Wednesday 27 July 2016

Google Races to Catch Up in Cloud Computing


When it comes to cloud computing, Google is in a very unfamiliar position: seriously behind.
Google is chasing Amazon and Microsoft for control of the next generation of business technology, in enormous cloud-computing data centers. Cloud systems are cheap and flexible, and companies are quickly shifting their technologies for that environment. According to analysts at Gartner, the global cloud-computing business will be worth $67 billion by 2020, compared with $23 billion at the end of this year.
“The world’s biggest maker of computer servers is making machines just for these guys,” said John Lovelock, a cloud analyst at Gartner. “It’s the nexus of things like big data, social networks and mobility, and the next big thing, which is artificial intelligence.”
For Google, a loss in cloud computing would be a rare misstep for a company that revolutionized media with its advertising business, and then made the world’s leading smartphone operating system.
A victory for Google, however, could change how we work, turning advanced computing into an everyday utility that we use to run factories, interact in virtual reality or read one another’s emotions. Given Google’s track record, it’s worth considering the prospect.
But it will be an uphill climb.
Amazon Web Services, which began its cloud product a decade ago, remains the leader. The company took in $2.6 billion, 9 percent of Amazon’s sales, in the first quarter of 2016. Profits from the service made up 56 percent of Amazon’s operating income. Those numbers may well be higher when Amazon reports its second-quarter earnings on Thursday.
Microsoft styled itself a cloud company, too, and the company said last week that revenue from Azure, its cloud business, which was founded in 2010, rose 100 percent over the last year. Cloud technology also figures in crucial businesses like Office 365.
In contrast, Google Cloud Platform does not even figure in the earnings reports of Alphabet, Google’s parent company. That has to sting, since the company owns perhaps the largest network of computers on the planet, spending close to $10 billion a year to handle services like search, Gmail and YouTube.
Google is moving rapidly to change things. Three announcements it made last week show how it hopes to gain ground on Amazon Web Services and Azure.
First, the company said it has used artificial intelligence to cut the power use in its data centers 15 percent, a huge decrease considering how efficient these data factories were already.
Power is probably the largest single cost for all three of the cloud companies. Google is almost certain to use its savings to reduce prices, much the way it won in search advertising by figuring out its competitors’ costs, then undercutting them.
That ability to find energy efficiency may be a powerful tool to sell to others over Google Compute. Mustafa Suleyman, the head of applied artificial intelligence at the company’s DeepMind subsidiary, said the techniques could be used at power plants, refineries and other big industrial systems.
“This certainly gives Google an edge,” he said. “Other people focused on narrow problems. We’ve focused on the widest possible problem.”
Google also released for public use ways to transcribe and analyze recorded speech for things like meaning, emotional content and whether a speaker was happy or sad. An outside company that worked on the project — Google would not say which one — used it to analyze over two billion minutes of customer service calls. It works in 80 languages, Google said.
At the same time, the company moved its customers on the United States West Coast from a data center in Iowa to facilities in Oregon. Google’s network, which it claims is larger than the internet, can send the equivalent of 375 hours of video a second.
A move like that only makes sense when a company wants to offer the kind of split-second performance needed for virtual reality or instantaneous customer interactions over networks of cellphones and sensors.
At a recent conference at Amazon Web Services for software developers, an executive from iRobot, which makes the Roomba vacuuming robot, talked about using the cloud to map homes and human behavior in houses with potentially “hundreds” of connected devices. Doing something like that would require instantaneous connections.
Can faster networks, lower prices and lots of artificial intelligence put Google ahead? Amazon’s lead seems to give it an edge for at least the next couple of years, as its cloud branch has perfected a method of developing hundreds of new cloud features annually. Yet while the company appears to have some basic artificial intelligence features, called machine learning, it seems to have little in the way of speech recognition or translation.
Mr. Lovelock, the Gartner analyst, predicted that Google would offer businesses the insights it has gained from years of watching people online. “Amazon views the customer as the person paying the bill, while Google believes the customer is the end user of a service,” he said. And Microsoft is promoting itself as the company that has products customers already know and use.
“Everyone has to play to the strengths the market already sees they have,” Mr. Lovelock said.