Broadcom Sold You Two VMware Bundles. You Probably Need One.

Broadcom finished rewriting the VMware catalog. Perpetual licenses are gone, and so are the standalone product SKUs that used to fill a renewal quote. You now buy a subscription priced per physical CPU core, and the core count on your hosts decides the bill.

What is not settled is what you actually receive. The new model sells you a stack in two bundles, and most buyers only ever deploy part of it. Understanding the packaging is the difference between a fair renewal and a subsidized one.

So the question stops being which editions to renew. It becomes which bundle matches the workloads you run, and how many cores those workloads touch. Miss that and you overspend in year one, then repeat the mistake at every renewal.

Your core count sets the price, not your host count

Every physical core on every licensed ESXi host counts toward capacity. Broadcom applies a floor of 16 licensed cores per physical CPU. A socket with eight cores still bills as sixteen. Cores you disabled in BIOS still count too.

The arithmetic punishes low-density hardware. Three hosts with one eight-core CPU each require 48 licensed cores, not 24. Modern processors ship with 32, 64, or more cores per socket. Core counts now balloon past the old socket math. Broadcom publishes the rule and worked examples in its licensing overview and again in its licensing model page.

Count your cores before you negotiate a price. Broadcom’s core counting guide walks through the method and ships a PowerCLI tool to automate it. Some sellers still describe a 72-core order floor as if it were a product rule. Broadcom’s public pages state a 16-core minimum per physical CPU. Treat any larger number as a commercial term and demand it in writing.

The two bundles share a naming trap. Broadcom kept Foundation in both names, so quotes blur together in a spreadsheet. Read the bill of materials, not the label, before you sign.

Component licenses vanished with the old SKUs. You no longer license NSX, HCX, or the automation tools separately. Assign the primary license to a vCenter instance and the connected components are licensed automatically. That is real simplification, and it also removes the old habit of trimming a quote by dropping individual products.

VCF is not a bigger VVF, and you rarely need both

Broadcom sells two primary bundles. VMware vSphere Foundation is the virtualization tier, built around vSphere with operations tooling and a small storage entitlement. VMware Cloud Foundation adds the full private cloud stack, including NSX networking, VMware vSAN, and VCF Operations.

The storage math differs too. The VCF documentation grants 1 TiB of vSAN capacity per licensed core. VMware vSphere Foundation grants 0.25 TiB, rounded up. Both bundles assume you will fire up the whole stack.

That assumption is where money leaks. Many estates run classic vSphere workloads and never touch NSX or the automation layer. Moving a quote from VMware Cloud Foundation to VMware vSphere Foundation is the single largest lever a buyer holds. It beats any discount percentage.

Add-ons extend the same logic. There is a storage add-on for capacity past the bundle entitlement, and a private AI add-on tied to cores. Each one is another line to justify against real usage rather than a roadmap.

Support compounds the same way. Both bundles include Production Support. A higher tier adds faster response targets and deeper engineering access. Compare what each quote actually contains rather than assuming parity.

Audit first, then negotiate the term

Reconcile the licensed core count against what you genuinely run. Teams that skip this step keep paying for decommissioned hosts for years. Pull the real inventory, confirm the sockets, and match it to the contract line by line.

Ask for the count in writing, then rebuild it on your own spreadsheet. Vendors make arithmetic errors, and those errors rarely land in your favor.

Term length is your second lever. A one year bridge term usually beats a three year lock-in signed before you have weighed alternatives. SUSE Rancher Prime and Red Hat OpenShift remain credible options for teams that want a real exit.

Time the work to your renewal window, not the seller’s quarter. Renewal is the one moment a buyer holds real power, and a modeled alternative is what makes that power credible.

Get the core count right, buy the bundle you will deploy, and keep your exit open. That is the entire 2026 negotiation.

The economics behind this migration wave, including the three year cost model and why exits run long, are in the 2026 State of Enterprise Infrastructure report.

Every announced commitment in this space is tracked with its source in our AI data centre power commitments record.

Related reading. A Full VMware Exit Costs More Than Three Years of Broadcom’s Bill. Cloud Native Transformation Is an Operating Model Change, Not a Lift and Shift. Bundling Is Not Unifying. Multicluster Kubernetes Management Still Has a Gap.. The 2026 State of Enterprise Infrastructure.