The Next Cloud Battle Will be Fought “On-Prem”

Predictions Going into AWS re:Invent 2018

Scott Coleman, Business Development Partner

As the seventh annual AWS re:Invent conference approaches, I’ve been thinking about what may surprise us in the 2019 cloud wars. In my mind, the big surprise will not be about cloud leadership, since the current and probable future pecking order appears clear, nor is it about the tech these days; all deliver AI tools and a range of data services, containers, kubernetes, serverless, edge and IoT, and security/identity services. Rather, the next surprise is likely to be that the battle for cloud will be found “on prem.”

Questions For The Changing Enterprise Landscape

As incumbents reshape the enterprise landscape via M&A, IBM buying Redhat, VMware buying Heptio, and SAP buying Qualtrics, there’s been much debate and commentary. Were these deals representative of the buyers waving the white flag or doubling down; did they over pay or pay just right; and whether the technical and market advantages that the acquirers have established will be squandered? These are questions that can only be answered with due course of time.

For me, these deals raise the question of whether and how the battlefield for capturing enterprise workloads is shifting. A few items shape my view:

First, enterprise cloud adoption has been slower than anticipated five years ago. The headline numbers for AWS and Azure are nothing short of phenomenal but discussions with enterprises suggest that less than 5% of core applications have been migrated to the cloud (excluding SaaS, Gmail or O365) for most of them.

Second, enterprises are likely to operate over time in hybrid, multi-cloud environments. No CIO I’ve spoken with wants to single source a cloud vendor in the long term, which is often based on the lessons learned by single-sourcing servers from HP, networking gear from Cisco, and databases from Oracle.

In addition, CIOs increasingly understand data gravity and that where you put the bits is the vendor lock-in of the cloud era. Enterprises dual/multi-source is to get the right features at the right price, and this axiom is as true in cloud as it was in the prior compute eras.

The On-Prem Prediction

As a way to mitigate these elements, I believe we will see cloud vendors push hard to deliver on-prem instantiations of their clouds over the next 12 months as a way to create new on-ramps.

There are already signs of this occurring. Microsoft announced Azurestack in 2016 as an extension of Azure that lived “on prem” and provides controls and features consistent with Microsoft’s cloud. While the idea is excellent, execution has been lacking, imo. AWS announced Snowmobile in 2017 as a way to enable large scale data transfer (look out for lock in!) and more recently there have been a spate of press stories noting that it will offer a white box switch via its partner VMware. Google has long had intranet servers in enterprise data centers and we assume they could be fast followers here, especially with Thomas Kurian taking the helm (Oracle has nearly perfected vendor lock-in, so expect that playbook to come back out at Google).

With this in mind, expect AWS at re:Invent (as well as Microsoft @ Build and Google @ Cloud Next) to start announcing more and more on-prem offerings.

What does this have to do with IBM+Redhat and VMWare+Heptio?

To remain relevant, traditional vendors need to own either layers of the new stack or have a critical role in cloud on-ramps. Win the on-ramp, win the traffic. On-ramps matter greatly in the cloud world as a way to capture workloads and customers. AWS’s initial on-ramp was winning with startups; Microsoft’s on-ramps include O365 and now Github in the developer world. Google’s seems to have been limited to date to marketing and a subset of big data apps. SAP’s intent to acquire Qualtrics signals that the cloud wars will increasingly shift to business applications. Clearly Salesforce, SAP and Oracle have been active in the app space but perhaps we will see traditional cloud vendors skew closer to infra deal’s focus on apps that can consume chunky cloud cycles.

Owning the open source OS layer may breathe new life into IBM (if they execute, naturally) as this is the lingua franca for most enterprises. And it is beyond dispute that Kubernetes has won the orchestration battle and will be the de facto orchestration layer for enterprises. In acquiring Redhat, IBM gained a toehold in the new stack; VMware, which via its partnership with AWS has changed its fortune, has potentially added greatly to its ability to be Switzerland in the enterprise cloud wars.

What does this mean for other incumbents?

More single threaded vendors like Cisco, HP and Oracle will likely see their businesses erode over time, though I think it’s also fair to expect them to jump hard into the M&A game (Nutanix, Okta, Splunk, or a range of intelligent app startups anyone?).

And for startups?

Enterprises have long had a focus on managing tool proliferation and don’t want separate tools regardless of on-prem or cloud. Owning a layer that cuts across clouds — security, monitoring, logging, orchestration, network, data prep, for example — will put startups in the demand vein that enterprises desire.

What are your thoughts about the next move in cloud leadership? Catch me, or one of the other Ignition attendees, in Vegas next week to chat about it in person. @colemanscott1