Two years of Broadcom: the mass VMware exodus that never happened (and what is actually going on)

On July 16, Broadcom put out a press release: Standard Chartered, a bank operating across 54 markets, is committing long-term to VMware Cloud Foundation and already runs 70% of its global infrastructure on it. That same week, industry surveys say 86% of VMware customers are shrinking their footprint. Both things are true at the same time. And that apparent contradiction holds everything you need to know before your next renewal.

What the data says, not the headlines

When Broadcom closed the VMware acquisition in November 2023 and started making moves — the end of perpetual licenses, the catalog compressed into a few bundles around VCF, per-core billing — the dominant prediction was a mass exodus. Two and a half years later, the data tells a different story. A February 2026 CloudBolt survey of 302 IT decision-makers in North America leaves three figures worth more than any opinion:

In other words: almost everyone is moving, almost no one has fully left. What we have is not an exodus; it is a reduction of dependency, customer by customer and workload by workload. We see the same thing in the companies that come to us: nobody switches off a production cluster because of a press release, but many have stopped putting new workloads on VMware.

The fear was 2x; the reality, 25–49% more (which is still a lot)

The same survey leaves another nuance that is uncomfortable for both camps. In 2024, 73% expected their VMware bill to more than double. In reality, only 14% have suffered increases above 100%; the most common bracket has been rises of 25% to 49% — although one enterprise respondent reports extreme cases of several hundred percent. And 2024 research from Omdia corroborates it from another angle: 72% of organizations reported cost increases on their virtualization platform due to the licensing model change.

The fact that the average increase is smaller than the initial panic does not make it reasonable: we are talking about paying a quarter to half more for the same software you already had, without a single new feature to justify it. And 85% of IT leaders expect further increases. That expectation, more than today's bill, is what is draining new workloads away from VMware.

Standard Chartered: the customer Broadcom designed all of this for

Back to the July 16 announcement. A global bank moving 70% of its infrastructure to an integrated private cloud on VCF is not empty marketing: for that profile — global scale, regulation across 54 markets, large teams, a budget to match — a single stack with top-tier support makes complete sense. Let's not kid ourselves: Broadcom's strategy is not to retain every VMware customer. It is to concentrate the business on fewer, larger customers who spend more. The Standard Chartered deal does not contradict the 86% who are reducing; it confirms it: they are two halves of the same plan.

The question that matters is not "is VMware good technology?" (it is; we have been operating it since the old vSphere versions and have never said otherwise). The question is: are you the customer the pricing was designed for? If you are a bank in 54 markets, probably yes. If you are a mid-sized company with three hosts and two hundred VMs, the suit was tailored for someone else — and you are paying for it anyway.

Why nobody leaves fast: the brakes are real

If 86% want to reduce and only 4% have left, something is applying the brakes. Coverage of the survey points to the usual ones: migration complexity, alternative costs higher than expected, and technical barriers. A public example with numbers: Danish manufacturer Danfoss is migrating about 1,100 servers off VMware, and the project runs from spring 2025 to the end of 2027. Two and a half years. A serious hypervisor migration is not a weekend with a disk converter; it is inventory, dependencies, testing, maintenance windows and a rollback plan for every wave.

There is another figure in the survey that almost nobody comments on, and to us it is the most expensive of all: nearly two-thirds of organizations have changed strategy at least twice since the acquisition. Every swerve is months of work thrown away. And one nuance about the destination: close to 75% of those migrating are going to public cloud, not to another hypervisor. Legitimate — but swapping an opaque license for a consumption-based bill does not spare you from doing the math; it only changes where the surprise can come from.

The pattern that actually worries us: finding out via a distributor memo

In March 2025, VMware customers found out — via a memo from the distributor Arrow, not an official announcement — that the minimum purchase would jump from 16 to 72 cores. For a small two-node cluster, that meant paying for cores you would never use. The backlash was strong enough that the measure was withdrawn before April 10, and a spokesperson went as far as stating that Broadcom "has never announced a price change". Formally true. In practice, plenty of companies spent two weeks running numbers on a rule that appeared and vanished without anyone signing it.

This, for us, is the underlying argument — more than today's price. You cannot budget three years ahead on a platform whose purchasing rules can change via memo. The rational response is neither panic nor loyalty: it is to reduce dependency to the point where a rule change cannot wreck your year. Which is exactly what the 86% are doing.

What we would do with a renewal due in 2027

If you signed your contract after the acquisition, it most likely comes up for renewal around 2027. And the decision is not made when the renewal quote arrives: it is made 12–18 months earlier, which is how long it takes for a real alternative to exist on the table. The three honest paths:

One note of honesty worth repeating whenever we talk about this: we are resellers of neither VMware nor Proxmox. We sell licenses for neither. When we recommend staying, we do not get paid for it; when we recommend migrating, we size the project with the math done, not with the commission in mind. It is the only way this advice is worth anything.

In short

Two years after Broadcom, neither mass exodus nor business as usual: 86% reducing, 4% fully out, real increases of 25–49%, and a vendor that has chosen its customers. The only bad position is having decided nothing and renewing in 2027 with the calendar against you. If you want us to look at your case — inventory, dependencies, the math of staying and the math of leaving — this is how we approach VMware to Proxmox migrations.

Sources (verified Jul 22, 2026): CloudBolt survey (Feb 17, 2026, 302 IT decision-makers in North America): 86% reducing footprint, 4% full migration, cost expectations vs reality, 85% expecting further increases — Channel Dive; Omdia 2024 (72% reporting cost increases), Danfoss case (~1,100 servers, 2025–2027) — TechTarget (Jun 26, 2026); Standard Chartered–Broadcom agreement (Jul 16, 2026, 70% of infrastructure, 54 markets) — Broadcom; 72-core minimum announced via Arrow memo and withdrawn — The Register (Mar 28, 2025) and Licenseware.