The Google Cloud Platform is one of the fastest growing businesses across Alphabet, company executives said Thursday on its first quarter 2017 earnings call.

Alphabet reported revenues of $24.8 billion for the quarter ended March 31, 2017, growing 22 percent versus the first quarter of 2016. Net income for the quarter increased to $5.42 billion compared to the same quarter in 2016 with net income of $4.21 billion.

Of the Google Cloud Platform (GCP), Google CEO Sundar Pichai said GCP saw the largest growth within the company in terms of Capex and in terms of headcount.

“The heavy lifting is around how we meet enterprises in the market,” Pichai said. “We created two new areas: one is the office of the CTO, which helps customers solve difficult technical problems. The second is our Advanced Solutions Lab, where customers can get help from machine learning experts.”

He mentioned that at its Google Cloud Next conference in March the company announced its acquisition of Kaggle, the largest community of data centers focused on machine learning and AI. Kaggle has become part of GCP.

Pichai said machine learning and artificial intelligence (AI) are fueling innovation across the entire company. And Google wants to leverage all its technologies across each service it offers. “Our approach is one Google,” he said.

In October 2016, Google created its Google Cloud Machine Learning Group to deliver cloud-based machine-learning software to businesses. Google’s Cloud Machine Learning is based on an open source library called TensorFlow, which lets developers build machine learning models and then scale them to production. On the earnings call, Pichai said the TensorFlow platform has spawned “more than 6,000 GitHub projects and counting.”

Pichai also mentioned the company’s open source AI project Deep Mind. The project goes beyond machine learning, which requires giving a computer thousands of examples and counter-examples. DeepMind is chasing the computer science discipline of deep learning, where the computer is able to educate itself.

Finally, he said Google introduced a new technique for training deep neural networks on mobile devices. The technique is called Federated Learning. It enables people to run a shared machine-learning model while keeping their underlying data stored locally on mobile phones.

“I'm really happy with how we are transitioning to an AI-first company,” said Pichai.

Perhaps worth noting is the fact that Google is eyeing opportunities in retail with all its various technologies, including cloud, machine learning, and AI. This comes at a time when traditional brick-and-mortar retail is suffering mightily.

“Ninety percent of retail is still offline,” said Pichai. “In the past year, local shopping queries have increased by 45 percent, and the number of retailers that provide us with their local inventory feeds has doubled. Retail overall also happens to be an amazing category for us from a Google Cloud standpoint.”