Amazon Is Doubling Down on AI

Amazon’s Q1 earnings call last night sounded somewhat similar to Google’s call just an hour earlier, highlighting a lot of the same AI and machine learning investment themes.  Jeff Bezos discussed increasing leverage of AI and machine learning to drive operational efficiencies both internally and with AWS customers.  On a smaller scale, the beta Amazon Go store in Seattle is one of several competing visions for the future of retail. Leveraging computer vision, sensor fusion, and deep learning, Amazon Go allows for a checkout-less shopping experience.  We suspect Amazon’s bookstore expansion, beyond its six stores today, may soon begin to resemble characteristics of Amazon Go.

The most immediate AI lift for Amazon is coming via increases in Prime member engagement from Echo products and, more recently, Alexa-enabled Fire tablets.  Commentary from the earnings call indicated Prime Music, Prime Now, and Amazon Fresh Grocery are all seeing meaningful lift from those voice-enabled devices.  We highlighted Alexa’s recent momentum in our work, Can Anyone Catch Alexa?, now with more than 12,000 skills (apps) and roughly 100 smart-home manufacturing partners integrating Alexa.  Contrast this with, by our count, only around two dozen device manufacturers using Google Assistant.  With Bezos indicating a “doubling-down” on investment in Alexa and Echo last night, we expect Amazon and Google to remain head-to-head for leadership in the digital assistant market.

Another example of Amazon extending it’s lead in AI and machine learning is the new Echo Look, an Alexa-powered device for your closet. Echo Look features a camera that takes pictures of your outfit, then uses AI to make fashion recommendations and catalog your wardrobe.

Source: Amazon

Echo Look shows us what screen-less computing will look like in the future and, at the same time, how early we are in the transition. Amazon is pushing speech-driven computing faster and further, but the Echo Look also reminds us that we’re in the first inning of devices that will be increasingly commonplace as the next computing paradigm emerges. Adding cameras, sensors, computing power, skills and form factors to devices like the Echo product line will dramatically improve the utility of the category and drive adoption. Clearly, Amazon gets it and is determined to lead the way.

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.

Google Earnings Reiterates AI, VR, AR Ambitions

Google CEO Sundar Pichai used tonight’s Q1 call to double-down on his “AI-first” mantra.  Several areas were highlighted, where Google will lead and define its mission to organize the world’s information with machine learning.  Beyond AI initiatives to enhance core consumer products like search, mail, maps, and Google Play, increased machine learning investment was highlighted in at least two segments of Google’s ad model.  These include ‘Smart Bidding’ where machines predict in real-time how an ad should perform in front of a particular target and adjust advertiser bids to maximize ROI.  Separately, in Google’s highly profitable app install ad business, namely Universal App Campaigns, machine learning is being used to best promote developer apps across Google properties including Search, YouTube, and the Display Network.

Management also made a couple of callouts on VR.  Google’s VR platform Daydream is seeing more than half its usage consuming video, with YouTube VR being its #1 app by time spent.  At a recent event, we heard from Google that people are using the Daydream headset to “hold the phone” for them as they watch video laying down.  While not an exciting VR use case, we heard that users are spending multiple hours in VR this way, so it is getting users comfortable with the experience.  VR investment near-term still appears focused on mainstream products like Daydream and accompanied VR produced content such as YouTube VR, Google Earth VR, and Tilt Brush.  We suspect Google is also experimenting with advanced VR and AR hardware, although no mention was made on the earnings call.  We continue to believe Google is the best positioned player to provide the AR operating system of the future, and the company’s leadership position in machine learning and AI only reinforces that view.

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 Dream Car: Hardware + Software

Friday’s news of Apple’s permit to test self-driving cars in California should not have come as much of a surprise given the poorly kept secret of Project Titan. But the permit begs the question of whether Apple is building a car or just building software for a car.

Apple’s overarching philosophy has been to own both the hardware and software to create superior product experiences. Typically, this means owning the technology stack from end to end for a given product; for example, Mac + macOS, iPod + iTunes, and iPhone + iOS; however, sometimes the company stops short of owning the entire experience. The Apple TV requires a third-party television panel, although we would argue that is the equivalent of plugging your Mac into a third-party monitor. Once the Apple TV is engaged, the experience is all Apple. Unlike a television, a car is not just a dumb panel. Modern vehicles require a tremendous amount of sensors and other electronics that monitor the exterior and interior. In an ideal world, Apple’s car project would involve the company building the actual automobile, combining hardware and software. In reality, the complexity of designing and manufacturing a vehicle may push the company to integrate deeply with an automotive partner or partners in an effort more similar to the Apple TV  — plugging Apple’s technology into an existing product.

From a software standpoint, building an automotive product beyond CarPlay is a no-brainer.  Apple has one of the better stacks of necessary components to build a great car OS:

  • Entertainment: iTunes/Apple Music
  • Assistance/Voice: Siri
  • Navigation/Local: Apple Maps
  • Image Processing/Recognition (Autonomous Driving): iPhone Camera
  • Security: Touch ID
  • Third-party Software: App Store (potential for OTA software updates)

Despite the obvious software advantages, the auto market poses challenges that Apple may not be able to overcome on the hardware side, i.e. building the car end-to-end. First and foremost, Apple would likely need to find a manufacturing partner because it is not a manufacturing company, but a design company. Foxconn and other manufacturing partners build iPhones, iPads, and Macs to Apple’s specs. A company like Magna Steyr may be a capable of building a car to Apple’s design specs. Aside from finding a partner, we note that the typical automotive design process takes 5-7 years. Even on an accelerated time table, Apple is likely multiple years away from a completed hardware design for a car. Finally, Tesla and Google have a multi-year lead on Apple in developing autonomous vehicle capabilities, including the associated testing mileage. We see autonomous driving capabilities as a key factor in auto desirability over the next five years.

Apple is almost certainly exploring how it could build an entire car, but as we learned the hard way with an Apple television, exploration does not mean a product comes to market. Apple is the best connected device maker in the world and the car is the biggest connected device in the world. There is a natural fit in the self-driving car market if Apple can figure out how to get past the challenges of making the hardware.

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.

Can Anyone Catch Alexa?

Amazon’s third-party developer strategy has Alexa looking like the Gingerbread Man of digital assistants. After taking over CES in January with multiple Alexa-enable devices, there’s been little pause in Alexa skills and device-enabled growth. As of this writing, we count more than 12,000 Alexa skills (apps) and roughly 100 manufacturers with integrated Alexa IP across multiple smart home categories. And yesterday, Amazon opened its Echo voice processing IP to third party developers, extending Alexa’s lead in smart devices. Given our belief that natural language processing is one of a few core technologies that will enable the screen-less future of computing, we think it’s important to track the pace of the key players in the field.

Source: Amazon

Skills growth impressive, but getting the basics right remains the key for scale.

One measure of Alexa’s increasing utility is the growth in skills that can be downloaded to Echo devices. In just the last 3 months, nearly 5,000 skills have been added to Alexa’s repertoire, which now tops 12,000. Roughly a third of Alexa’s skills are knowledge-based – from education apps like the Old Farmer’s Almanac to trivia categories like Lesser Known Star Wars Facts. Other growth categories include health and fitness with skills like answers to common medical questions and workout suggestions. In addition, nearly 100 smart home skills are available today, an important catalyst for scaling the Alexa-enabled device ecosystem. It’s too early to tell how much scalable utility these skills bring to Alexa usage, particularly the nearly 500 knowledge/trivia skills categories. There is a fun-factor with a lot of these skills and voice access is seamless relative to paging through dozens of apps on your phone. However, Alexa needs to get better at answering basic information-related queries, which we believe will produce sustained utility and growth in Alexa-enabled devices. In a recent test, we found the Echo answered only 41% of information queries correctly.

We found the Echo answered only 41% of information queries correctly.

Device integration pacing well ahead of Google Home.

We count close to 100 manufacturers across several categories that are compatible with Alexa IP today. Smart home coverage, perhaps the most seamless hands free utility Alexa offers, continues to grow – from lighting to locks to thermostat control. Included on this list is Google’s own Nest device, a collaboration that began early last year; however, the relationship has been anything but seamless as a preponderance of Nest skill reviews suggest. We wonder if after a year of collaboration, whether Alexa and Google will ever nest together. Contrasting Echo’s ~100 smart home partners with Google Home, we find a much shorter list. Only around two dozen device manufacturers are integrated with Google’s Assistant IP. Amazon has taken advantage of Echo’s head start on Google Home by pushing integration across many manufacturers and platforms. Google Home will likely follow, but has a long road ahead.

Read More

AI’s Busted Bracket

The Loup Ventures NCAA bracket contest isn’t as hotly contested as we thought it would be. We entered Bing’s AI bracket into our pool, and it’s just as busted as the others. In fact, Bing’s bracket will finish at the bottom of our pool, in 7th place, regardless of the outcome of tonight’s game. We would like to think that we outsmarted AI, but the reality is that predicting the outcome of the NCAA tournament is more a matter of luck than skill. Bing’s performance doesn’t mean it’s broken, just unlucky this year.

* Bing Predicts 2017 NCAA Basketball Bracket

To date, Bing has chosen 39 out of 67 games correctly, including the opening round. Bing was 2 of 4 in the opening round, 24 of 32 in the 1st Round, 9 of 16 in the 2nd Round, 4 of 8 in the Sweet Sixteen, before going 0 for 4 in the Elite Eight and ending its chances at victory. If you look at Bing’s bracket now, it will show a different story, because it re-picked winners for matches after each round. Even with this adjustment, it only picked 47 of 66 games correctly, leading into tonight’s game. In the adjusted rounds, Bing chose Final Four weekend right with Gonzaga and UNC as winners, with UNC ultimately taking home the crown.

Read More