Google announced some cool consumer electronics devices at its San Francisco event yesterday hashtagged #MadeByGoogle: Google Home personal digital assistant, two new flagship phones under the new Pixel brand, Chromecast Ultra (capable of 4K video), Google Wi-Fi, and a VR headset for the Pixel phones. The usual sales channels—Verizon, Best Buy and Google Play—will distribute them.
It sounds like the consumer electronics business, but it is not.
Each product is thoughtfully designed with features that appeal to consumers. But they also enable Google to extend its infrastructure to increase revenues in existing businesses, to create new lines of business, and in the case of the Google Home digital assistant, to acquire a deeper understanding of users than its search engine provides.
+ Also on Network World: Amazon Echo, Google Home and their competitors will be a $2.1 billion market by 2020 +
Google’s hardware partners in past September announcements would deliver a complete product that Google would add its name and brand and sell it through its Play Store and through its channels. Google changed the relationship. Google engineers sat elbow to elbow with partner engineers, designing and engineering responsive application-specific devices. In the short term, custom-designed hardware built for specific applications perform better than commodity hardware. However, in the longer term, commodity hardware gets faster and less expensive, eliminating the need for customer hardware.
Looking at these devices from Google’s perspective, the company has a good reason for its #MadeByGoogle hashtag—to deliver high-performance hardware that enables Google’s services.
Google Home: Google Home pictured to the left in the photo below is an attractive, vase-shaped personal assistant that doubles as a streamed music sound system. Somewhat like Amazon’s Alexa, it can answer direct questions based on the 70 billion entries in Google’s knowledge graph ranked internet sites and interface with home control systems to, for instance, turn on lights when asked. It can also conversationally clarify a request for action, such as call an Uber.
Plus, it is the front end to machine learning systems in Google’s cloud to conversationally interact with users. Google Home has to be responsive, understand what a user said, filter noise from the background, and with very low latency, send the user’s part of the conversation to Google’s cloud for interpretation and a response.
Google Home is powered by Google Assistant, which also was demonstrated on the new Pixel phone. Peeling away the layers, there are public machine learning APIs that let independent developers create their own Google Assistant app. And finally, underneath all of the layers is the embedded Google Assistant software developer kit (SDK), due to for release next year, that lets other consumer hardware makers build Google Assistant features into their products.
Google CEO Sundar Pichai said Google’s intention is to shift from being a mobile-first company to become an AI-first company. AI-first means humans interact with machines conversationally, not just with phones. There is no reason why a conversational agent has to run on one type of device, such as a Pixel phone or the Home device. Google Assistant could be built into a refrigerator, a vacuum or a car with the SDK.
The Home device is a stable, responsive platform. It is better suited as a personal assistant than a smartphone because it is plugged in, so it is always ready. And its dedicated hardware makes it responsive to interpreting users’ voices.
During the Home demonstration, it appeared as though the conversation was taking place locally. However, it communicated with machine learning systems in Google’s cloud to understand and respond.
Like Google search uses machine learning to optimize the interpretation of user searches and rank results learned from search history, Google Assistant running on Home, Android and elsewhere will learn to respond by better understanding what the user has asked and better predict the response that best serves the user.
If Google Assistant can be optimized to better understand and predict, Google can use the same technique to predict which ads will appeal to a consumer and perhaps create new ways to monetize its free services. Home also will learn in a different context than search: in users’ homes. And if it makes itself an indispensable personal assistant, it will know everything about everyone living in these homes, improving its ability to predict which ads are most appealing to users.
Google Pixel and Pixel XL: These two beautifully engineered phones with a 5-in. and 5.5-in. screen, respectively, have been optimized to serve two purposes. First, the Pixel is Google’s new entry into the flagship phone race. It has a smooth, beautifully designed, glass-and-metal housing that looks and feels like the best flagship phones. Like all flagships, it has an amazing camera that achieved a top score of 89 from camera researcher DxOMark. And its specs, reported by GSM Arena, reflect excellent engineering and curation of top-tier components into a polished device.
It is also the first Daydream-ready, VR smartphone designed to be inserted in the Daydream View headset (pictured below). The Pixel design has been optimized when worn with the Daydream View headset with a very low latency gyroscope and inertial measurement unit (IMU) to track the user’s head movement and a very fast Qualcomm Snapdragon 821 SoC designed for VR that can render VR video on the screen that match the user’s changing head orientation.
More engineering and more expensive high-performance components have gone into the Pixel than would be required even for a flagship to make VR apps perform well while keeping the price in the flagship price range. A smartphone maker would not ordinarily pay for more expensive VR-ready components unless they were investing in a new application like Google is in VR.
Google Wi-Fi and Chromecast Ultra: These two devices are best explained together. Home Wi-Fi routers aren’t keeping pace with the steadily rising number of consumer devices in the home that stream audio and video. And they are hard to set up.
Google does not want to be a cable company like Comcast, but it does want users to stream more audio and video content from its Play Store and video content from YouTube, creating a bigger workload as content shifts from HD to Ultra-HD 4K resolution and 3D VR video. For most consumers, buying the right routers, configuring them and setting them up, and placing them throughout the home to provide a strong, evenly distributed Wi-Fi signal is beyond their skill levels.
Google’s Wi-Fi business model differs from that of a typical router company that tries to make a small profit in the cutthroat router market. Google has an incentive to solve the setup, configuration and signal distribution problem for the consumer because a better, faster and evenly distributed Wi-Fi signal will enable users to stream video and audio, as well as stream YouTube where its margins are much greater compared to routers.
Google Wi-Fi automatically configures multiple access points. Network management by its smart Network Assist automatically helps avoid Wi-Fi congestion and transitions the user to the closest Wi-Fi point for the best signal.
Chromecast Ultra fits into this network scheme. It is almost twice as fast as the second-generation Chromecast, and it supports streaming of 4K video to the latest high-resolution TVs and Daydream VR headsets.
Google doesn’t want to be Apple or Comcast. It wants billions of users to spend more time in front of every type of screen and immersed in VR consuming its content. Right now, the hardware needs to be optimized for this task. If successful, many consumer device makers will build devices like these to expand Google’s ecosystem.