Estates that ran a serious exit evaluation still kept 60 to 80 percent of workloads on VMware

Two years after Broadcom closed the acquisition, installed base and momentum point in different directions. VMware still leads where workloads sit today, while net new, SMB, and edge deployments move elsewhere. Reading those as one number is what produces both of the sales pitches on offer.

Prepared by Redress Compliance · August 16, 2026 · Broadcom VMware advisory. 47 renewals and exit evaluations advised, 2024 to 2025.

Churn concentrates at the edges: SMB estates priced out by bundle minimums, edge and branch deployments, and discrete workload tiers inside large estates. Core production clusters in regulated enterprises move last, if at all.

• Leaving fastest: SMB estates priced out by bundle minimums, edge sites, VDI estates, and cost sensitive public sector. These are the segments the share data actually reflects.
• Moving partially: large enterprises tiering workloads across two platforms, which is the most common 2026 outcome rather than an exception to it.
• Staying put: regulated core production, dense VCF estates, and anyone mid hardware cycle, where the operational gravity is heaviest.
• Operational gravity is the mechanism, not inertia. Runbooks, integrations, skills, and certified configurations accumulate over fifteen years and do not port inside a budget cycle.
• Most exit programs moved tiers in a fixed order: test and development first, edge second, production last and often never. Plan the sequence that way rather than as a single cutover.
• Time the evaluation to the refresh, because a migration decision made away from a hardware or renewal moment has to fund itself twice.

The common narrative says VMware is collapsing and every sensible estate is migrating now. Broadcom's counter narrative says the platform is stable and the churn is noise. We disagree with both, and the number that settles it is the retention figure rather than any share estimate. Across the 40 to 55 VMware engagements advised in 2024 and 2025, most estates that ran serious exit evaluations, with real costing and real board attention, still kept 60 to 80 percent of workloads on VMware. Not because the evaluation was insincere, but because the migration math failed for core production once runbooks, integrations, certified configurations, and skills were priced honestly.

That figure explains something that otherwise looks contradictory. A costed exit plan moved renewal quotes by 20 to 50 percent, with a median around 35, and it moved them in engagements where most of the plan was never executed. The leverage did not come from the threat of a full departure, which account teams have learned to discount, and it did not come from a migration deck, which is cheap to produce and therefore cheap to dismiss. It came from a document that showed exactly which tiers cleared the three year math and which did not, because that document is credible in both directions. It concedes what will stay, which is what makes the part that would move believable.

This is also why the two narratives are both saleable. The collapse story is true of SMB, edge, VDI, and cost sensitive public sector, where bundle minimums genuinely price estates out. The stability story is true of regulated core production and dense VCF estates, where nothing has moved and little will. Anyone selling you an alternative quotes the first segment; anyone selling you a renewal quotes the second. Neither is describing your estate unless your estate happens to sit entirely in one of them, and most large estates sit in both, which is precisely how the two platform outcome became the norm.

So the buyer side move is to build the credible partial exit plan, use it to negotiate the renewal hard, and execute only the tiers where the three year math genuinely clears. Bank the rest as leverage for the next cycle rather than treating an unexecuted plan as a failed one. Time the evaluation to a hardware refresh, since that is when platform decisions actually get made. The renewal arithmetic sits in the 2026 cost breakdown, the alternative comparison in VMware against Nutanix, and the wider library in the Broadcom practice.

Across roughly 40 to 55 VMware renewals and exit evaluations advised, the gap between migration talk and migration action was the defining feature of the market:

The third pattern matched the published share data. SMB and edge estates departed at far higher rates than core enterprise datacenters, which is why aggregate share numbers describe the market's edges rather than its centre. Installed base measures where workloads sit today and VMware leads that comfortably; momentum measures where new ones land and that splits four ways.

Broadcom consolidated the portfolio around VMware Cloud Foundation subscriptions, trading breadth of customer base for depth of revenue per retained customer. Churn at the edges is therefore part of the model rather than evidence against it, which is worth knowing before you assume a departure threat carries weight it does not.

Watch the briefing · 4:135 Tactics That Move the Number in a Broadcom NegotiationWhat actually shifts a Broadcom quote once the exit plan is costed and on the table.

Is VMware losing market share in 2026?

On momentum, yes. On installed base, not materially. VMware remains the largest enterprise server virtualization platform by installed base, with the heaviest concentration in large regulated estates. Net new deployments, SMB, and edge workloads are where share is shifting, and that split is widening.

How many estates actually leave VMware entirely?

Few. Across the engagements we advised, most estates that ran serious exit evaluations kept 60 to 80 percent of workloads on VMware. Full exits remain the minority, and the most common 2026 outcome is a two platform estate rather than a migration.

Why are full exits so rare?

Operational gravity. Runbooks, integrations, skills, and certified configurations accumulate over fifteen years and do not port inside a budget cycle. Most exit programs moved tiers instead: test and development first, edge second, production last and often never.

Where do departing workloads go?

By estate profile. Windows heavy estates favor Hyper V, hyperconverged refresh buyers favor Nutanix AHV, cost driven and SMB estates favor Proxmox VE, container forward organizations consolidate on OpenShift Virtualization, and elastic workloads go to public cloud.

Does Broadcom care about the churn?

Not at the edges. Broadcom consolidated around VMware Cloud Foundation subscriptions, trading breadth of customer base for depth of revenue per retained customer. Departures among SMB and edge estates are part of that model rather than evidence against it.

Does an exit plan still move the renewal?

Yes, materially. Renewal quotes moved 20 to 50 percent, median 35, when a costed and board reviewed exit plan was on the table. The leverage comes from a document that shows which tiers clear the three year math and which do not, because conceding what will stay is what makes the rest credible.

Should we believe alternative vendors' migration win claims?

Selectively. Every alternative claims record VMware wins and each claim is selectively true, because wins concentrate in the segments already predisposed to move. No alternative has demonstrated mass displacement of regulated core production at scale.

When do platform decisions actually get made?

At hardware refresh and renewal moments, almost never mid cycle. The platform decision is a calendar decision, and the share shift tracks refresh cycles rather than announcements, so time the evaluation to a moment when the capital decision is open anyway.