VCF components, license the operating model and not the slide
VMware Cloud Foundation in 2026 is one integrated bundle, vSphere, vSAN, NSX, and the operations and automation layer, sold per core on subscription with the pricing assuming you use all of it. The bundle composition drove more cost surprise than the headline per core price, because most estates paid for the platform vision while running the hypervisor reality.
Prepared by Redress Compliance · August 7, 2026 · Broadcom and VMware advisory. Based on 30 to 40 VCF licensing engagements advised 2024 to 2025.
Executive summary
Two thirds paid for components nobody deployed.
Most estates ran vSphere and vSAN only while paying full VCF rates, with NSX and the automation layer sitting unused in roughly two thirds of the estates we reviewed: NSX is the bundle's biggest swing component.
Transformative where network virtualization is real and pure shelfware where the network stays physical.
Deployment of individual components does not reduce the bill; only the tier decision does.
The 16 core minimum taxes the distributed estate.
A two socket edge host with 8 core CPUs licenses as 32 cores, and across dozens of sites the phantom cores added 10 to 20 percent to effective spend in the estates we modeled: host hardware decisions are now licensing decisions, and sometimes consolidation beats licensing at the edge entirely.
Paid optionality is shelfware with a subscription.
The standard reseller advice, take full VCF because the delta looks small and the extra components are free optionality, inverted in 25 of the 30 to 40 estates: the unused components anchored renewal baselines, inflated support.
And weakened every future downgrade conversation, because the vendor priced loss aversion rather than usage.
License the operating model you run, and keep the upgrade as your concession to trade.
The negotiation order is tier, cores, term, and the order matters.
The tier decision sets the baseline every other number scales from and is the one the account team least wants reopened: demand a priced vSphere Foundation alternative for every cluster where NSX and automation are undeployed.
Verify the core inventory yourself from entitlement data and RVTools before accepting Broadcom's count, and trade term for rate only after tier and count are fixed, with documented renewal protection.
The bundle composition against typical deployment
| Component | In VCF | In vSphere Foundation | Typically deployed |
|---|---|---|---|
| vSphere compute | Yes | Yes | Universal |
| vSAN storage | Yes, with a capacity allowance per core | As an add on | Common |
| NSX networking | Yes | No | A minority |
| Operations and automation | Yes | Operations only | Partial |
| HCX migration | Yes | No | Project phases only |
| vDefend and Avi | Add ons | Add ons | Rare |
The leak is paying VCF rates for vSphere usage.
The catalog collapsed to essentially two tiers, VCF as the full stack and vSphere Foundation as the compute centric package, and the composition decision moved from the SKU list to the bundle tier: if NSX and the automation layer are not in the operating model.
The estate is buying the platform vision while running the hypervisor reality, and the delta compounds at every renewal.
The tier match, workload by workload
- Full private cloud: VCF where vSAN, NSX, and self service automation are genuinely deployed, or funded for deployment with dates and owners.
- Compute plus storage: vSphere Foundation with the vSAN add on where networking stays physical, at the structurally lower point.
- Edge and small sites: the phantom cores counted honestly, where consolidation sometimes beats licensing altogether.
- Exit candidates: stable workloads with no VMware dependency, priced into the alternatives case that doubles as negotiation leverage.
The VMware Cloud Foundation licensing brief
The bundle decisions end to end: the component audit, the tier arithmetic, the core verification, and the negotiation order that holds.
Get the white paper →The negotiation order, and why it holds
Tier first, because the bundle tier sets the baseline every discount scales from, and the priced vSphere Foundation alternative per undeployed cluster is the demand the account team least wants on the table.
Cores second, verified from your own entitlement data and RVTools before accepting the vendor count, because the 16 core minimum and the refresh driven density drift both inflate the base by default.
Term last, traded for rate only after tier and count are fixed, with renewal protection documented, because a long term on the wrong tier locks the shelfware in.
The rate structure the order negotiates against sits in the VCF pricing brief, the portfolio wide mechanics in the Broadcom changes guide, and the exits that discipline the whole conversation in the Nutanix and Proxmox comparisons.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across VCF engagements, 2024 to 2025
Across roughly 30 to 40 VMware estates Fredrik Filipsson advised through Broadcom transitions in 2024 and 2025, the bundle composition drove more cost surprise than the headline per core price:
Estates paying for network virtualization and automation that never entered the operating model.
Workloads fitting vSphere Foundation quoted VCF, with nobody on the buyer side challenging the bundle.
The downgrade path finding is the strategic one: the missed vSphere Foundation opportunities were not pricing errors but unchallenged defaults, and the anchoring works exactly as designed, once the estate is on VCF.
Every renewal starts from the full stack baseline and the downgrade conversation fights loss aversion the vendor priced deliberately.
The correction is the component audit before every renewal, what is actually deployed per cluster, with the tier demand built from it, because the operating model, not the roadmap slide, is what the subscription should license.
Your first five moves
- Audit component deployment per cluster, because two thirds of estates paid for NSX and automation that never ran.
- Demand the priced vSphere Foundation alternative for every cluster where the full stack is not in the operating model.
- Verify the core inventory yourself, entitlement data and RVTools, with the 16 core minimum's phantom cores counted honestly at the edge.
- Negotiate in the order that holds: tier, then cores, then term, with renewal protection documented before any multi year signature.
- Keep the upgrade as your trade, funded NSX deployment for concessions, never the default. The Broadcom practice runs the audit with you.
Frequently asked questions
What is included in VMware Cloud Foundation in 2026?
One integrated per core subscription bundle: vSphere for compute, vSAN with a capacity allowance per core, NSX for network virtualization, and the operations and automation layer formerly sold as Aria, with HCX migration tooling included and vDefend and Avi as add ons.
The catalog collapsed to VCF as the full stack and vSphere Foundation as the compute centric tier.
Do you pay for VCF components you do not deploy?
Yes, fully: the bill is set by the bundle tier, the core count, and the support level, and deployment of individual components does not reduce it.
NSX and the automation layer sat unused in two thirds of the estates we reviewed while billing at full VCF rates, which is the single largest leak in the post Broadcom estate.
What does the 16 core minimum cost in practice?
10 to 20 percent of effective spend in distributed estates: every CPU licenses at least sixteen cores, so a two socket edge host with 8 core processors bills as 32 cores, and the phantom cores multiply across sites.
Host hardware choices are now licensing decisions, and edge consolidation sometimes beats licensing entirely.
When is vSphere Foundation enough instead of VCF?
Wherever the operating model is compute centric: vSphere Foundation covers compute plus the vSAN add on at a structurally lower point for clusters where networking stays physical and self service automation is not deployed.
Workloads fitting it were quoted VCF by default in our reviews, with nobody challenging the tier, which is why the priced alternative is demanded per cluster.
Is taking full VCF for the optionality a good idea?
No: in 25 of the 30 to 40 estates we reviewed, the unused components anchored renewal baselines, inflated support, and weakened every future downgrade conversation, because the vendor prices loss aversion rather than usage.
Optionality paid for annually is shelfware with a subscription, and the upgrade path is a concession to trade, not a default to accept.
How should a VCF renewal be negotiated?
Tier first, with a priced vSphere Foundation alternative for every cluster where NSX and automation are undeployed; verified cores second, from your own entitlement data and RVTools before accepting the vendor count.
And term last, traded for rate only after tier and count are fixed, with documented renewal protection.
The tier sets the baseline everything else scales from.
