VMware is about to lose its status as the undoubted leader of the virtualization market after a 20-year reign – a result it doesn't mind one bit, but which also signals an industry-wide end to significant innovation for traditional server virtualization.
The end of VMware's reign will come slowly, then suddenly, between now and October 2027, a period that includes three dates that tens of thousands of VMware customers will have circled on their calendars to remind them of ideal deadlines by which they should change virtualization platforms or reduce their use of VMware. And at VMware's annual user conference, which kicks off today, the Broadcom business unit probably won't make a new pitch for them to stay.
The VMware users thinking about an exit mostly relied on the virtualization pioneer for the vSphere and vCenter products that together let them virtualize and manage a modest fleet of servers. Broadcom doesn't sell those products anymore, other than as components in VMware Cloud Foundation (VCF), a bundle of compute, storage, and networking virtualization tools that together assemble a private cloud.
Broadcom insists it sells VCF for less than pre-acquisition VMware did, but the bundle is nearly always considerably more expensive than a VMware user's last bill – and for many that means paying for VCF even when they don't plan to use all of its components. Many VMware users are therefore looking to reduce their VMware footprint so that if they acquire VCF, they can shift VMs that don't need it to other platforms.
Before acquisition, VMware had about 350,000 customers, and more than half of the server virtualization market. Broadcom has all-but-said it is only interested in 10,000 to 30,000 of those customers who need VCF and will go all-in on it as their future infrastructure and DevOps platform.
In case any VMware users haven't got the message, Broadcom's position will become abundantly clear on three imminent dates.
The first is November 22, 2026 – the third anniversary of Broadcom's acquisition of VMware. Because Broadcom has a reputation for making big and hard-to-digest changes after buying software companies, smart VMware customers rushed to acquire multi-year subscriptions before the acquisition closed. Many of those deals will expire soon.
The second date is March 31, 2027, on which VMware's contracts with many members of its Cloud Service Provider (VCSP) program expire, leaving them unable to provide cloudy VMs powered by VCF. Customers who rely on VMware-powered clouds therefore need to find a new home, almost always after buying a VCF license direct from Broadcom. Plenty of former partners are trying to bring their clients with them to another platform.
The last date is October 11, 2027, the last day of support for version 8 of VMware Cloud Foundation. Again, this is a moment that will force VMware users to make a decision about their virtual estates – do they upgrade to VCF 9, or look elsewhere?
That many users are either considering a move or trying to avoid a new VMware purchase is not in dispute. Third party support outfit Rimini Street tells me its VMware support offering is currently its fastest-growing service, albeit off a low base.
Cisco created its own hypervisor to give its customers an alternative.
Adam Centorrino, CEO of Australian services outfit Centorrino, moved from VMware to SUSE before the March 2027 deadline that unilaterally ends his VMware partnership, and is trying to bring his clients with him. He told The Registerhe is not angry with VMware and Broadcom, because he understands the decision to cut him from the partner program is a dispassionate business decision. But he is baffled by Broadcom's strategy because his clientele includes substantial government agencies that other vendors would consider a prize.
And then there are the high profile departures like Tesco and Allstate who are decidedly angry with Broadcom – so much so that they'll fight it out in court.
Whatever the motivation for quitting VMware, users know they won't find a better virtualization platform.
VMware's rivals admit this. I've spoken to plenty of them in recent months– Red Hat, Acronis, Sangfor, Nutanix, SUSE, and more – and all admit their products can't completely match VMware's, and that their teams don't include people who can match Virtzilla's corps of virtualization wizards.
At the lower end of the market, VMware's rivals aspire to assure buyers that their platforms are good enough for essential server virtualization, and their businesses are solid enough to stick around for the long haul. I've met tiny vendors who have put a pretty front end on Linux KVM, quickly found enough VMware refugees to sustain a 10 or 20 person server virtualization business, but don't have huge ambitions. The likes of Acronis and Parallels are trying to win over former VMware partners and provide them with a way to offer cloudy VMs to rent for VMware quitters.
Sangfor is almost alone in having a memory tiering offering to rival VMware's and is keen to point that out, but less enthusiastic about discussing its Chinese origins and any objections that some buyers might raise.
At the higher end, vendors promote their Kubernetes distributions, because they see virtualization as a mature technology that won't benefit from innovation other than making sure it can handle the latest hot workload, which is currently AI. Big players like SUSE, Nutanix, and Red Hat therefore position themselves as ideal for future workloads, and thoroughly competent at hosting existing virtualized applications.
That argument is going quite well. Nutanix is winning hundreds of former VMware customers each quarter. Red Hat has won over $680 million in virtualization orders from a standing start. HPE has also entered the market with its VM Essentials product, which as the name implies is closer to vSphere than a private cloud. The company reports "high double-digit new logos growth" for the product.
Proxmox is the challenger brand. Its platform handles basic server virtualization comfortably, the project is pushing into bigger datacenters thanks to Kubernetes integration, and admins get genuinely enthusiastic after using its wares.
All of the abovementioned vendors are fighting for, and will pick up, hundreds or thousands of mostly small VMware customers. Nutanix will also go toe-to-toe with VMware for big buyers, and peel some away. So will Red Hat, which I'm told is the only rival VMware truly fears because it is perceived as the most mature platform for Kubernetes and now also a competent platform to host VMs.
VMware will emerge with most of the customers it wants, and perhaps as the single largest virtualization vendor by customer count. But more users will rely on rival tech – probably KVM in one form or another – than will rely on ESXi. VMware will therefore emerge working with most of the most lucrative users – fulfilling Broadcom's ambition for the brand. But in many of those users, VMware will be just one provider of infrastructure software, rather than owning an account outright. And to keep the parts of a customer's business it retains, VMware will need to do more with modern and cloud-native apps, not just VMs.
The Broadcom business unit is well-positioned to succeed, because Nvidia's core software products are containerised and agentic AI workloads will benefit from running inside a lightweight VM.
I suspect that VMware will therefore not announce a significant strategy change at its conference this week. Further enhancements to make VCF a stronger container and AI platform are likely, and the Broadcom business unit will pitch its memory tiering prowess as the antidote to high hardware prices – and perhaps sweeten the deal by making it easier to run its wares on either older hardware or hyperscale clouds' Arm-powered servers.
VMware will also let the three dates mentioned above pass without mention. Even though it knows its strategy will give rivals a chance to win some smaller customers, and pieces of business at others, it also knows none will ever offer them a truly better way to run, manage, or protect their VMs.
(Score: 3, Insightful) by Anonymous Coward on Wednesday September 02, @10:53PM (3 children)
1. If you're still using VMware, it's budgeted, and it will stay budgeted, and those users will continue using it. It's been _years_, everyone who's using it now is going to continue using it. Some will go out of business. There will be no new users. The population will continually dwindle.
2. Broadcom bought it to hurt the long-term, can't-escape people. Those people who didn't escape before the end-of-year support renewal are exactly what they were after. Especially, large Fortune-500 co's, governments, and the like.
3. Broadcom has already made back its purchase price?
4. VMWare was undisputably the most performant hypervisor -- Xen, KVM, qemu don't even come close to the performance that it offered. It's really sad to see it go. (My most recent VM had 34% cpu usage *while idle*, and constant stuttering etc when trying to use it. It's really sad to see VMware gone.)
5. the "Workstation" version is free, even for commercial use. That won't get ESXi installed anyplace though -- can't afford it. Comparably, Virtualbox is *not* free for commercial use. At least there's a viable contender to Hyper-V (never used it).
(Score: 2, Insightful) by shrewdsheep on Thursday September 03, @08:33AM
I wonder where performance differences come from? There is hardware support for virtualization which levels the playing field already meaning that virtualizers start form the same interrupt to emulate/arbitrate the hardware. VMware then seems to have highly optimized this process, I simply can't see how this can explain large performance differences. Does VMware patch the code to directly access hardware in some cases?
(Score: 5, Insightful) by ledow on Thursday September 03, @11:47AM
As the article itself links:
https://www.theregister.com/virtualization/2026/06/17/tesco-is-sprinting-to-quit-vmware-and-broadcom-despite-rapid-migration-risks/5256133 [theregister.com]
The UK Tesco's chain of stores were utterly reliant on it. They are BIG.
It's a 107 year-old company and "In 2023, it was the 15th largest retailer in the world by revenue"
At the moment, not only did they transition their entire services to use something else very quickly, but they are also suing Broadcom for breach of contract over the changes to the licensing they were initially sold.
So... no... those "long-term, can't-escape" people? They absolutely can and will just walk away, and sue your arse on the way out, demanding their money back that they paid many years ago.
Broadcom are doing the single, dumbest thing you could do to every one of their customers, and they're all just going to walk unless they're absolute idiots.
(Score: 2) by tbuskey on Saturday September 05, @02:43PM
If your workload can't be converted to a 12 factory app, you probably are stuck. This is an event to push refactoring.
VMs are mainly used to keep doing pets instead of cattle. VMware let you keep running your pet through hardware changes. If you have a gitops workflow, you can recreate your VMs on another platform, but I bet most companies that replaced expensive individual hardware with cheapinfividual VMs decades ago just kept the same practices.
(Score: 5, Insightful) by Zoot on Thursday September 03, @04:35AM
...it's pronounced "Computer Associates".
In the 1990s CA (acquired by Broadcom in 2018) had a reputation for evil, buying up successful software companies, especially ones whose products were expensive and deeply integrated into big companies, firing virtually all the employees, jacking up support and maintenance fees, and basically squeezing all the blood out of the customer base.
One substantial software company I was familiar with was down to like one original employee, an old developer with advanced kidney disease who would not live long if he lost that job. He was the only person working (like three days a week) in a large otherwise empty building in Alameda which had no lights on. CA hired a small five-person company to help support and "maintain" the software until all the customers went somewhere else.
According to Wikipedia, CA somewhat rejuvenated their image in the 2010s though whether that was just better PR I don't know. It feels like Broadcom looked at the old 90's business model and said "we gotta get ourselves some of that" so they acquired CA. The VMWare purchase looks like a classic 90's CA operation. Buy something great, jack up the prices, shed overhead and development costs, and rake in MASSIVE profits at least for the short term.
They'll likely continue to maintain the current level of functionality (at least core functionality) as long as there are big customers rich or dumb enough to keep paying for it in exchange for not needing to figure out how to replace it. Maybe for those customers it will still be the best option for them.
(Score: 0) by Anonymous Coward on Saturday September 05, @08:17PM
I migrated 30 servers from licensed VMware to ProxMox over a weekend. It requires a bit more of effort, of course, but like Slackware, it simply works. F*ck off Broadcom.