Now that Google has infused the brand of artificial intelligence known as machine learning into everything from search to speech recognition, it’s making the technology it views as tech’s next big wave available for other companies to use.
The search and advertising giant today announced an early “alpha” test version of a new Cloud Machine Learning product family that will be available on its public cloud service. The announcement came at GCP NEXT, the company’s first global cloud users conference in San Francisco.
The service will provide ways for businesses to tap into Google’s systems for machine learning, a fast-emerging way to let computers use artificial neural networks along with huge amounts of data to learn tasks on their own, rather than by humans programming them. Google made it clear that it aims to make Cloud Machine Learning a key differentiator from cloud leader Amazon Web Services, which has a huge lead in cloud services, and others such as Microsoft Azure that Google trails far behind in the $23 billion market.
“This is the next transformation” in the technology industry, said Eric Schmidt, executive chairman of Google parent company Alphabet. He suggested that machine learning will be responsible for the next wave of technology startups and initial public offerings in a few years.
Sundar Pichai, Google’s chief executive, also sought to position the industry at an “inflection point” at which it’s becoming apparent that much of the computing done by everything from small businesses to large enterprises will be done in the cloud. “We’re at the beginning of what’s possible in the cloud,” he said.
The surprise appearance of Pichai and Schmidt, in the top tier of Google management along with co-founders Sergey Brin and Larry Page, Google’s CEO, was a pointed sign that the company is serious about becoming more of a force in cloud computing. Potential customers have had their doubts about that, as Google’s cloud strategies have shifted often in recent years even as Amazon.com bolted ahead to capture a commanding lead.
Diane Greene, new senior vice president of Google’s enterprise and cloud business, addressed the perception head on during repeated questions from the press about Google’s commitment. “We are dead serious about this business,” she said. “Google’s all-in on what we can do with the enterprise. This is a long-term forever business.”
She also made it clear that she intends to make money at it, despite Google’s emphasis on undercutting rivals on the price of its cloud services. That point was seconded by Urs Holzle, senior vice president of Google’s technical infrastructure, who repeated a claim he made last year that cloud revenues eventually surpass those of Google’s massive ad business.
“We could not scale this up if we were losing money every time we deployed services,” Greene said with a laugh at the question of whether Google expected to earn profits from the cloud. “This is a business for us. We happen to have extraordinarily efficient data centers that allow us to offer lower prices.”
Beyond price, Google made it clear that it intends to pitch the Cloud Machine Learning services announcement as a key differentiator. In just the last couple of years, Google has significantly improved services ranging from search and YouTube recommendations to speech and image recognition to language translation.
Google already had introduced in test mode models for computer translation and vision that are already trained on massive amounts of data. Today, it also debuted a Cloud Speech API, or application programming interface for software developers, in test mode. Google also offers an open-source library of software code, called TensorFlow, that allows companies to create their own machine learning models to create their own services. “This will be the foundation for this next layer of programming not just at Google but everywhere else,” Schmidt said.
Beyond machine learning, Google introduced a number of feature improvements that, taken together, underscore a central Google push in the cloud: making it much easier for companies to use cloud services, especially hybrid systems using both their own cloud and public clouds. Although cloud services are intended to simplify a company’s computing infrastructure, the reality is that they require a lot of expertise and engineering resources to run.
“We decided to meet you where you are versus where we think you should be,” Schmidt told the audience. The phrase “meet you where you are” was repeated by other Google executives in what is clearly a marketing theme the company intends to hammer home to wary potential customers.
Indeed, several customers of Google’s cloud, such as Spotify Ltd. and Coca-Cola Co., appeared onstage to underscore that point. (However, there was no mention of Apple, which is reported to have signed a large cloud deal with Google.) Mike White, chief technology officer of Disney Interactive, said the company need to get access to more emerging technologies to make its movies better and more engaging. To do that, he said, “we need to be hooked into the cloud.”
Google will need more customers like those to counter the persistent doubts. “Google Cloud initiatives are too late and too little – 5 years too late and insufficient breadth of services to compete and win against either AMZN AWS or MSFT Azure,” Trip Chowdhry of the investment bank Global Equities Research, said in a note to clients.
Brian Gracely, an analyst with Wikibon, said Google’s technology is superior on a number of fronts to Amazon’s. But he thinks Google is still struggling to deliver it in ways that information technology people at large companies are accustomed to.
“The question is: Can Google transfer their knowledge to customers?” said Gracely. “The technology is super-powerful, but I don’t walk away today thinking Diane Greene has solved the enterprise.”
But it was clear today that Greene and Google are making a much more serious attempt to try.
Adapted from my post on SiliconANGLE.
This article was written by Robert Hof from Forbes and was legally licensed through the NewsCred publisher network.