Last year, the world was excited about the prospect of Docker lowering cloud complexity and removing the need to run virtual machines in the cloud. This year, however, the industry seemed to awaken to the need to be cross-cloud-capable to avoid vendor lock-in.

That industry-wide realization came at a time when Microsoft’s Azure was turning into not just an Amazon alternative, but a legitimate competitor. The company doubled its revenues on Azure this past year, but that doesn’t mean it was slowing down. In fact, Microsoft spent a billion dollars in 2016 alone on data centers in Europe to expand its cloud operations there.

Which isn’t to say Amazon is not still the leader in the cloud. That company’s web service offerings continued to expand in 2016. Among the new features were HIPAA compliance for more of its services, price reductions, and new features for Storage Gateway. And at the end of the year, support for desktop Windows 10 deployments. That last addition gave users access to desktop computer use from the cloud.

Amid all the wild news from Microsoft and Amazon, you’d expect Google to be racing alongside with its own big announcements. But Google spent most of 2016 preparing for a revising of sorts of the Cloud Platform as a whole. After acquiring her startup in 2015, Google installed VMware founder and former CEO Diane Greene as vice president of Google’s cloud business.

This hire was supplemented later in the year with the acquisition of Apigee and the hiring of former Apigee strategy chief Sam Ramji. The underlying shift in the platform may not actually be a shift in the platform itself, but rather is a major shift in the strategy behind bringing Google Cloud to market. The company has already made moves to keep the platform competitive, and as of this writing, Google Cloud is about half the price of Amazon for certain services. But we expect 2017 to be the year the dial begins to move on the somewhat left-behind Google Cloud.

No matter how Google does in 2017, it’s a safe bet that OpenStack continues to grow and mature. This open platform for data centers has grown significantly over the past few years. Today, it’s got an appealing position inside the enterprise as a bridge between the public and private clouds.

Interestingly enough, OpenStack’s original value proposition of offering internal developers the same on-demand cloud-like hosting capabilities available in AWS and Azure has now given way to a much more mundane use case: maintaining infrastructure portability.

Developers are now using OpenStack as a way to push their internal deployments, and are then bursting to external clouds when demand peaks. This ensures applications cannot simply be written to work in Amazon and nowhere else.

And that’s a major trend we expect to see continue in 2017: avoiding vendor lock-in. While Amazon has provided developers with some of the easiest and most powerful APIs and services inside its cloud, using them is tantamount to pouring concrete on your application’s feet and shoving it into the ocean: Once you’ve built with a specific cloud’s services, it’s unlikely you’ll be getting away from them any time soon.

That’s the one thing every software developer and manager learned to avoid back in the 1990s. Only now is the industry starting to see it is possible to build multi-cloud applications without having to invest in significant internal development to make it happen.