First things first

Analyst firm Gartner released a report this week predicting that spending on cloud computing services will increase 21.8 percent in 2018, to $186.4 billion. The infrastructure as a service portion of that will grow at an even faster clip, up to $40.8 billion from $30 billion in 2017. In fact, Gartner's estimates have IaaS as the fastest-growing of cloud services, eventually reaching $83.5 billion by 2021.

It's not surprising that cloud usage and revenue are growing, but just how fast they continue to grow is notable. I find this particularly interesting in the context of the public fight happening right now among cloud providers vying for a multi-billion-dollar Pentagon contract -- a contract that Amazon Web Services' competitors say heavily favors that company winning the entire contract. (A collection of links to new stories on this is below.)

Even taking into account the bias inherent in comments from AWS competitors, I do think there's some logic in the United States not wanting a single point of failure for something as sensitive at defense data and workloads. Nor would I blame a cloud provider for not wanting to take on that much risk all by itself, despite the size of the deal. (But, I guess I also wouldn't blame a cloud provider for wanting that money, being confident it could handle the job, and making the argument that trying to integrate systems across multiple clouds could add unnecessary complexity.)

Here's the latest in that controversy:

On a broader scale, though, this whole situation begs the question of whether cloud computing contracts -- especially IaaS contracts -- should be viewed as a zero-sum game, where there's only room for one provider. At least from the customer perspective, there are plenty of good arguments that they shouldn't be, including an avoidance of lock-in and the benefits around cost, flexibility, latency, performance and resiliency that can come from decentralization.

Furthermore, with the ever-shifting currents of corporate muck-ups and public outrage -- and cloud-provider entries into new industries -- pressure to cut ties with a company like Amazon, Google or Microsoft could come from anywhere, and some companies might actually want to be in a position to do it.

(Speaking of Facebook, I thought these two stories -- "AI is an excuse for Facebook to keep messing up" and "Facebook uses artificial intelligence to predict your future actions for advertisers, says confidential document" do a good job capturing its awkward relationship with artificial intelligence and user data.)

I don't suspect large cloud providers will jump on the cooperation bandwagon anytime soon (although some smaller providers are) because there's just too much money at stake to go for the kill with every deal. But I do think we're going to see (well, more than we're already seeing) savvy buyers start forcing the issue themselves through clever architectures and tooling, like container orchestration platforms, that can abstract infrastructure and open the door for better third-party applications.

Because if there's one thing the Pentagon-contract controversy is underscoring, it's that cloud computing is no longer a boutique business or the alternative to cumbersome legacy IT processes. Cloud computing is IT, and the vendors leading that industry are larger, more powerful and more all-encompassing than anything we've seen in decades of enterprise computing. The pros and cons of this situation haven't been fully realized yet, but a little freedom of cloud choice might go a long way as they become more clear.


AI and machine learning

Cloud and infrastructure

Sponsor: Replicated

Data and analytics