Google Taps Machine Learning to Lure Companies to Its Cloud

  • Alphabet unit grabs Disney, Coke as new cloud customers
  • Nuance shares fall on concern about competition from Google

Google will create business tools and products based on its own artificial intelligence technology, seeking to entice more companies to rent its cloud-computing services.

The Alphabet Inc. unit plans to offer services such as audio transcription and image identification built around its machine-learning software. Google has used this technology for its own products, and is now making the capabilities available for other companies to rent and access over the Internet. 

The company also said it won several new large cloud customers, including the interactive division of Walt Disney Co., which now runs a web-subscription service on Google’s cloud, and Coca-Cola Co., which rented Google servers for a World Cup marketing campaign.

Google wants to broaden the appeal of its cloud services to more corporate customers. After letting Inc. get an early lead in the booming market, Google hired Silicon Valley legend Diane Greene to conjure a comeback. On Tuesday, the Internet giant said it will expand the number of places people can buy its cloud services, from four regions to 16 in the next 18 months. That revealed twin strategies for gaining share: developing new products based on its powerful internal technology, while meeting the sometimes-idiosyncratic needs of large companies. Google is third in cloud revenue versus Microsoft Corp. and Amazon Web Services, according to Morgan Stanley.

“In the future almost everything will be done in the cloud,” Google Chief Executive Officer Sundar Pichai said Wednesday in a speech at the company’s Google Cloud Platform NEXT conference, held at San Francisco’s Pier 48. “For years we have been investing in scaling up our infrastructure to do this.”

While modern cloud systems are based on “decades-old” technology, Google said, the company’s forthcoming products and services are designed for the next wave of cloud computing. To catch Amazon and Microsoft, Google must use any existing advantage it has -- and AI is an area of expertise for the online search provider.

Artificial intelligence “is the next transformation,” said Alphabet Chairman Eric Schmidt. “I’ve decided to spend as much time as I can on it because it’s so profound.”


Google’s plan to open its speech-recognition services to third parties sent shares of Nuance Communications Inc. lower on concern about increased competition in the voice-recognition software market. Nuance dropped 6.3 percent to close at $17.95 in New York.

Google’s machine-learning services will compete with similar ones from Microsoft, Amazon, and International Business Machines Corp. 

Startups “don’t have to hire a bunch of researchers and do research anymore. Amazon and Microsoft and Google and IBM have it as a service,” said Adrian Cockcroft, a technology fellow at venture capital firm Battery Ventures. 

Google’s approach, which pairs specific services like its vision and speech systems, with more general ones based on its free TensorFlow software, broadens what companies can build on its cloud. 

“It’s the layers of things that will be built on top that will be interesting,” said Cockcroft, who helped build many of Netflix Inc.’s Web video services on rented Amazon servers. 


Google will charge for these capabilities, a departure from its typical consumer-focused approach in which AI technology supports free Web-based services and apps such as Photos and Translate.

For its Cloud Vision system, which lets companies automatically detect faces, landmarks, logos and other features in images, monthly prices range from 60 cents to $5 for every 1,000 uses. The more times companies use the service each month, the lower the per-unit cost. There’s also a free tier for 1,000 or fewer uses during a month. The speech-recognition service is free during a preview phase, but after that the company will charge, it said.

Greene said the new machine-learning cloud services will turn Google into “an applied R&D team for our customers, in effect.”

Schmidt said the Mountain View, California-based company will put “thousands of people” to work on the systems that support its cloud over the next few years, in part to build the data center and network infrastructure needed to handle many large corporate customers.

Along with the machine-learning products, Google unveiled new ways to monitor the software that runs on its cloud, analyze the data held within it and link different products together with networking.

Before it's here, it's on the Bloomberg Terminal.