Why Do So Many of the Future Technologies Google Shows Off Appear Once and Never Again? | Google I/O

In the end, this comes down to how the tech industry finds its next breakthrough.

Yesterday morning, Google I/O opened at the Shoreline Amphitheatre in Mountain View.

More than 10,000 attendees packed the outdoor seats and the surrounding lawn as if they had come to a music festival, and sat through a two-hour opening keynote.

The mood was as charged as a festival's, too. Especially when Google announced Google Home, its artificial-intelligence speaker, and played this video on the big screen: the applause and cheering ran a good 30 seconds.

Most people in the crowd had used Siri, and plenty had used Amazon's Echo. But watching Google Home hold a conversation like a person — about live traffic, about astronomy and geography — and show off one ability after another that Echo and Siri lack, the whole place caught the excitement.

Sitting in the press seats, I had goosebumps, and typed a message to colleagues watching the livestream in the Beijing office: "Doesn't this speaker feel like an extra person in the house?"

They wrote back that, next to the "food delivery war" going on at home, this is what a technology company ought to be making. You could feel the same excitement in comments across social networks.

Two hours later, still excited, I walked around the venue looking for a chance to try the Google Home that had just been announced. If I hadn't run into a confused visitor at the Nest booth, the excitement might have lasted a while longer.

"No, no, we didn't make that speaker. This is Nest." The staffer from Nest, Google's smart home division, looked embarrassed as he explained that they had no Google Home of their own to show.

If you weren't watching closely during the keynote, you would have missed that Google Home is in fact the work of the Chromecast team — the team that used to make the dongle the size of a USB stick that throws your computer screen onto a TV.

Before this conference, though, Nest had always been treated as Google's smart home division, full stop.

In early 2014, Google paid $3.2 billion for Nest Studio, which made smart thermostats, pulling Silicon Valley's hottest smart home venture — and its founder, the "father of the iPod" Tony Fadell — into the company.

That same year, Fadell spent $550 million on DropCam, another smart home darling.

At last year's conference, Nest took part in launching Brillo, a smart home operating system, and Weave, an internet-of-things protocol — both gestures toward building a smart home platform. The audience sitting with me then didn't stint on applause and cheers either.

Everyone outside the company assumed Google was building a genuinely mainstream smart home hub around Nest. It looked that way: more than a dozen partner manufacturers plugged into the Nest-centered platform. The Nest team itself grew to 1,100 people in 2015.

Nest's fall from favor came with warning signs. At CES in January this year, appliance makers no longer mentioned the platform and backed Amazon's smart assistant instead. Meanwhile the infighting inside Nest spilled out unexpectedly in March.

In this year's exhibition area, the Nest booth was tiny, squeezed in with Android Wear and other products, taking up two not-very-large tables.

To Build a Smart Home, Google Has Already Gone Through Three Teams

Nest is not even the first Google smart home project to fail. At Google I/O in 2011, the company unveiled Android@Home, a smart home concept.

A year later, Android@Home's first product — the oddly shaped Nexus Q set-top box — launched alongside Google Glass at the 2012 conference. You could use an Android phone as the remote and play audio and video from the Google Play store and YouTube.

This little sphere, led by the Android team, was the first piece of hardware Google designed entirely in-house, and it was made in the United States.

But it drew doubts for its thin feature set from the day it went on sale: Apple and Roku both sold comparable devices at a third of the price that did more. The Nexus Q was pulled less than six months after it went on sale.

Over the past five years, Google's smart home project has changed hands three times:

Google's smart home project has changed hands repeatedly

2011

Led by the Android team: the Android@Home software concept, the Nexus Q hardware

2014

Led by the Nest team: the Nest hardware line, the Brillo operating system, the Weave communication standard

2016

Led by the Chromecast team: the Google Home smart speaker, the Chromecast TV dongle

This Is Not the First Attempt That Went Nowhere

When a project stalls, swapping in a new team and trying again is standard practice at Google.

At this I/O, Google introduced Allo, a text chat app with a built-in AI assistant and automatic replies, plus a companion video chat app, Duo.

Allo was another product that drew heavy applause and cheering on site. Once the excitement faded, though, plenty of people asked the same thing: what happens to Hangouts?

Google has overhauled its chat tools more than once:

• Google Talk (2005–2015), a chat tool built on Gmail accounts;
• Google Hangouts (2013–?), the chat app installed on Android phones outside China, combining text messaging with the chat features of the Google+ social network. Gtalk was folded into it.
• Allo & Duo (2016), with AI built in and end-to-end encryption. Both use a phone number as the account, where the earlier chat tools used an email address.

Allo's rival is obvious: Facebook's WhatsApp. WhatsApp has more than a billion people using it every month, also leads with encrypted messaging and phone-number accounts, and has recently added some AI features of its own.

To attract new users, Google simply built two new apps rather than improve the ones it had.

Look further back and there are many more fields Google announced it was entering and has made little progress in since:

Social networks. Google Plus was the most recent social failure, and it was already Google's fifth attempt at social.

Car operating systems. The Open Automotive Alliance, a sweeping plan for Android to take over every piece of software and every system inside a car, has so far reached only simple infotainment (Android Auto).

Wearables, Android Wear. The Apple Watch hasn't become the next iPhone either, but Android Wear, which came out earlier, sells even less.

Try First, Explain Later Has Become a Google Habit

Google's advertising business will not grow forever, least of all now that Facebook competes with it head-on, and it needs new sources of revenue. That is the motive behind its constant experiments.

Entering so many new markets, and replacing a failed team with a new team and a new idea, does not look good — someone even built Google a product graveyard.

But among the countless attempts there are successes: Gmail, YouTube and Chrome, each past a billion users, and the biggest of all, Android.

Each failure costs a limited amount, while any one of those products can cover the losses of a thousand failures. It is the same logic by which a venture capital firm can be wrong about hundreds of companies and still make a lot of money, as long as it backs a few big ones.

Today Alphabet's whole corporate structure encourages independent businesses to build their own brands, which gives small teams more reason to experiment.

Since Android, though, we have not seen another success like it.

In Fact, the World's Two Biggest Companies, Run Very Differently, Have Both Yet to Find Their Future

Today Apple and Google, the two Silicon Valley companies, are the most valuable companies on earth. As of this writing their market caps are $521.9 billion and $487 billion.

Numbers three through seven — Microsoft, ExxonMobil, Berkshire Hathaway (Warren Buffett's investment company), Facebook and Amazon — are all under $400 billion.

With iPhone growth topping out, Apple needs a new market more urgently than Google does. But Apple's method is to put the whole company's strength behind a handful of products.

That follows from how the two are managed. Unlike Google, which encourages teams to run themselves, Apple does not even give the iPhone, the Mac or the iPad their own divisions. Its departments are drawn by function — design, software, marketing — and every new product needs all of them working together.

Jobs set up this structure after he returned to Apple in 1997. It makes sure the company concentrates on a few new products each year, and it keeps a business unit that has already grown large from crushing the one that carries the future — the way Microsoft's Windows division did to its phone business years ago.

This is a shape better suited to hardware. That is also why Apple's failures in new hardware are never that bad. When we say the Apple Watch has not succeeded, it still sells 10 million watches a year, far ahead of Samsung, which has been selling smartwatches for years longer. "Not successful" means it cannot be a breakthrough on the scale of the iPhone, big enough to fill the hole the iPhone's decline leaves.

But Apple's structure also stops it from entering several brand-new markets at once. Last year it truly entered only two: wearables, with the Apple Watch, and online music, with Apple Music.

By their different methods, the world's two biggest companies have spent the past few years exploring one new market after another — wearables, the living room, tablets, cars…

But since the iPhone and Android remade the smartphone, none of those efforts has broken through.

This May Be the Whole Tech Industry's Problem

Perhaps we should ask whether this is not about what Apple and Google are capable of, but a sign that the whole tech industry has finally hit a wall.

Since the 1970s, when Jobs and Wozniak's Apple Computer got started, the personal computer industry has never stopped moving.

Personal computers spread in the 1980s, the internet spread in the 1990s, and smartphones and the mobile internet rose fast from around 2007.

Each of those shifts pushed computers deeper into daily life. In the decade that smartphones spread, one field after another — social networks, e-commerce, mobile games, mobile reading, online video — won hundreds of millions of users.

But today everyone has a smartphone, and every year brings fewer new features. Every phone maker faces the same problem.

And the next generation of computing devices — virtual reality, the smart home, wearables, driverless cars — has yet to show that it can become the next smartphone.

The new form of computing has not arrived, and software and internet services have been mined nearly clean: in the familiar categories — social, messaging, video, ride-hailing, reviews, classifieds — the top few companies already have hundreds of millions or more than a billion users. Building something new and taking their users away is not easy.

No wonder Google turned this conference toward artificial intelligence, saying "machine learning" — one way for AI to teach itself — over and over.

This was not just talk on stage. Before I/O opened, Otavio Good, an engineer on Google Translate, told QDaily that every department in the company is now trying to fold machine learning into its products, whatever those products are. His own team recently used machine learning to let the translation app read the text in a photo.

With a new platform nowhere in sight, Google's plan is to use what it has built up in artificial intelligence to make better products than its rivals.

Allo is the plain example. Google is trying to use AI to give people a better experience and to take users, and users' time, from Facebook's chat tools — the foundation of mobile advertising.

In other words, fight for a market someone else already holds, rather than create a brand-new one.