27 November 2017

Google, Amazon, or Microsoft: which cloud is right for your AI?


Businesses who begin exploring artificial intelligence (AI) technology will soon become aware that an affordable and relatively easy solution exists for addressing specific problems: leveraging cloud-based AI.

Instead of installing new software at considerable expense, or beginning the lengthy process of developing your own solution, today your company, or you as an individual, can access APIs readily made in the cloud to do your heavy lifting for you.


It comes as no surprise that three of the big names in cloud-based AI are Google (Google Cloud), Amazon (AWS cloud), and Microsoft (Azure).

They all offer AI platforms where you can access similar things: machine learning (ML) toolkits for your natural language processing (NLP), image recognition, and data crawling needs, as well as much more.

Picking your side

It is a common difficulty when choosing between the product offerings of the big three tech giants: they all offer a similar service, and all of them appear to be of a high quality. How do you choose?

Ian Massingham, Technical Evangelism Leader for AWS, told Access AI, “We have a distinctive set of services for developers that are building AI and ML systems because we operate at several different levels within the stack. We have abstracted services, making it really easy for developers to incorporate pre-packaged AI features directly into their apps.”

AWS count Stanford University, Pinterest, and Duolingo among their many customers using the AWS Cloud for its AI services.

Yuval Dvir, Head of Global and Online Partnerships at Google Cloud, told us that, “The advantage of Google is that we’ve been doing ML and AI for more than 15 years. It’s been injected in our products and services since before the peak of AI and ML hype that exists today. We also have the new innovations being introduced with Deep Mind, and in search and Gmail”.

Google Cloud count Coca Cola, AXA, and Spotify amongst their customers.

Microsoft have also announced that they are building on decades of research into AI in order to develop new tools for Azure.

Elon Musk recently made headlines after Open AI defeated the world Dota 2 champion by tweeting appreciation to Microsoft for the use of Azure, in which the Dota 2 playing bot was trained, saying, “This required massive computing power”.

Clearly Amazon, Google, and Microsoft are all focusing on cloud-based AI services, and there is no company with a clear lead on the others.

Dvir expressed to us that this situation can only be good for customers. “I don’t think one company should have all the answers”, he said.

Co-operation and integration is the key to success

Dvir told us that, while there are many variables influencing the decision customers make between Google, Amazon, and Microsoft, “Google has always opted for an open ecosystem and environment to benefit all. A hybrid cloud strategy makes sense for everyone. Some workloads on Google Cloud enable easy migration to AWS and back and forth.”

Previously of Microsoft, Dvir echoes the sentiments of many in the AI space who are calling for greater co-operation and less rivalry.

The idea would be if you are using AWS Cloud solutions, then you should also be able to use elements of Google Cloud in tandem without having to go through an overly complex conversion or worse still having to buy new hardware.

This would allow the democratisation of AI, giving customers choice and encouraging technology companies to work together and with the open source community to provide the best possible offering.

Breaking bad habits

Openness and co-operation has been a focus of Microsoft’s since Satya Nadella became CEO in 2014, a significant change for a company which a federal judge ruled “kept an oppressive thumb on the scale of competitive fortune” in 2000.

“I was there when Satya took over”, Dvir told us, “since then [Microsoft] have made strides to open up”.

This rang true at an open source round table that Microsoft hosted at their Future Decoded event on October 31st, which featured Mark Shuttleworth of Canonical, Scott Johnson of Docker, Amr Awadallah of Cloudera, and Michael Ferris of Redhat.


“They take and never give back. They’re like vampires essentially. Which pisses all of us off”

Microsoft was widely praised by the attendees for their contributions to open source, and it was AWS who came under fire for being closed.

“AWS, while I really admire Jeff Bezos”, Awadallah said, “when it comes to open source, they only take. They take and never give back. They’re like vampires essentially. Which pisses all of us off”.

CNBC reported earlier on October 12th that AWS had partnered with Microsoft and rolled out new tools that will make it easier for developers to use open-source artificial intelligence software.

It was also suggested that the AWS and Microsoft partnership had significantly excluded Google, and that they were “teaming up”.

Perhaps the titans of tech haven’t shaken off their competitive habits, but as Dvir suggested openness is the best way to encourage participation in developing AI.

Whether you choose Microsoft, Amazon, Google, or one of the other large AI cloud providers which include Oracle, IBM, and Alibaba, as the cloud solution for your business, you should expect to increasingly see the potential for mixing and combining their separate offerings.

If you would like to hear more from Yuval Dvir, Head of Global and Online Partnerships at Google Cloud, then you can meet him this Thursday (24th) in Spitalfields, London, at a free Access AI breakfast briefing, which you can register for here.

No comments: