What Broadcom's VMware actually costs you now — and when migrating pays off

The renewal quote is the reason you are reading this: it arrived, it was a multiple of last year's, and now someone has to decide whether to pay it or leave. Both sides of that decision cost money, and the panic version — migrate immediately — is as expensive a mistake as the passive one — pay whatever they ask. Here is the honest cost math, the way we run it with clients before anyone touches a hypervisor. If you have already decided to move, the migration mechanics and the platform comparison are separate pieces; this one is only about the money.

Why the bill jumped

Three changes stacked on top of each other:

• •Subscription, not perpetual. You used to buy a licence once and pay a modest annual support fee. Now you rent, per year, forever — so the recurring number that used to be "support" is now the whole licence.
• •Per-core, with floors. Licensing counts every CPU core, with a minimum of 16 cores per CPU — and, from 2025, a much higher 72-core minimum per order layered on top where it applies (reports on how universally it stuck differ, so your own quote is the number that counts). A small host with lightly-used sockets now pays for cores it barely touches.
• •Bundles, not à la carte. The old long menu collapsed into a few packages (VMware Cloud Foundation and the smaller vSphere Foundation), so you often buy — and pay for — capability you did not ask for to get the piece you need.

The reported increases vary too widely to quote a single multiplier — some contracts landed at roughly double, others far higher — which is exactly why a headline number is useless and your own quote is the only one that counts.

Get your real number first

Before comparing anything, write down what staying actually costs, three years out:

• •Your core count across all hosts (sockets × cores), not VM count — that is what you are billed on.
• •The bundle you are being quoted into and what it includes that you will not use.
• •The three-year subscription total, because a one-year figure flatters staying and hides the recurring nature of the new model.

That number — three-year cost to stay — is the thing every alternative gets measured against. Most "VMware is now unaffordable" reactions are to a one-year sticker, not a three-year plan.

The other side of the ledger

Migration is not free, and pretending it is produces the second expensive mistake. A fair comparison prices the whole move:

• •Engineering effort — the platform build (cluster, storage, networking, backup), disk conversion and driver work, testing, and the cutover. This is the biggest line and it is labour, whether yours or a partner's.
• •The target platform's own cost — Proxmox's optional support subscription, or Hyper-V's Windows Server licensing. "Free" hypervisors still cost something to run supported.
• •Retraining and risk — your team learning a new platform, and the real possibility of a rough cutover. Both are costs even when they do not appear on an invoice.
• •Replacing the VMware-only pieces — vSAN, NSX, SRM, Horizon each need a substitute, and those substitutes are where migration budgets quietly blow up.

The break-even

Put the two columns side by side over three years: *(subscription to stay)* versus *(one-time migration cost + three years of the new platform's running cost)*. The migration pays off when the second column is clearly smaller — and the more cores you have, the faster it does, because VMware's cost scales with cores while the migration is a mostly fixed, one-time project.

That gives three honest outcomes:

• •Migrate when your core count is high enough that three years of VMware subscription dwarfs a one-time move — the common case for anything beyond a couple of small hosts.
• •Renew once when the delta is real but you need time to plan, or a contract or project makes this year the wrong time to move. Buying twelve months to do it properly is a legitimate, often wise, choice.
• •Stay when your estate is small, your VMware dependence is deep (heavy vSAN/NSX/SRM), and the migration effort genuinely exceeds the licence saving. It happens, and we will say so.

Hidden costs on both sides

The honest ledger includes the things that are not on either quote. Staying carries lock-in and the near-certainty of the next increase — you are pricing not just this renewal but your exposure to the one after it. Leaving carries the cutover risk, the retraining, and the temptation to under-scope the vSAN/NSX replacement. Neither side is free; a good decision counts both.

What we do

We build this comparison for clients before recommending anything, because we resell no one's licences and have no reason to push a migration or a renewal. We will size your three-year cost to stay, price the move honestly against it, and give you the outcome the numbers actually support — migrate, renew once, or stay. When it is migrate, we do the server and virtualisation work end to end; when it is stay, we tell you plainly. The worst decision here is a fast one made on a headline; the second worst is no decision made out of dread of the quote.