Cloud AutoML highlights Google's ability to build open ecosystems and monetize them

Google has always made money from open ecosystems, but its cloud-based machine learning tool shows the lengths to which the company will go to foster those environments.

One cynical view of Google is that it's an advertising business that basically innovates new ways to get more people using the web, thereby viewing its ads. A more generous view is that Google has cracked the code on how to gain competitive advantage from open ecosystems. That is, Google generates significant benefit and is content to skim just a bit off the top to fuel further efforts.

It's hard to escape the fact that Google does indeed make massive bank from selling ads. It's equally inescapable, however, that en route to selling those ads Google does, in fact, generate massive value for consumers. With its cloud business, however, Google seems to be finding its way to a model that both makes money and happy customers, without selling a single ad. Nowhere is this clearer than Google Cloud AutoML, an algorithm for building machine learning models.

You win, I win

Some of us still remember buying a "yellow pages" for the internet. There were so few web pages initially that it was possible to catalog them all and sell a paper index to the world not-so-wide web.

As the websites boomed, it became harder and harder to discover new ones, or even keep track of changes on the old ones. Search engines were born. I still remember my default, called Metacrawler, because it allowed me to employ several search engines at once. At some point, however, Google became the true metacrawler, helping us find our way through the labrynthian web, collecting a small advertising tax along the way.

It was a genius move, and it made the web as we know it possible.

SEE: Internet and Email usage policy (Tech Pro Research)

The tool for using the web—the web browser—started to buckle under the weight of Microsoft's Internet Explorer. As the risk of one company effectively gating access to the web grew, Google responded with Chrome. While the irony is that Chrome now threatens that same domination of our online behavior, over the years it has largely had a positive influence on the growth of the web. Along the way, Google has profited from more web use.

Later, this same scenario played out in mobile, with one company (Apple) growing so dominant that competition got choked. Google introduced the open source Android operating system and, as ever, collected some coins for its troubles.

Meanwhile, much the same strategy has played out in the enterprise. In response to growing interest in cloud computing, coupled with growing risk of a single company dominating that market, Google open sourced Kubernetes, essentially giving away the keys to its container-driven kingdom, thereby fostering an open cloud ecosystem.

As machine learning (ML) increased in importance, Google open sourced Tensorflow and Kubeflow to open up an ML ecosystem. The hope is that, as enterprises get cozy with Kubernetes or Tensorflow, they will turn to Google Cloud Platform to run more of their workloads.

Helping companies run like Google

It's an interesting strategy, and it's just that: A strategy. Google isn't a charity. It doesn't use open source to foster greater love and harmony in the universe. It's very much a capitalist tool for a capitalist company, one that happens to create a lot of free-of-charge consumer and enterprise value along the way. In fact, that value to end users must come first for Google to be able to monetize the abundance it helps to create.

Google Cloud AutoML is the latest instantiation of Google's open ecosystem strategy, and it's a signal that Google's cloud ambitions are finally maturing. That is, while Google has been a great place to "figure it out yourself," Cloud AutoML is Google's recognition that most of us are mere mortals and need a lot of help to crack the ML code.

SEE: Special report: How to implement AI and machine learning (free PDF) (TechRepublic)

As 451 Research analyst Nick Patience wrote: "Right now, a lot of knowledge is required to build custom models, and given there are only roughly one million data scientists in the world (compared to about 26 million developers of all kinds), many companies, including Google, see a major opportunity in enabling developers other than data scientists to be able to get their hands dirty with machine learning."

To lower the bar to machine learning adoption, Google has made the interface more drag-and-drop, less command line.

In addition, rather than forcing enterprises to train their algorithms using Google's data, Cloud AutoML ingests enterprise data assets and tunes the model accordingly. The key here is that Google helps enterprises to customize a model without having to do so de novo: There's already a great deal of training baked in. Though initially focused on image data, Google plans to roll out the service to tackle text, video, and more.

Google isn't alone in this ambition to enable developers to do gee-whiz things with data. AWS has SageMaker (as well as a suite of other AI/ML services), and Microsoft has Azure machine learning, but no one has quite what Cloud ML offers today. I doubt that lead will last, but I'm also not sure it has to: Google has a long history of figuring out how to build open ecosystems and monetize them, with Cloud AutoML demonstrating a newfound ability to make complex technology straightforward for mainstream enterprise adoption.

Also see

Image: iStockphoto/monsitj

About Matt Asay

Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.

Editor's Picks

Free Newsletters, In your Inbox