VMware market share in 2026. Why most exits stay partial.

VMware still leads on installed base while new, SMB and edge deployments go elsewhere. Most organizations that costed an exit kept the bulk of their workloads, and the common 2026 outcome is two platforms.

Is VMware losing market share in 2026, and which number should you trust?

VMware is losing share on momentum and holding it on installed base, and most confusion comes from reading the two as one figure. Installed base counts where workloads run today, and VMware still leads it comfortably. Momentum counts where new workloads land, and that is now split four ways between Hyper-V, Nutanix, Proxmox and container platforms.

The gap between the two is widening. Broadcom closed the VMware acquisition on November 22, 2023, and every renewal since then has forced a platform question onto the calendar. Some customers answered it by leaving. More moved part of their workloads and kept the rest on VMware.

Why do the headline share figures describe the edges of the market?

In our engagements, small and midsize businesses and edge deployments left VMware at far higher rates than core enterprise datacenters, which matches the published share data. Aggregate numbers therefore describe the market's edges and say little about the large datacenters where most VMware workloads run.

Why does Broadcom accept customers leaving?

Broadcom rebuilt the portfolio around VMware Cloud Foundation (VCF) subscriptions and chose depth of revenue per retained customer over breadth of customer base. Churn among small customers is priced into that model, so a departure threat from a small customer carries less weight than many buyers assume.

The subscription terms themselves push small customers out. Broadcom's core counting rules require at least 16 physical cores to be licensed for each CPU, even when the CPU has fewer. Bundle minimums like this land hardest on SMB environments and branch sites, which is exactly where the share loss shows up.

Which VMware customers are leaving, and which are staying?

The customers leaving fastest are the ones the price model was never built to keep. Large enterprises are mostly tiering, and regulated core production is mostly staying put. The segments break down like this:

• Leaving fastest. SMB environments priced out by bundle minimums, edge sites, VDI deployments and cost sensitive public sector. These are the segments the share data mostly reflects.
• Moving partially. Large enterprises splitting workloads across two platforms. In 2026 this is the most common outcome of all.
• Staying put. Regulated core production, dense VCF deployments, and anyone in the middle of a hardware cycle.

Why does core production stay on VMware?

Operational gravity keeps it there, and that is a different thing from inertia. Runbooks, integrations, team skills and certified configurations build up over fifteen years. None of them port inside a single budget cycle, and each one has to be rebuilt, tested and signed off before a regulated workload can move.

The cost of that rebuild is what decides the outcome. Organizations that priced it properly found the migration case failed for core production, even when they were motivated to leave.

In what order do workloads leave VMware?

Most exit programs we saw moved workloads in the same order. Plan for that sequence from the start instead of drawing up a single cutover.

1. Test and development first. Low risk, few integrations, and a safe place to build skills on the new platform.
2. Edge and branch second. Small hosts that the per core minimum makes expensive, often with simple workloads.
3. Production last. Often never, once the rebuild cost is on paper.

How does the picture change with the size of the environment?

A company running a few dozen hosts, mostly Windows, with a hardware refresh due can often leave entirely. The minimums hit it hardest, and it has the least operational gravity. A global organization with hundreds of hosts, regulated workloads and VCF components in production almost always ends with two platforms, moving lighter tiers and renewing the core.

Where do workloads go when they leave VMware?

The destination follows the profile of the workload and the team that runs it. Windows heavy environments lean to Hyper-V, hyperconverged refresh buyers to Nutanix AHV, and cost driven SMB and edge buyers to Proxmox VE. Container forward teams consolidate on OpenShift Virtualization, and elastic workloads go to public cloud.

Hyper-V has a specific pull for Windows shops. Windows Server Datacenter already licenses unlimited virtual machines on a licensed host, so part of the target platform is paid for before the migration starts.

Should you believe alternative vendors' migration win claims?

Believe them selectively. Every alternative vendor claims record VMware migration wins, and each claim is true within its segment. The wins cluster where customers were already inclined to move: SMB, edge, VDI and cost sensitive public sector. No alternative has yet shown mass displacement of regulated core production at scale.

When do workloads actually move?

They move at hardware refresh and renewal dates, almost never mid cycle. The platform choice is a calendar decision, and the share shift tracks refresh cycles more closely than any press release. A migration decided away from a refresh or renewal has to fund itself twice, once for new hardware and once for the migration itself.

What have we seen in VMware renewals and exit evaluations since the acquisition?

The gap between migration talk and migration action defined the market. Across roughly 40 to 55 VMware renewals and exit evaluations we advised in 2024 and 2025, most customers that ran a serious exit evaluation still kept 60 to 80 percent of their workloads on VMware.

Those evaluations were not for show. They had real costing and board attention. The retained share stayed high because the migration case failed for core production once runbooks, integrations, certified configurations and skills were priced properly.

Why are the collapse story and the stability story both sales pitches?

One story says VMware is collapsing and every sensible organization is migrating now. Broadcom's version says the platform is stable and the churn is noise. The retention figure above settles it better than any share estimate, and it supports neither.

• Where the collapse story holds. SMB, edge, VDI and cost sensitive public sector, where bundle minimums price customers out.
• Where the stability story holds. Regulated core production and dense VCF deployments, where little has moved.

Whoever sells you an alternative quotes the first group, and whoever sells you a renewal quotes the second. Most large organizations sit in both, which is how two platforms became the norm.

Why a threat to leave VMware entirely works less well than a partial plan

The usual advice is to tell Broadcom you are leaving and let the threat do the work. We disagree, because in our engagements the price moved for a different reason. Account teams have learned to discount threats of a full departure, and a migration deck is cheap to produce and so easy to dismiss.

What moved the quote was a document showing which tiers cleared the three year cost case and which did not. Conceding what will stay is what made the rest believable.

An exit plan that was never fully executed still moved the price, because it was costed, board reviewed and honest about what would stay.

How does a partial exit plan change the VMware renewal price?

A partial plan changes the price because it is credible in both directions. It shows the account team which tiers you will move and proves it with a three year cost for each tier. The renewal arithmetic behind the quote is broken down in the 2026 cost breakdown.

A worked example: tiering a hypothetical 2,400 core environment

Say your renewal quote is $1.2 million a year for 2,400 licensed cores, or $500 per core per year. These figures are illustrative, not Broadcom list prices. The three year VMware cost is $3.6 million. You then cost each tier's alternative over the same three years, including migration labor, the target platform, support and retraining.

Two tiers clear. Moving them takes 720 cores, or 30 percent, off VMware and leaves 70 percent in place, inside the range we saw in practice. At the quoted rate the retained 1,680 cores cost $2,520,000 over three years. Add $640,000 for the two migrated tiers and the total is $3,160,000, a saving of $440,000.

The larger effect comes in the negotiation. If the price on the retained cores moved by the median we observed, it would fall to $325 per core. The retained cores would then cost $1,638,000 over three years, and the combined total would be $2,278,000.

What will the Broadcom account team say, and how should you answer?

• "Customers who tried to leave are coming back." Agree that core production is hard to move, then show the two tiers whose three year cost clears. The plan never claimed more.
• "This price is only valid until quarter end." Ask for the offer in writing with its validity date, and tie your decision date to the hardware refresh. The quarter end timing guide covers when pressure is real.
• "A smaller core count means a higher rate per core." Show the general production tier, which sits $80,000 short of clearing over three years. A rise of about $25 per core per year on those 1,080 cores is enough to push it across the line.
• "The partner cannot go any lower." Ask for Broadcom's own deal desk to join the conversation, since the partner rarely sets the price.

Which contract terms protect a two platform outcome?

• Right to reduce cores. Reduce the licensed count at renewal as tiers move, without the remaining cores being repriced.
• Renewal cap. A ceiling on the per core price for the retained cores in the next term.
• Term length. A term that ends near your next hardware refresh, so the next platform decision lands on a date you control.
• Host by host core schedule. The agreed count attached to the order, so the next renewal starts from your figure.
• Licensing by cluster. VCF where you run its components and vSphere Foundation where you do not.

When should you run a VMware exit evaluation?

Run it so that its conclusions are ready at a hardware refresh or a renewal, whichever comes first. Platform decisions get made at those dates, and the capital decision is already open. Work back from the date:

How do you check your own VMware position?

Start with Broadcom's own License Counting PowerCLI tool, published in its knowledge base article on counting cores for VCF and vSphere Foundation. It reports licensed cores per host after the 16 core minimum. Compare its output with your vCenter inventory or an RVTools export, and with the hardware register that holds each host's refresh date.

Then map applications to clusters. The tiering only works if you know which workloads are regulated, which carry certified configurations and which have owners willing to move. The wider research sits in the Broadcom knowledge hub.

What to do next

1. Separate the two measures in your own environment. List where workloads run today against where the next three years of new workloads will land.
2. Count your cores. Run the Broadcom counting tool and reconcile it with vCenter and your hardware register.
3. Tier by exit feasibility. Test and development, then edge, then production, with a three year cost for each tier.
4. Take the costed partial plan to the board. Board attention is what makes it credible in the negotiation.
5. Time the evaluation to a refresh or renewal. A decision made away from those dates has to fund itself twice.
6. Move only the tiers that clear. Keep the rest of the plan current as your position for the next cycle. The Broadcom practice costs the plan with you.

Frequently asked questions