Broadcom did not change the discount. It changed the unit of sale.
Perpetual licenses and the standalone SKU catalog are gone, replaced by two subscription bundles priced per physical core with a 16 core minimum per processor. That single rule reshapes every estate differently, and it means the first saving on a Broadcom renewal is not a discount at all. It is a corrected core count and the right bundle, because a percentage off an inflated basis still overpays.
Prepared by Redress Compliance · August 10, 2026 · Broadcom advisory. Based on 40 to 50 VMware renewals benchmarked, 2024 to 2025.
Executive summary
The comparison is now two bundles, not dozens of products, and the per core unit is what reshapes the bill. Broadcom retired perpetual licensing and the standalone catalog, leaving vSphere Foundation for compute focused virtualization and Cloud Foundation for the full private cloud stack.
Both are priced per physical core as an annual subscription, and both carry the same 16 core minimum per processor.
That minimum is the rule that decides most estates: a socket with fewer than 16 cores still bills at 16, so low density hosts pay for capacity they do not physically have while high density hosts absorb the minimum with far less waste.
Core counts were wrong on 20 to 35 percent of estates, which is the first saving and it costs nothing to take. Quotes routinely arrived built on inflated or stale counts: decommissioned hosts still carried in the inventory, sockets counted at the wrong density, environments long since retired.
Because the unit of sale is now the core, an error in the count multiplies through the entire term. Auditing true physical cores and removing dead hosts corrects the basis before anyone argues about a percentage, and it is the one move that does not require Broadcom to agree to anything.
VCF was proposed where VVF would have covered the workloads in 40 to 60 percent of cases. Cloud Foundation adds NSX networking, storage, and the broader management suite on top of the compute layer, and it is the costlier default the account team leads with.
Most virtualization estates do not consume that stack. If you are not running NSX and the wider management suite, vSphere Foundation usually covers the same workloads at a materially lower cost, so match the bundle to what actually runs rather than to the platform ambition in the slide deck.
Multi year subscription totals ran 1.5 to 3 times the prior perpetual plus support spend, and the lever that moves them is a credible alternative.
The increase is structural rather than negotiable in principle, so the defensible response is to fix the basis, negotiate term length and uplift caps rather than only the headline discount, and price a partial migration so the renewal is visibly a choice.
Costing a move of test and development hosts to Hyper V, Nutanix, or Proxmox moves a Broadcom quote more reliably than any spreadsheet argument, because it converts a captive spend into a contested one.
vSphere Foundation against Cloud Foundation
| Dimension | vSphere Foundation (VVF) | Cloud Foundation (VCF) | Buyer note |
|---|---|---|---|
| Scope | Compute and virtualization | Full private cloud stack | Match to real workloads |
| Includes | vSphere, vSAN entitlement | NSX, vSAN, Aria, more | Do not pay for unused layers |
| Pricing | Per core subscription | Per core subscription | Both use the 16 core minimum |
| Best for | Classic vSphere estates | Private cloud platforms | VCF is the costlier default |
The shift from perpetual to subscription means cost is recurring and tied to cores rather than a one time purchase plus maintenance, so the estate's shape now drives the bill more than its size.
Three host profiles behave very differently under the minimum: high density hosts absorb it with low waste, low density hosts bill at 16 cores per socket even where fewer exist, and stale counts inflate everything because decommissioned hosts sitting in the quote are billed like live ones.
Work out which profile dominates your estate before you read a price, because the same discount produces wildly different outcomes across them. The full bundle mechanics sit in the VMware licensing guide and the platform detail in the VCF pillar.
The moves that actually move the quote
- Audit true physical core counts and strip out dead hosts, because 20 to 35 percent of estates in our file were quoted on inflated or stale numbers, and the error compounds across every year of the term.
- Default to VVF unless the VCF stack is genuinely in use. If NSX and the wider management suite are not running, the compute bundle covers the same workloads at materially lower cost.
- Model the 16 core minimum against your host density, since a low density estate is paying for phantom cores and consolidation onto fewer, denser hosts is worth more than a discount point.
- Negotiate term length and uplift caps, not just the headline percentage, because a strong first year rate with an uncapped uplift is a deferral rather than a saving.
- Cost a partial migration so the option is real. Even pricing a move of test and development hosts to Hyper V, Nutanix, or Proxmox signals the renewal is a choice. The full route comparison sits in the VMware alternatives guide.
The Broadcom VMware negotiation brief 2026
The bundle comparison, the per core arithmetic, the minimum core rule, and the renewal moves that hold against a Broadcom opening position.
Get the white paper →Fix the basis before you argue the price
The common advice is to accept the new bundles and push for a bigger discount. That advice misreads what changed.
Broadcom did not tighten discounting so much as move the unit of sale to the physical core, and then quote many estates on inflated counts and on the costlier bundle by default, which means a percentage off the wrong basis still overpays by a wide margin.
Fixing the basis is a sequence rather than a negotiation. Verify true physical core counts against the current inventory and remove every host that has been decommissioned since the last renewal, because those hosts bill exactly like live ones.
Confirm whether the estate consumes the Cloud Foundation stack or only needs the compute bundle, and be specific about which workloads justify NSX rather than accepting the platform story wholesale.
Model what the 16 core minimum does to your particular host density, since a fleet of low density sockets is quietly buying capacity that does not exist, and consolidating onto fewer denser hosts removes that waste permanently.
Only then does a discount conversation mean anything, because at that point the percentage is applied to a number that reflects your estate rather than Broadcom's opening position.
The pricing posture across the wider catalog sits in the Broadcom pricing analysis, and the exit route in the VMware exit plan.
- 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 VMware by Broadcom renewals, 2024 to 2025
Across roughly 40 to 50 VMware renewals we benchmarked between 2024 and 2025, the Broadcom restructure changed the unit of sale faster than buyers changed their estates, and the gap between those two speeds is where the money went:
Share quoted on inflated or stale core counts, where a corrected inventory cut the basis before any discount conversation began.
Cases where Cloud Foundation was proposed but vSphere Foundation would have covered the workloads at materially lower cost.
Three patterns recurred: core overcount on 20 to 35 percent of estates, bundle mismatch in 40 to 60 percent of cases, and sticker shock as multi year subscription totals ran 1.5 to 3 times the prior perpetual plus support spend. The buyer side move is to correct the basis first and negotiate second.
Audit the cores, remove the dead hosts, default to the compute bundle unless the private cloud stack is genuinely running, model the minimum against your host density, then take term length, uplift caps, and a costed migration option into the conversation.
Win the core count and the bundle choice and the price follows. The wider library sits in the Broadcom and VMware practice.
Your first five moves
- Audit true physical core counts and remove decommissioned hosts from the quote, because 20 to 35 percent of estates were priced on a number that no longer described them.
- Confirm whether the estate uses the VCF stack or only needs VVF, since the costlier bundle was proposed unnecessarily in 40 to 60 percent of the cases we benchmarked.
- Model the 16 core minimum against your host density, then consolidate low density sockets, because that waste is structural and recurs every year of the subscription.
- Negotiate term length and uplift caps alongside the discount, so a strong opening rate is not undone by an uncapped increase in year two.
- Cost a partial migration to make the alternative real, even for test and development hosts, because a documented option is the strongest lever in a Broadcom renewal. The Broadcom practice runs the core audit and the renewal with you.
Frequently asked questions
What changed in VMware licensing after Broadcom acquired it?
Broadcom moved VMware to subscription only licensing sold in two main bundles, vSphere Foundation and Cloud Foundation, priced per physical core with a minimum core count per processor.
Perpetual licenses and standalone SKUs were retired, so buyers now compare two packages rather than dozens of products, and the per core unit reshapes each estate differently.
What is the difference between VVF and VCF?
vSphere Foundation is the compute focused bundle for virtualization estates. Cloud Foundation adds the full software defined data centre stack including NSX networking, storage, and the broader management suite.
VVF suits classic vSphere workloads and VCF suits genuine private cloud platforms, and the price gap between them is large enough that the choice usually outweighs the discount.
How is VMware priced per core in 2026?
Per physical core, billed as an annual subscription, with a minimum of 16 cores per processor. A server with fewer than 16 cores per socket still bills at 16, so core density and socket count now drive the bill more than raw host count.
That is why a low density fleet pays for capacity it does not physically have.
Is the subscription model more expensive than perpetual VMware?
For many estates, yes. In our file multi year subscription totals ran 1.5 to 3 times the prior perpetual plus support spend, particularly where servers carry fewer than 16 cores per socket.
The change is structural, so the defensible response is to right size the core basis and benchmark alternatives rather than treating the quote as fixed.
Can you still negotiate VMware pricing under Broadcom?
Yes, but the leverage has moved. Discounting still exists and is won through core count accuracy, bundle right sizing, term length, and a credible alternatives position rather than through SKU mixing.
Buyers who audit their true core counts and model a migration routinely move the Broadcom quote; buyers who only ask for a bigger percentage rarely do.
Why is the core audit worth more than the discount?
Because the discount applies to the basis, and the basis was wrong on 20 to 35 percent of the estates we benchmarked. Decommissioned hosts still in the quote, stale inventories, and miscounted sockets all inflate a number that then multiplies across every year of the term.
Correcting the count costs nothing and does not require Broadcom to agree to anything.
What are the main alternatives to VMware in 2026?
Microsoft Hyper V and Azure Local, Nutanix, Proxmox, and public cloud rehosting. None is a drop in replacement, and the migration cost is real.
But a credible plan for even part of the estate, priced out rather than asserted, is the strongest single lever in a Broadcom renewal, because it converts a captive spend into a contested one.
The VMware Estate After the Repackaging
Part 2 of the Negotiating Broadcom series. Two bundles, per core with a sixteen core floor, three year terms paid up front, and a support horizon in October 2027 that decides your timing more than your renewal date does. What the estate actually looks like now, and which numbers are real.
