Back in 2016, we suggested that the IT industry had reached a point of saturation in server virtualisation. In a post entitled “Peak VMware”, we looked at the announcement of VMC, the first instantiation of VMware software on the public cloud. Six years later, Broadcom looks to be set to acquire VMware for $61 billion, plus some debt. Does this point to the end of server virtualisation growth, and if so, what becomes of VMware as a company?
Although I’d already been using VMware on the desktop, the first time I saw VMware products in an enterprise setting was at Lehman Brothers in 2004. My first (perhaps facetious) question was to ask whether VMware had permission from IBM to use the term “VM”. After all, IBM effectively invented virtualisation back in 1972 with a product called VM/370. It’s incredible to think that technology is now 50 years old.
However, the VMware solution was different to IBM’s mainframe virtualisation because it abstracted x86-based operating systems, initially in software and then through hardware extensions like Intel’s VT-x and AMD-V. This enhancement put VMware right into the heart of the data centre, virtualising Windows and Linux systems to consolidate inefficient single-server applications.
Through a wonderful piece of serendipity, the advancement of Moore’s Law directed the processor manufacturers to build out multi-core CPUs, aligning perfectly with a technology capable of multi-tasking many virtual operating systems across a processor complex. This architecture aligns very much with the mainframe, where VM/SP and successors were able to virtualise many mainframe operating systems at the same time (this was also possible directly in hardware with features like PR/SM).
Standardisation is a significant driver of IT practices, and server virtualisation offered the ability to optimise on two operating systems – Windows and Linux. Low-cost and low-power rack-mounted servers quickly resulted in massive growth for VMware, while “big iron” solutions from Sun, IBM and HP began to fade away. Today, vSphere and ESXi are the cornerstones of any on-premises technology deployments.
Of course, VMware didn’t become the de facto standard for server virtualisation merely on consolidation benefits alone. Technologies like vMotion and Storage vMotion enabled high availability, reduced maintenance, and allowed businesses to eliminate many legacy practices like forklift storage upgrades. The golden age for VMware was surely the early 2010s, with enhanced security and software-defined networking adding to an expanding ecosystem of products and services.
We should remember that throughout this period of growth, VMware wasn’t an independent company but was owned by EMC, which had acquired VMware in 2003/4 for $635 million. The company continued to operate independently but formed part of the EMC “Federation”, including the core storage products (EMC-II), Pivotal, RSA, Virtustream and VCE. Joe Tucci played a masterstroke by allowing VMware to operate independently, including working with companies that could be considered competitors.
In 2015, when Dell Inc announced plans to acquire EMC, we looked at the structure of the Federation, which showed that the core EMC business and VMware delivered around 95% of the revenue. With an 87% gross margin, VMware was the jewel in the crown of the EMC acquisition. We thought that Dell (the man and the company) would see VMware as the tip of the spear, leading the company forward.
This leads us to 2016 and our “Peak VMware” article and this conclusion:
So I think we have reached or are about to reach Peak VMware. Without a major pivot, VMware will be resigned to being a vendor of legacy software. Mr Dell and Co will continue to get good cashflow for a number of years, however VMware won’t be a growth business for the Dell empire unless something changes. That change perhaps is to accept server virtualisation isn’t the only route forward and to look to container-based future revenue streams. That would mean acquiring a company like Docker or CoreOS. Making a move like that would be radical, however I’m not sure VMware has the mindset to achieve it. It would be nice to be proved wrong though.Peak VMware
Why was this the inevitable outcome? First, all of VMware’s offerings centre around server virtualisation and the vSphere ecosystem. This is still true today, where Tanzu has been built into ESXi, delivering containers within a virtualisation framework. VMware chose to integrate rather than build out a separate product line or family.
Second, the public cloud has gained most of the growth in IT over the last decade. Increasingly, that focus is directed towards containerisation and has been accelerated by the popularity of Kubernetes. In parallel, cloud-native and open-source software has made great advances.
VMware did make a minor pivot with the introduction of the VMware platform to all the major cloud providers. This development was a significantly better approach than trying to compete directly with the public cloud (through vCloud Air). However, this direction still feels like a stopgap for customers that will eventually move to cloud-native applications (more on that in a moment).
Another factor that must be considered is the degree of complexity and overhead VMware now introduces. Initially, Tanzu required quite a significant amount of resources (and additional software) to implement. Even the Community Edition wasn’t that straightforward to deploy. Complexity translates to additional costs in hardware, licensing and skills. At the same time, Kubernetes has become trivially easy to deploy in the public cloud and relatively simple on-premises.
So now we’re at the heart of the problem for VMware. The company has continued to develop products and services focused on the core virtualisation technology. VMC and the other public cloud solutions are short-term fixes for customers that don’t want to operate data centres. There is no benefit to be gained by building out a competing public cloud infrastructure other than to partner with Equinix and other vendors looking to implement small-scale private clouds.
Elsewhere, the focus has been to use containerisation within the core products. This might work for some customers, but I doubt the majority will want to wrap complexity around a solution that can easily be implemented with a few physical servers or virtual instances.
Effectively the big pivot we suggested was needed in 2016 hasn’t materialised.
We should remember that VMware is still a large and profitable business. At this point in time, the company is not in decline but does appear to be moving from a growth to a maintenance phase. As we highlighted in 2016, Dell (the man and company) benefited significantly from dividends that paid down debt and justified the giveaway of VMware to Dell shareholders in November 2021.
The Broadcom acquisition aligns nicely with the strategy applied by CEO Hock Tan over recent years. In 2018, Broadcom acquired CA Technologies, picking up Symantec (or at least part of the company) in 2019. CA owns many software assets, including the UCCEL products I used in the late 1980s and 1990s. Software solutions like CA-1, CA-7 and CA-11 seemed legacy when I was using them 30 years ago, but they still exist and continue to return revenue.
The general industry consensus believes that Broadcom will slash VMware costs through reductions in R&D and redundancies, following a pattern seen with the CA and Symantec acquisitions. This approach is no secret and is part of the acquisition announcement. VMware made around $4.7 billion in earnings (EBITA), which Broadcom expects to reach $8.5 billion within three years. Either there are new revenue streams coming online, or costs will be cut dramatically to achieve this.
VMware will become the Broadcom Software Group following the acquisition, with a merged portfolio of existing and VMware solutions. So much of the increase in revenue will be from savings made by consolidation.
Can we expect another business to step in and outbid Broadcom? We can’t see who that could be. If future application growth is coming from the public cloud and/or containerisation, acquiring VMware could only be for either ongoing revenue (the Broadcom model) or to protect an existing business. On-premises vendors such as IBM and HPE already partner with VMware, so don’t need to acquire (they’re no different in outlook than Dell Technologies). At a lower level, perhaps Intel or Cisco might be interested, but this seems a long shot, as neither business is compatible nor has the “synergies” offered by Broadcom.
Ultimately, VMware will transition to a maintenance business while innovation occurs elsewhere. We will see a long-tail transition from server virtualisation, which will likely never be fully complete, just like the maintenance business for IBM mainframe.
The Architect’s View®
The information technology industry never stands still. IBM dominated the landscape in the 1960s and 1970s, but by the 1980s, new upstarts like Sun Microsystems were changing the landscape with smaller, nimbler, and cheaper solutions. Even this market was disrupted by cheap rack-mounted servers, Linux and, of course, server virtualisation. Now VMware is being disrupted by containerisation and Kubernetes, driven by the ease of implementation in the public cloud. Server virtualisation has become ubiquitous to the point that it’s just part of the furniture.
Meanwhile, the public cloud has innovated more quickly, with hardware-based virtualisation and processor diversification. Businesses can translate infrastructure costs directly to applications while scaling up and down on demand. VMware hasn’t innovated fast enough and now pays the price through an acquisition that can’t be denied. After all, shareholder value is paramount.
That leaves us with one last thought. VMware has been an integral part of so many IT professionals’ lives. As a result, the news of VMware being acquired has been met with shock and disappointment. I can empathise with those worried about job security but remember that VMware had a great run, and there are many other IT vendors out there to choose from. IT professionals are like Madonna and must reinvent themselves many times over the course of a 40-year career. The VMware acquisition represents just another one of those inflection points.
Post #ebc1. Copyright (c) 2007-2022 Brookend Ltd. No reproduction in whole or part without permission. The Architect’s View is a registered trademark of Brookend Ltd.