What Changed: Socket vs Core Licensing

Under the legacy VMware model, vSphere was licensed per physical CPU socket. A dual-socket server required two vSphere licences, regardless of how many cores each socket contained. A 4-socket server required four licences. The per-socket model was simple, predictable, and — as server hardware evolved to pack more and more cores per socket — increasingly favourable for buyers.

Broadcom has moved to per-core licensing. Every physical core in every licensed host now has a licence cost. This is not merely a price increase — it is a structural change to the licensing model that fundamentally alters the economics of VMware in a dense compute environment.

This change is part of the broader Broadcom VMware commercial transformation that has affected the entire VMware product line since the acquisition. The per-core model aligns VMware with how other enterprise software vendors (Oracle, IBM, SQL Server) license compute-intensive software — and it is equally controversial in those contexts.

Free Guide

Broadcom VMware Licensing Guide

Navigate Broadcom's VMware pricing overhaul: VCF bundles, subscription mandates, and migration options.

Download Free Guide → Broadcom VMware Negotiation Service

The 16-Core Minimum Rule

The per-core model includes a critical provision: each physical CPU socket is subject to a minimum of 16 licensed cores, regardless of actual core count. This means:

The minimum rule disproportionately affects organisations running older, lower-core-count hardware. Modern servers with 32- or 64-core processors are above the minimum on every socket, so the minimum is irrelevant — but these servers carry the full core count at the per-core rate, which is still materially higher than the old per-socket cost in most estates.

Calculating Your Cost Impact

The most important thing any VMware customer can do right now is calculate their specific cost exposure under the new model. This exercise takes 30–60 minutes with access to vCenter or your CMDB and is essential context for any renewal negotiation.

Step-by-Step Calculation

Example: 50-Host vSphere Estate
Physical Hosts (vSphere licensed) 50 hosts
Sockets per Host (avg) 2
Cores per Socket (avg) 24 cores
Licensed Cores per Host (24 cores × 2 sockets) 48 cores/host
Total Licensed Cores (50 hosts × 48) 2,400 cores
VCF Per-Core Rate (negotiated) $160/core/year
Annual VCF Subscription Cost $384,000/year
Prior Perpetual (annualised ~15%) ~$120,000/year
Year-on-Year Cost Increase +$264,000 (+220%)

This illustrative example — which is representative of mid-size VMware deployments — shows why enterprises are reacting strongly to Broadcom's pricing model. A 220% cost increase on a mature infrastructure platform that has not added materially new value is not a commercial proposition that most enterprises can accept at face value.

Stay Ahead of Vendors

Get Negotiation Intel in Your Inbox

Monthly briefings on vendor pricing changes, audit trends, and contract tactics. Unsubscribe any time.

No spam. No vendor affiliations. Buyer-side only.

How Hardware Generation Affects Your Exposure

Hardware Generation Cores per Socket (typical) Licensed Cores (with minimum) Cost Impact vs Old Model
Server refresh 2018–2020 (older)8–12 cores16 cores (minimum applies)3×–4× cost increase
Server refresh 2020–202216–24 cores16–24 cores2×–3× cost increase
Server refresh 2022–2024 (current gen)32–64 cores32–64 cores1.5×–2× cost increase

The counterintuitive insight here is that organisations with older, lower-core-count hardware face the worst cost impact — the 16-core minimum creates a significant overpayment relative to actual hardware capacity. Organisations with newer, higher-core-count hardware face a still-significant but proportionally smaller increase.

VVF as a Cost Reduction Strategy

For organisations primarily using vSphere without significant vSAN or NSX deployment, the VMware vSphere Foundation (VVF) tier offers the same core compute virtualisation at approximately half the VCF per-core price. If your estate is effectively a vSphere-only environment, VVF is the right tier — and moving from VCF to VVF proposals can halve the per-core cost.

VVF does not include vSAN or NSX. If you are actively using these products, VCF is likely the correct tier. If you are paying for VCF but not deploying vSAN or NSX, you are subsidising components you do not use. Our article on VCF vs VVF licensing provides detailed guidance on this decision.

Negotiation Leverage: Your core count calculation is your primary commercial weapon in the Broadcom renewal conversation. A precise, documented core count — including identified decommission candidates — forces Broadcom to negotiate on specific numbers rather than broad estimates. Never enter a renewal conversation without this data.

Core Count Reduction Through Decommissioning

One of the most direct ways to reduce your VCF cost is to reduce the number of licensed cores — by decommissioning hosts before the contract start date. This is not always straightforward (workload capacity must be maintained), but in many estates there are underutilised hosts that could be consolidated before the subscription window opens.

A host decommission analysis should consider: current utilisation rates (vCenter performance data), workload consolidation opportunities, DR and HA headroom requirements, and any planned hardware refreshes in the near term. Even a 10–15% reduction in licensed hosts translates directly to a 10–15% reduction in annual subscription cost — a material saving on large estates.

Negotiation Implications of Core-Based Pricing

The shift to core-based licensing has specific implications for how you negotiate with Broadcom. Key points to address in any commercial conversation include:

For a complete negotiation playbook, our guide on Broadcom VMware negotiation tactics covers the full commercial process. The IT Negotiations Broadcom advisory service provides hands-on support for enterprises facing significant renewal decisions. Our team has delivered negotiated outcomes that reduce the core-based pricing impact by 25–40% compared to initial Broadcom proposals.