Sitting in the cavernous general session at AWS re:Invent 2017 in Las Vegas, it’s easy to imagine that 2018 will see more of the same AWS dominance as the last 10 years have. The truth is more nuanced, however.

Gartner currently pegs AWS’ market share at 44% of the Infrastructure as a Service (IaaS) market, with Microsoft Azure grasping at 7% in second place. That’s not likely to change much in 2018. What has already begun to change, however, is how enterprises think about the cloud. With containers becoming ever more mainstream, and Kubernetes there to enable orchestration thereof, more enterprises will start to tinker with moving applications between clouds. Though this won’t kill AWS domination, it will introduce real choice into public cloud adoption.

Why would you leave?

The irony, of course, is that choice has always been the primary driver for AWS, specifically, and public cloud, generally. Developers, disappointed in the thin gruel IT kept serving them, took their credit cards and bought infrastructure elsewhere. If anything, developers in 2018 will find even more reasons to open their wallets to AWS.

But not because they must. After all, given the rise of Kubernetes, developers no longer need to feel constrained by where their application starts. As Brian Gracely, director of Red Hat OpenShift product strategy, has pointed out, “Containers and Kubernetes have made hybrid cloud and application portability a reality.” Indeed, new 451 Research survey data suggests that 69% of enterprises will be running multi-cloud workloads by 2019:

Thanks to platform abstractions and portable APIs, it’s never been easier to embrace multiple clouds and move workloads between them.

Sort of.

SEE: Hybrid cloud 2017: Deployment, drivers, strategies, and value (Tech Pro Research)

As Rishidot analyst Krishnan Subramanian told me, “In 2018, multi-cloud will gain more traction and, at the same time, infrastructure will become less relevant.” That is, basic infrastructure services like compute and storage.

This, however, is no longer where the game is played. As Subramanian goes on to say, “The applications will be deployed on the clouds based on the service dependencies of the application rather than any other factors.” That is, if your application calls AWS Lambda functions, you’re going to run that part of the application on AWS. If it needs a service specific to Google or Microsoft Azure, that part of the application will live on those clouds.

Which brings us back to portability, or the lack thereof. While it’s certainly true that containers give enterprises real portability between clouds, what they do not do is eliminate the lock-in associated with cloud-specific services. AWS, Microsoft Azure, and Google Cloud will continue to drive the bulk of enterprise innovation in 2018, just as they did in 2017, as Bernard Golden has highlighted:

[AWS, Microsoft, and Google] are the new pacesetters for IT. All of the growth in the tech industry resides with them, and they will increasingly define what is meant by IT….Simply stated, they have capabilities that no other user or tech vendor on the planet has, and those capabilities can be leveraged to deliver innovation unobtainable anywhere else.

In 2018, we should expect to see this “AMG” innovation hit overdrive as these companies compete to own IT dollars transitioning to the cloud. In this way, while containers give us more portability between clouds, cloud-specific services will somewhat blunt that potential portability. Given that AWS continues to drive the biggest portfolio of services, it will keep the cloud crown in 2018. Container-driven portability won’t change that.

SEE: The future of Everything as a Service (free PDF) (ZDNet/TechRepublic special report)

Meanwhile, beyond containers…

Containers and application portability are not the entire story for 2018. Another huge cloud issue in the coming year will be the relentless push toward the edge. In the words of Subramanian,

Data gravity and the need for seamless user experience will drive this push towards the edge. As IoT and technologies like self driving cars become a reality, the edge will play an important role. Unlike previous efforts to do computing at the edge, this iteration will see even higher order services like Serverless operating from the edge and working with the central cloud.

AWS was early to market with such edge services, including Cloudfront and Greengrass, but Microsoft followed suit with Azure IoT Edge, and Google has its own portfolio, including Google Cloud IoT. Across each of these different approaches, the idea is to push intelligence to the device at the edge, rather than relying on costly bandwidth to move data back and forth between a central repository and dumb nodes. In 2018 we’ll see the cloud, hitherto conceptualized as a central brain in the sky, turn into omnipresent services that run where they’re most needed.

Just like those central services like Lambda, however, the more a developer ties herself to cloud-specific services like Google Cloud IoT, the less likely she’s going to be able to port that application anywhere else. In sum, 2018 promises an unprecedented level of choice… and an unprecedented unlikelihood that you’ll actually be able to exercise it.