Well played, EMC.

EMC (s emc) might be smarter than we thought it was. You already know the news — Paul Maritz out as VMware CEO, in as EMC chief strategist. Pat Gelsinger, former EMC COO, now in as VMware CEO. The companies are planning to spin out their cloud computing and big data assets into a separate company. If EMC handles this transition well, the companies under its banner could do great things. Of course, there are a lot of moving parts here and the transition won’t be easy.

I was surprised, if not shocked, when I first heard the rumor of Gelsinger replacing Maritz as VMware CEO. Without details, the natural connotation of such a move is, “We’re firing you because we don’t think you’re doing a good job.”

Which was strange in this case, because most people bought into Maritz’s vision of an application-centric future whole-hog. The premise is simple and, more importantly, true: Servers, virtual machines, databases, operating systems — everything below the application — will become commodotized, leaving all the room for differentiation at the application layer.

It seems EMC bought into it, too, just maybe not as the ideal business for an infrastructure company such as VMware. Read between the lines of EMC CEO Joe Tucci’s new job descriptions for Maritz and Gelsinger:

Pat will now lead Cloud Infrastructure at VMware, and Paul will look across our technology strategy with a particular focus on Big Data and the next generation of cloud-oriented applications that will be built on top of these foundations.

The cloud computing and big data spin-out we reported Monday (or something very much like it) seems destined to happen, and Maritz will stay on board to ensure it stays true to his vision. Tucci has even given him the M&A and strategic investment reins. Well played, EMC. I think.

Give innovation room to roam

The cloud spin-out could do a lot of good by helping get EMC out of the way both technologically and culturally. As my colleague Stacey Higginbotham noted in covering the spin-out rumors on Monday, EMC’s insistence that VMware build Cloud Foundry on its pricey, absolutely-not-commodity storage gear rankled a lot of the team and might have helped catalyze the idea for a separate entity. That’s just not how successful cloud services are built.

There’s also a general feeling among the platform-as-a-service world that while Cloud Foundry is a great open-source project, VMware’s hosted version is just a ploy to sell more software licenses.

And as I explained in a post on Monday, VMware is losing (and seemingly not replacing) a lot of key talent that are leaving to free their innovative sides. Large companies under the thumb of even-larger companies have rarely (never?) been a breeding ground for innovation.

The proposed cloud spin-out could solve all these problems. Cloud Foundry, Greenplum and, it appears, VMware’s current suite of cloud-based applications, will be able operate as they see fit (i.e., initially building out large ecosystems of application developers) without having to worry about padding EMC’s or VMware’s bottom line and keeping shareholders happy.

Free from the large-enterprise constraints and operating in some white-hot areas, a spin-out might actually lure top talent and create some compelling products.

The infrastructure ball is in VMware’s hands

Of course, VMware and EMC stand to gain from this, too. If Maritz really did take his eye off the hypervisor ball to focus on applications, new CEO Gelsinger can put all his focus back on the infrastructure. VMware will still make a lot of money from its virtualization-management software — all that gear and all those virtual machines atop which applications run aren’t going to manage themselves — but it’s going to take some serious attention to find the right strategy for the future.

I would argue the answer to receding market share as more companies adopt the Microsoft (s msft) Hyper-V, Citrix (s ctxs) XenServer and KVM hypervisors for new workloads is to make vSphere entirely free and ship it on as many physical servers as it can. VMware CTO Steve Herrod’s software-defined data center vision is legit, and it’s where VMware has to make its infrastructure money as hypervisors become commodities. But when customer suspect they’re being bent over the barrel by a vendor trying to lock them in at every level and extract as much money as possible, they get understandably leery about buying anything at all.

Or maybe VMware envies Oracle (s orcl) and wants to go even further in that direction (although its DynamicOps acquisition suggests otherwise). Perhaps, as some Wells Fargo analysts suggested in a note this morning, the answer is for EMC to buy back VMware and bring it entirely back into the mothership. Whatever the case, the infrastructure business is in Gelsinger’s court now. He’s free to do what he thinks is best without having to worry about things like applications and cloud platforms.

Or is he …?

Products without homes

Some commentators have already questioned the wisdom of putting both an infrastructure-as-a-service cloud and a PaaS offering within the control of the proposed spin-out, and they might be right. According to my sources, Project Rubicon (the code name for VMware’s IaaS cloud) is already up and running and more closely resembles a hosted private cloud option than it does a commodity cloud such as Amazon Web Services. In fact, I’m told, it’s being headed up by former Terremark (s vz) executives Jason Lochhead and Randy Rowland.

That sounds a lot more like a Gelsinger business than it does a Maritz business. Even relatively recently, it sounds like VMware wasn’t sure how to position the service (don’t forget, it already has hundreds of vCloud Data Center partners) or where it should live.

SpringSource, too, seems like something of a motherless child. It’s not inherently cloudy and it’s a Java application platform that kind of lets VMware compete with Oracle and IBM in what’s still a big market. My assumption, though, given that it also comprises the data components of VMware’s current application-centric focus, is that if a spin-out happens, SpringSource goes with it.

At some level, everything EMC and VMware have announced in the past day and are rumored to be doing make sense. There’s the potential for a big business if an EMC-VMware spin-out can become the leader in helping developers build next-generation cloud and big data applications. There’s also  plenty of business yet to do managing companies’ virtualized resources. But there’s a lot of gray area between those two businesses, and navigating that will be the hard part.

Feature image courtesy of Shutterstock user nokhoog_buchachon.