2020 to 2040. Today, companies such as Uber and Lyft dominate the ride sharing market, but we anticipate other leading tech companies (Tesla, Waymo, etc.) and traditional car OEMs to introduce a ride sharing services as well. We also envision a future where individuals that own an autonomous car are able to deploy the system to a fleet service when they are not using the car (e.g., while at work), extracting value from the car’s dormancy.

Bottom Line

The global automotive industry is quickly approaching a paradigm shift, and the types of vehicles on our roads and the competitive landscape in the car market is going to change significantly by 2040. Level 1 and 2 automated vehicles will still be sold, we estimate that >94% of new cars sold will be fully automated in 2040. Some of the traditional auto players will successfully transition to these emerging themes, but several tech companies are most likely to become leaders in the space. Over the next several decades, the biggest headwinds to full autonomy will likely be legislative rather than technical, but the safety and efficiency benefits of autonomous cars will provide a strong tailwind to broad public acceptance and rapid market growth.

Disclaimer: We actively write about the themes in which we invest: artificial intelligence, robotics, virtual reality, and augmented reality. From time to time, we will write about companies that are in our portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making investment decisions. We hold no obligation to update any of our projections. We express no warranties about any estimates or opinions we make.

HomePod Uniquely Positioned to Win Smart Speaker Market

In December, Apple will ship HomePod, a smart speaker with a unique focus on music. Don’t be fooled, however, by HomePod’s music-focused marketing; Apple has a grander vision than delivering a better sounding Echo. The company is making Siri a ubiquitous, ambient presence that connects and controls all your connected devices and services – and making a leap forward in the transition to voice-first computing.

The importance of natural language interface. The way humans interact with computers is changing. Today, we use our keyboards, mice, and touchscreens to interact with computers. In the future, we’ll simply rely on our voice, gestures, or even our thoughts. In the near-term, voice is quickly becoming a preferred interface. At Google I/O in may, CEO Sundar Pichai said that 20% of mobile queries are now made via voice search. Moreover, 42% of people in a MindMeld study said they have started using voice commands in the last six months. Natural language as a computing input is not only a more natural way to interact with our devices, but it can also be remarkably more efficient. When typing or clicking, users will be very brief, leaving the computer with little information to act on. Asking a verbal question, however, allows for more involved queries with which a machine can much more easily determine intent and deliver more specific information. This is one area in which Siri excels. Siri is able to process commands with multiple steps, such as, “make a note called Slide 4 in my Presentation Notes folder that says: change transition.” Users will also be able to say, “send directions to Steve’s house to my phone,” or, “turn on the TV and play the newest episode of Westworld.”  These functionalities are not unique to Siri, but Apple’s seamlessly integrated ecosystem of devices puts them in a position to employ voice-first computing in ways their competitors can’t match.

SiriKit is important for the future of the HomePod. Siri, which will be the AI brains inside HomePod, has recently extended an olive branch to developers with the introduction of SiriKit. SiriKit allows third-party developers to add voice capabilities to their apps. Consumers will be able to do a lot more with Siri than set a timer or ask for the weather. As Apple’s vibrant community of developers works to integrate voice into third-party apps, users will be able to get real work done with verbal inputs, marking a turning point in voice-first computing. However, lining up Siri against Alexa and Google Home reveals measurable gaps in ability early on – but the Siri we have come to know on our iPhones and the upcoming Siri that lives in HomePod with third-party integrations are two very different animals.

How does HomePod stack up? The smart speaker market has undergone impressive growth and rapid adoption in recent years, growing 62% in 2016 alone. When you use one, it’s easy to see why – the verbal interface is very natural and serves as a clear glimpse into the future of our interactions with computers. Meanwhile, this market continues to be flooded with products from new entrants, and from the continuing dominance of Amazon and Google.

Amazon’s Echo, released in November of 2014, costs $179. The Echo is part of a broader family of devices that also includes the Echo Show, Dot, Tap, and Look, each with their own distinctive features and price tags. Alexa-enabled devices command over 70% of the smart speaker market. Between licensing Alexa’s software to third-party hardware manufacturers, Amazon’s aggressive sales and marketing efforts, and allowing developers to augment user experience with Alexa Skills, the Echo family has solidified itself as the de facto voice platform of today. Alexa Skills, which integrate voice capability into an expansive range of third-party applications, are Amazon’s number one advantage going forward. While Amazon remains the market leader today, the sustainability of Amazon’s dominance comes into question going forward without an existing base of integrated phones.

The Google Home, which came along a full two years after Alexa, costs $129 and has a 24% market share. Google has opened its voice platform to various third parties, but does not have the exposure of Alexa Skills, or SiriKit. Google’s natural language processing, which is reported on extensively and tested in our research, is best in class, and may propel Google to the forefront of voice-first computing in the coming years.

In the small sliver of market share that remains, numerous alternatives have entered, hoping not to be left behind as voice becomes a major computing interface. Some prominent recent and upcoming entrants include the Alibaba Tmall Genie, Lenovo Smart Assistant (powered by Alexa), Harmon Kardon Invoke (powered by Cortana), and Samsung’s Bixby Assistant. With the underlying technology in its fledgling days, early leaders and laggards are bound to appear, but the core offerings remain fundamentally similar.

Apple is well-positioned for long-term success. As the technology improves, which our research suggests can happen quickly, competitors will converge, and the long-term winner will be the product that provides its user with a heightened experience and improved efficiency. We believe Apple is uniquely positioned to do so, as Apple’s device ecosystem delivers a frictionless experience, which will only get better with the adoption of voice-first computing.

Apple’s device ecosystem delivers a frictionless experience, which will only get better with the adoption of voice-first computing.

Interestingly, Apple has included an A8 chip in its HomePod, the same chip included in an iPhone 6. The A8 chip is much more powerful than the chips competing home assistants run on, which poses the question: what else is Apple planning with the HomePod?

Disclaimer: We actively write about the themes in which we invest: artificial intelligence, robotics, virtual reality, and augmented reality. From time to time, we will write about companies that are in our portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making investment decisions. We hold no obligation to update any of our projections. We express no warranties about any estimates or opinions we make.

Apple’s Core ML Brings AI to the Masses

AI is one of the core themes on which we focus at Loup Ventures. And as analysts, we heard Google, Facebook, Amazon, and Apple emphasizing their focus on AI over the last several years. Google CEO Sundar Pichai has commented on each of the past three Google earnings calls that Google is transitioning from mobile-first to AI-first. Facebook has recently spent a lot of time and resources developing chat bots on its platform, and has utilized AI to create a better news feed, and improve photo recognition. Amazon uses AI extensively with recommendations, and is integrating third-party AI models into AWS. While Google, Facebook and Amazon are each making significant progress as it relates to AI, it’s worth noting that Apple was the first company of the four to embrace it.

Apple’s AI roots date back to the mid 1990s with handwriting recognition on the Newton. In June Apple announced Core ML, a platform that allows app developers to easily integrate machine learning (ML) into an app. Of the estimated 2.4m apps available on the App Store, we believe less than 1% leverage ML today – but not for long. We believe Core ML will be a driving force in bringing machine learning to the masses in the form of more useful and insightful apps that run faster and respect user privacy.

Apple’s history in ML. Apple’s history in ML dates back to 1993 with the Newton (a PDA Apple sold from 1993 to 1998) and its handwriting recognition software. While not a complete list, Apple has since used AI in the following areas:

  • Facial recognition in photos
  • Next word prediction on the iOS keyboard
  • Smart responses on the Apple Watch
  • Handwriting interpretation on the Apple Watch
  • Chinese handwriting recognition
  • Drawing based on pencil pressure on the iPad
  • Extending iPhone battery life by modifying when data is refreshed (hard to imagine that our iPhone batteries would be even worse if not for AI)

Core ML. Core ML was announced at Apple’s June 2017 WWDC conference. It’s a machine learning framework that sits below apps and third-party domain specific AI models, but above processing hardware inside of a Mac, iPhone, iPad, Apple Watch, or Apple TV.

Source: Apple

Core ML allows app developers to easily incorporate third-party AI models into their apps. App developers don’t need to be experts in AI and ML to deliver an experience powered by AI and ML within their app. In other words, Apple will take care of the technical side of incorporating ML, which allows developers focus on building user experiences.

At WWDC, Apple outlined 15 ML domains that can be converted to work on apps:

  • Real Time Image Recognition
  • Sentiment Analysis
  • Search Ranking
  • Personalization
  • Speaker Identification
  • Text Prediction
  • Handwriting Recognition
  • Machine Translation
  • Face Detection
  • Music Tagging
  • Entity Recognition
  • Style Transfer
  • Image Captioning
  • Emotion Detection
  • Text Summarization

What’s different when it comes to ML between Apple vs. Android? Google provides developers with TensorFlow compiling tools that make it easy for Android developers to integrate ML into their apps. Developer blogs suggest that Core ML makes it easier to add ML models into iOS apps, but we can’t compare the comparative ease of adoption. However, we can say they are different when it comes to speed, availability, and privacy.

  • Speed. ML on Apple is processed locally which speeds up the app. Typically, Android apps process ML in the cloud. Apple can process ML locally because app developers can easily test the hardware running the app (iOS devices). In an Android world, hardware fragmentation makes it harder for app developers to run ML locally.
  • Availability. Core ML powered apps are always available, even without network connectivity. Android ML powered apps can require network connectivity, which limits their usability.
  • Privacy. Apple’s privacy values are woven into Core ML; terms and conditions do not allow Apple to see any user data captured by an app. For example, if you take a picture using an app that is powered by Core ML’s vision, Apple won’t see the photo. If a message is read using an app powered by Core ML’s natural language processor, the contents won’t be sent to Apple. This differs from Android apps, which typically share their data with Google as part of their terms and conditions.

AI for the masses. In the years to come, iPhone users updating their favorite apps will experience a step function improvement in utility, but may never know that Core ML is behind the curtain making it all possible. We can all look forward to continually improving apps thanks to Core ML.

Disclaimer: We actively write about the themes in which we invest: artificial intelligence, robotics, virtual reality, and augmented reality. From time to time, we will write about companies that are in our portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making investment decisions. We hold no obligation to update any of our projections. We express no warranties about any estimates or opinions we make.

Humans Are a Bigger Existential Risk Than AI

Elon Musk continues to warn us of the potential dangers of AI, from debating the topic with Mark Zuckerberg to saying it’s more dangerous than North Korea. He’s called for regulating AI, just as we regulate other industries that can be dangerous to humans. However, Musk and the other AI debaters underestimate the biggest threat to humanity in the AI era: humans.

For the purposes of the current debate, there are three potential outcomes debaters of artificial intelligence propose:

  1. AI is the greatest invention in human history and could lead to prosperity for all.
  2. A malevolent AI could destroy humanity.
  3. An “unwitting” AI could destroy humanity.

There are few arguments in between worth considering. If the first possibility was not the ultimate benefit, then the development of AI wouldn’t be worth exploring given the ultimate risks (2, 3).

There’s certainly a non-zero chance that a malevolent AI destroys humanity if one were to develop; however, malevolence requires intent, which would require at least human level intelligence (artificial general intelligence, or AGI), and that is probably several decades away.

There’s also a non-zero chance that a benign AI destroys humanity because of some effort that conflicts with human survival. In other words, the AI destroys humanity as collateral damage relative to some other goal. We’ve seen early AI systems begin to act on their own in benign ways where humans were able to stop them. A more advanced AI with a survival instinct might be more difficult to stop.

There’s also a wild card relative to the first outcome that the two sides of the AI debate. On the road to scenario one, the positive outcome and probably the most likely outcome, humans will need to adapt to a new world where jobs are scarce or radically different than work we know it today. Humans will need to find new purpose outside of work, likely in the uniquely human capabilities of creativity, community, and empathy, the things that robots cannot authentically provide. This radical change will likely scare many. They may rebel with hate toward robots and the humans that embrace them. They may band behind leaders that promise to keep the world free of AI. This could leave us with a world looking more like the Walking Dead than utopia.

Since the advent of modern medicine, humans have been the most probable existential threat to humanity. The warning bells on AI are valid given the severity of the potential negative outcomes (even if unlikely), and some form of AI regulation makes sense, but it must be paired with plans to make sure we address the human element of the technology as well. We need to prepare humans for a post-work world in which different skills are valuable. We need to consider how to distribute the benefits of AI to the broader population via a basic income. We need to transform how people think about their purpose. These are the biggest problems we face as we prepare to enter the Automation Age, perhaps even bigger than the technical challenges of creating the AI that will take us there.

Disclaimer: We actively write about the themes in which we invest: artificial intelligence, robotics, virtual reality, and augmented reality. From time to time, we will write about companies that are in our portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making investment decisions. We hold no obligation to update any of our projections. We express no warranties about any estimates or opinions we make.