VMware market share, two years after Broadcom.

VMware's market share — once the unquestioned default for enterprise virtualisation — is no longer the static line it was for a decade. Two years on from Broadcom's acquisition, the data shows clear share erosion in specific segments and equally clear share retention in others. Reading the pattern matters: it tells you which migration narratives are real and which are vendor talking points.

This piece synthesises what is publicly known from market research, customer announcements, hyperscaler disclosures, and our own engagement data. It is not an argument for or against staying on VMware; it is a structured look at where customers are actually moving, where they are not, and what the data implies for IT teams planning 2026 capacity decisions.

The baseline: where VMware market share stood at acquisition

At the November 2023 acquisition close, VMware held what most analyst houses described as roughly 70%–80% of x86 server virtualisation by deployed footprint, with vSphere as the dominant hypervisor and a long tail of ancillary products — vSAN, NSX, Aria, Horizon, Tanzu — bolted to the same install base. The competitive position was extraordinary: Hyper-V held the second-largest share, KVM-based stacks (Red Hat, Nutanix) held smaller slices, and Citrix Hypervisor held a shrinking residual.

That baseline was anchored on three reinforcing properties. First, vSphere was deeply integrated into every adjacent IT discipline — backup, monitoring, security, networking, storage. Replacing it meant replacing or revalidating dozens of dependent tools. Second, VMware certifications were widespread; the operational labour market favoured retention. Third, perpetual licensing meant the unit economics of staying on VMware looked predictable: pay support, defer capex, stay put.

Broadcom's pricing and packaging changes broke the third leg of that stool. They did not break the first two — but for some customer segments, the third leg alone has been enough.

Where share has moved

The mid-market segment

The clearest share erosion is in the mid-market: customers running roughly 50 to 500 hosts, often with simple vSphere deployments and modest use of higher-tier products. For this segment, Broadcom's pricing changes — bundling vSphere into Cloud Foundation or VVF SKUs, raising minimums, removing the lowest-tier offerings — have made the per-VM economics materially worse without delivering capabilities the mid-market actually uses. The result has been a measurable migration to two alternatives: Proxmox VE for smaller and more technically self-sufficient organisations, and Nutanix AHV for organisations that want a packaged hyperconverged platform.

The migration is real but slow. Mid-market organisations rarely have the platform engineering bandwidth for fast hypervisor swaps. What we see in practice is a "no-new-VMware" decision made in 2024 or 2025, followed by 18–36 months of incremental migration as new workloads land elsewhere and old hosts are retired.

Smaller enterprise edge deployments

A second segment showing meaningful share loss is the edge / remote-office tier within large enterprises. Where central data centres remain on VMware, the small ROBO deployments — three or four hosts per location, dozens of locations — have become economically painful under the new packaging. Many enterprises are quietly replacing these with Hyper-V (where Windows licensing already covers it) or with hyperscaler-backed thin deployments.

Net-new workloads

Across all segments, the strongest signal is at the margin: net-new workload growth is not landing on VMware at the rate it used to. Where a 2022 enterprise might have spun a new application platform on vSphere by default, in 2026 that platform increasingly lands on Kubernetes (on bare metal, OpenShift, or hyperscaler-managed), or on a hyperscaler-native compute service. VMware retains the installed estate; it is losing the growth slope.

Where share has held

Top-tier enterprise data centres

The very largest VMware customers — Fortune 100 enterprises with deeply integrated NSX, vSAN, and Aria deployments and multi-thousand-host footprints — are mostly staying. The cost of migration at that scale is genuinely prohibitive: a credible exit plan is 36 to 60 months, the operational risk is considerable, and the migration cost frequently exceeds three to five years of even Broadcom's elevated subscription pricing. What these customers are doing instead is hard-negotiating the subscription deal, retiring lower-value workloads off vSphere, and demanding contractual protection against future repricing.

Regulated and government workloads

Regulated environments — defence, classified, healthcare, certain financial — show very low migration rates. The recertification burden of changing hypervisor in those environments is large enough that even significant price increases are absorbed.

VMware Cloud Foundation as the consolidation play

Counter-intuitively, the customers Broadcom most wants to retain are also the ones moving to a deeper VMware footprint — VCF subscription with all components in one bundle. Some of this is forced by packaging; some is genuinely rational, where the full stack is in use. Either way, it preserves VMware share within the customer estate while shifting the revenue model.

Hyperscaler VMware: a separate share story

VMware on AWS, Azure VMware Solution, and Google Cloud VMware Engine occupy their own market share narrative. AWS terminated its direct VMware Cloud on AWS service in 2024 in favour of Broadcom-led delivery, which has slowed net new growth on that platform. Azure VMware Solution has retained momentum, helped by Azure's broader hybrid story and Microsoft's willingness to absorb some of the customer-facing complexity. Google Cloud VMware Engine remains a niche play.

For market share accounting, the hyperscaler VMware services are best understood as VMware footprint expressed through someone else's infrastructure. They do not reduce VMware/Broadcom share; they shift where the host sits. Some customers use them as part of an exit ramp, others as a permanent landing pattern.

Hyper-V's quiet comeback

Microsoft has not aggressively marketed Hyper-V as a vSphere alternative, but the deployment data tells a quieter story. For Windows-heavy estates — particularly in mid-market segments where Hyper-V is functionally free under existing Windows Server entitlement — the platform has regained traction. It will not displace vSphere at enterprise scale, but it is materially increasing its share at the mid-tier and the edge.

Nutanix as the consolidated alternative

Nutanix has been the most visible commercial beneficiary of Broadcom's pricing posture. Their AHV hypervisor, packaged with HCI and a coherent management plane, gives mid-market and lower-enterprise customers a single-vendor alternative. Nutanix is publishing strong new-customer numbers that align with what we see in advisory engagements: enterprises with 100–1,000 hosts evaluating Nutanix as the destination, with a real migration plan over two to three years.

KVM-based open stacks

For technically sophisticated customers, KVM-based options — Proxmox VE in particular, but also Red Hat OpenShift Virtualization for OpenShift houses — are the rising open alternative. These platforms are growing share in the segment of customers who view virtualisation as a commodity layer beneath their actual application platform.

The geography of share movement

Share movement is not uniform across regions. The same underlying economic pressure produces different responses depending on local IT culture, alternative-vendor presence, and regulatory context.

North America

The US market is showing the most measurable mid-market migration, with Nutanix and Hyper-V both gaining ground in the 100-to-1,000-host segment. Canada follows a similar curve but with stronger Hyper-V representation, reflecting Microsoft's entrenched federal-government position. Top-tier enterprises in both markets are largely staying on VMware with hard-negotiated subscription deals.

Western Europe

The UK, German, and French markets are seeing similar mid-market dynamics, with the additional factor that several Western European governments have explicitly mandated open-source-first procurement policies that favour KVM-based stacks. Proxmox is gaining unusually strong traction in the German mid-market, and OpenShift Virtualization is taking measurable share in the regulated-services sector across Western Europe.

Nordic countries

The Nordic markets are an early-adopter signal worth watching. Migration projects there tend to start earlier, complete faster, and be more publicly documented than in other regions. We see Proxmox, Nutanix, and hyperscaler-native winning roughly equal shares of the migrations that complete.

Asia-Pacific

APAC is more bifurcated. Australia and Singapore look similar to Western Europe in their mid-market dynamics. Japan is moving more slowly — VMware retention is unusually high — but where movement is happening, it is concentrated toward Nutanix and hyperscaler-native. India's large IT-services exporters are running aggressive alternative-platform evaluations driven primarily by per-customer cost pressure.

The role of hyperconverged infrastructure

One subtle structural shift is that the rise of Nutanix and similar HCI platforms is changing the unit of analysis. Where VMware customers historically thought in terms of "hypervisor + storage + networking + management", HCI platforms collapse those into a single procurement decision. Customers replacing VMware often replace the entire stack at once — including vSAN and elements of NSX functionality — rather than just the hypervisor.

This has implications for how share is measured. Pure-hypervisor share is one metric; full-stack-platform share is a different one. Nutanix's gains are most pronounced when measured against the full-stack VMware platform (vSphere + vSAN + Aria), and less pronounced when measured against vSphere alone.

What stays even when share moves

An important dynamic we observe in real engagements is that even customers who decide to move find themselves keeping a long-tail VMware estate indefinitely. Three categories of workload reliably stay:

Highly stable production systems with no business case for migration. Workloads that have run reliably for years, have known performance characteristics, and have no near-term modernisation roadmap. The cost of migrating them is real; the benefit is largely theoretical.

Workloads with deep VMware-specific integration. Anything tightly coupled to NSX micro-segmentation, vRealize automation, or Site Recovery Manager has a non-trivial replatforming cost. For some workloads this is acceptable; for others, it is the reason they stay.

Compliance-validated environments. Anything where revalidating the underlying platform triggers a full compliance recertification — pharma manufacturing, certain defence workloads, certain payment systems — tends to stay on whatever platform was originally certified.

For most enterprises, the realistic five-year picture is therefore a hybrid: VMware shrinks but does not disappear. The strategic question is how large the retained VMware estate should be, where it should sit, and how to negotiate around it.

Implications for Broadcom strategy

The segmented share picture is consistent with what Broadcom's revenue strategy implies. Losing some share at the mid-market and at the workload margin was always factored into the integration thesis; the value of the deal sits in the strategic 2,000 customers at the top of the market, who are mostly staying. Whether the share movement at the margins becomes large enough to threaten that thesis is the central long-run question for the Broadcom franchise.

The current data suggests the answer is "not yet". Net VMware revenue is up; customer count is down but predictably; the retained accounts are paying meaningfully more per host. The strategy is working on its own terms even as the market headline is "VMware is losing customers". Both statements can be true simultaneously, and both should be considered together when shaping customer-side strategy.

The signal worth tracking

If you want one metric to track over the next 18 months, it is the share of new enterprise workloads landing on VMware. Installed base moves slowly; new workload placement moves quickly. A sustained decline in new-workload landing rate is the leading indicator of long-term share movement. That data is harder to find than headline market-share numbers, but it is the most predictive thing available. Customers who design their own internal version of this metric — tracking what their own teams choose for net-new deployment — get the earliest read on whether their platform standard is durable.

Capability gap analysis: where alternatives still lag

Honest comparison work requires acknowledging where the alternative platforms still trail VMware. Three areas show up repeatedly in side-by-side evaluations.

Software-defined networking sophistication. NSX-T's combination of micro-segmentation, distributed firewalling, and service mesh integration is more mature than the equivalent functionality in any of the alternative platforms. Nutanix Flow, OVN-based stacks, and hyperscaler-native networking all have credible answers, but none of them match NSX's depth at enterprise scale.

Storage feature parity. vSAN's combination of stretched cluster, encryption, dedup-and-compression, and policy-driven placement is best-in-class. Alternative HCI platforms have closed most of the gap, but specific features — particularly around DR with stretched clusters — still favour vSAN.

Ecosystem certification depth. The breadth of third-party software, hardware, and integration certifications against VMware remains larger than for any alternative. For workloads with strict vendor support requirements, this matters.

For most workloads, none of these gaps is decisive. For specific high-stakes workloads, they can be. The right read is workload-specific, not platform-specific.

The role of total estate consolidation

One pattern we are seeing increasingly is that customers use the Broadcom commercial pressure as a forcing function for broader estate rationalisation. The conversation that begins as "what do we do about VMware renewal?" often expands into "what should our overall infrastructure platform strategy look like?".

That expansion is healthy. Customers who use the moment to rationalise their full estate — not just the VMware portion — frequently end up in a stronger long-term position than customers who address only the immediate Broadcom question. The work is larger; the payoff is durable.

A note on data quality

Market share data for enterprise virtualisation is genuinely hard to pin down. Vendors report selectively, analyst houses define their methodology differently, and customer self-disclosure is sparse. The picture in this article is built from multiple sources triangulated against engagement data; it is directionally reliable but not precise to the percentage point. Decisions worth making on a one-percent precision are not the decisions on which the answer here should turn. Decisions on the shape of the trend are exactly the decisions this synthesis is useful for, and the trend has been consistent enough across sources to plan against with confidence.

What the share data implies for IT planning

For IT leaders, the right read is not "VMware is dying" and not "VMware is fine". It is more nuanced. VMware retains its position at the top of the enterprise market; the platform there is still the practical default. At the same time, the centre of gravity for net-new workload deployment is moving away from VMware, and the mid-market is genuinely repricing the relationship.

For an enterprise running VMware today, the practical implications are: protect the installed estate with hard negotiation and audit defence, treat any new workload as a deliberate platform choice rather than an automatic vSphere landing, and build an internal capability — even a small one — for at least one alternative hypervisor, so the cost of strategic optionality is paid in advance rather than under audit pressure.

The share picture in 2026 is one of slow movement, not collapse. But slow movement at this scale compounds. The five-year picture is meaningfully different from the one-year picture, and the licensing decisions made under audit pressure in the next 18 months will shape which side of that compounding curve each enterprise sits on.