VMware Cloud Foundation pricing in 2026. What the per core rate becomes after negotiation.

VCF has a list rate per core, but almost every enterprise deal is negotiated below it. Term length, scale and a credible alternative cut the price, while the 16 core minimum per socket bills small hosts for cores they do not have.

How much does VMware Cloud Foundation cost per core in 2026?

The list price for VMware Cloud Foundation (VCF) is $350 to $400 per core per year on a one year subscription. That is already well below the roughly $700 per core Broadcom opened with after it closed the VMware acquisition in early 2024.

List is the ceiling, and few enterprise customers pay it. What you actually pay depends on three things: how many cores Broadcom counts, which bundle each cluster carries, and how much discount your term, size and alternatives earn. The first two usually move the bill more than the third.

What does the per core price include?

One VCF core license covers the full private cloud stack on that core, whether or not you run every part of it. Broadcom's documentation lists these components for VCF 9:

• Compute. ESX hosts and vCenter, the part most customers bought VMware for.
• Storage. vSAN, with 1 TiB of vSAN capacity included per licensed VCF core.
• Networking. NSX Manager and NSX Edge for software defined networking and security.
• Management. VCF Operations, VCF Automation, SDDC Manager and VCF Operations HCX for migration.

VMware vSphere Foundation (VVF), the smaller bundle, keeps ESX, vCenter and VCF Operations but drops NSX and VCF Automation, and includes only 0.25 TiB of vSAN per core. The price gap between the two bundles is why the tier decision belongs cluster by cluster.

Why is there no public VCF price list?

Broadcom publishes no official rate card for enterprise customers. Every deal is negotiated through Broadcom's account team or a partner, so the discount structure matters more than the list figure itself.

Ask for the one year price per core in writing at the start. It becomes the comparison point for every multi year offer, and it stops the conversation from sliding between list, "standard" and "special" prices. Our note on which VCF price list applies covers the versions that circulate in quotes.

How much do term length, scale and ramps take off the VCF list price?

Each factor has a fairly predictable discount band, and together they take committed enterprise deals well below list. Term length earns the first discount, size earns the next, and a credible alternative adds the most negotiable part on top.

Combined, these bands explain where the market settles. Large customers with committed multi year terms reach effective rates around $100 to $130 per core per year. One year renewals with no alternative on the table pay close to list.

On a 20,000 core environment, the gap between those two outcomes is roughly $5M a year. That is a fair measure of what the negotiation effort is worth, and of how much preparation it justifies. The discount bands by core count tier break the scale factor down further.

A worked example: 60 small hosts from first quote to signature

Say you run 60 hosts, each with two 12 core processors. That is 1,440 physical cores, but the 16 core minimum bills each socket at 16, so Broadcom counts 1,920. The table uses $350 list, a 25 percent three year discount and 10 points for a documented alternative. All figures are hypothetical and annual.

At the same 35 percent discount, the subscription costs $436,800 on the inflated count and $327,600 on the corrected one. The $109,200 a year between them comes entirely from the 480 phantom cores.

Over a three year term that adds up to $327,600, a full year of the corrected subscription. Set that figure against the cost of the replacement hardware.

How does the price differ for a small and a large environment?

• Smaller environments. The scale band is out of reach, so term and a credible alternative do most of the work. Small hosts are often a large share of the count, so fixing the core minimum can be worth more than the last few points of discount.
• Mid sized environments below 10,000 cores. Term bands apply in full and the account team has room to move. Cluster by cluster tier decisions start to pay because some clusters will not need NSX or VCF Automation.
• Above 10,000 cores. The scale band comes into play, and the lowest effective rates we see are only reached with a committed term. Here the contract terms matter as much as the rate, because a renewal cap protects a seven figure annual bill.

When does the five year band pay off?

It pays off when you are confident VMware will still run those workloads in year five. Take the 1,440 cores from the worked example and say the three year band gives 25 percent off and the five year band 35 percent. That is $262.50 against $227.50 per core, which saves $50,400 a year.

Now say a migration retires half those cores at the start of year four. Paying for 720 idle cores in years four and five at $227.50 costs $327,600, more than double the $151,200 the deeper band saved in years one to three. Without reduction rights, the longer term is the more expensive one.

How does the 16 core minimum raise the VCF bill?

VCF is licensed per physical core with a minimum of 16 cores per CPU socket. A host with two 12 core processors has 24 cores and is billed for 32, so 8 cores you do not have are in the subscription for every year of the term.

On modern high core count processors the minimum makes no difference. The cost falls on small two socket hosts with 8 and 12 core parts, which typically run edge sites, management clusters and legacy workloads. In our engagements the minimum added 10 to 25 percent of phantom cores, mostly where budgets could least absorb it.

How do you check your own core count before Broadcom does?

Build your own host level count before the quote arrives, then reconcile Broadcom's number against it line by line. These sources give you what you need:

• RVTools. The vHost tab lists CPU sockets and cores for every host vCenter manages. Export it and add a billed cores column at 16 per socket minimum.
• vSphere API. Each host's CpuInfo reports numCpuPackages (sockets) and numCpuCores (physical cores). Count physical cores only, never hyperthreads.
• Broadcom's License Counting PowerCLI tool. Broadcom documents it for larger environments with vCenter. Run it yourself so you see the count Broadcom will see.
• Disabled cores. Broadcom notes the tool can report inaccurate results if cores are disabled in the BIOS. Do not rely on disabled cores to lower the count without written agreement.

Our guide to VMware core licensing covers the counting rules in more detail.

What should you fix before the quote is built?

Both corrections belong before the quote. Otherwise the commitment is sized on the inflated count and the phantom cores become part of the contract.

• Consolidate the small hosts. Fewer sockets with more cores per socket turn phantom cores into real capacity. A hardware refresh that looked marginal on its own often pays for itself once the licensing difference is included.
• Scope the count to what will still exist. Hosts retiring during the term, edge sites moving to another platform and clusters planned for cloud migration belong in the ramp discussion. Keep them out of the committed core count.

Why does the first Broadcom quote come in so far above the old vSphere bill?

The product, the bundle and the metric all changed at once. In our engagements, first quotes opened at 2 to 4 times the prior vSphere run rate, and that multiple was the product transition working as designed rather than a pricing error.

Perpetual vSphere licenses with annual support were converted into a VCF subscription that bundles vSphere, vSAN, NSX and the management layer, used or not. The per core metric with its socket minimum then repriced the same hardware at a higher count.

Which clusters need full VCF, and which fit vSphere Foundation?

Scoping comes before any talk of percentages, because every cluster moved to the smaller bundle shrinks the base the discount applies to. The VCF versus vSphere Foundation comparison covers the feature differences in detail.

• Clusters running NSX in production. Microsegmentation, overlay networks or distributed firewalling usually justify VCF.
• Clusters with heavy vSAN storage. The 1 TiB per core entitlement in VCF can offset buying vSAN capacity separately. Check the actual TiB used against 0.25 TiB per core under VVF.
• Compute only clusters on external storage. These are usually VVF candidates, since they use neither NSX nor much vSAN.
• Management and edge clusters. Look at them separately. They are often small hosts where the core minimum hits hardest.

What will the Broadcom account team say, and how should you reply?

• "VCF is the standard offer for customers your size." Reply with your per cluster usage and ask for a VVF quote on every cluster that runs no NSX or VCF Automation.
• "The quote uses the core count in our records." Send your host level count and ask Broadcom to reconcile theirs against it before any pricing discussion.
• "This discount is only available until quarter end." Your timeline is set by your own term end date. Our note on Broadcom quarter end timing shows when that pressure is real.
• "Migrating off VMware will cost you more than renewing." Share the costed evaluation for the workload slice you have scoped. A plan with dates and a budget is harder to dismiss than a threat.

What have we seen in recent Broadcom VCF renewals?

Across roughly 35 to 50 Broadcom VMware renewals and migrations Fredrik Filipsson advised between 2024 and 2026, final VCF prices landed 30 to 55 percent below list once term, scale and competitive pressure were combined. The spread between prepared and unprepared customers was wider than in any other vendor practice we run.

• An alternative changed the result. A documented exit evaluation, even a partial one covering a single workload slice, moved the final discount 8 to 15 points against customers who negotiated with nothing on the table.
• Opening quotes came down in two stages. Scoping reduced the base first, then negotiation reduced the rate on what was left.
• Phantom cores were found too late. The core minimum was almost always discovered after the quote arrived. Customers who ran their consolidation numbers first negotiated from a smaller, accurate base.

The licensing rules behind these prices are set out in our VCF licensing guide.

Why we do not start with the per core rate

The usual advice is to push the account team for the biggest possible percentage off list and hold out for quarter end. We disagree with that order, because in our renewals the base held more recoverable money than the rate.

A hard won discount on an inflated count leaves phantom cores and unused NSX in the contract for the whole term. Correct the count and the tier per cluster first, then negotiate the rate with a costed alternative in hand.

The list rate is the least negotiable number on a VCF quote. The core count underneath it is the one you control.

Which contract terms should come with a multi year VCF subscription?

Ask for price protection and flexibility in the same agreement as the discount. Broadcom does not offer these by default, so they have to be requested and traded for term length.

• Renewal cap. A maximum increase on the per core rate at the next renewal. Without it, the discount ends when the term does. See our note on the VMware price cap negotiation.
• Price hold on growth cores. Cores added during the term at the contracted rate, so expansion does not reopen the price.
• Ramp schedule. Core quantities that rise or fall by year in line with your migration plan, written into the order form.
• Tier flexibility. The right to move clusters from VCF to vSphere Foundation at renewal as their use changes.
• Reduction rights. The ability to drop cores for retired hosts at renewal instead of carrying them forward.
• Portability language. Clear terms on where the subscription can run if workloads move to a hosted VMware service.

When should you start preparing for a VCF renewal?

What to do next

1. Split the quote into base and rate. List the core count by host and socket, flag every minimum, and note the bundle per cluster. If the quote arrived as a single total, ask the partner to reissue it in this format.
2. Run the consolidation numbers on small hosts before you commit, so phantom cores become real cores or leave the count.
3. Decide VCF or vSphere Foundation per cluster, and refuse the full stack where the smaller bundle does the job.
4. Cost an exit for one workload slice and document it, even if you intend to stay. It changes the tone of the meeting as well as the discount.
5. Trade term length for protections: a renewal cap, price holds on growth cores, and ramp years matched to the migration plan.
6. Get a second view on the quote. Our Broadcom VMware practice and the VMware assessment tools run this sequence with you.

Frequently asked questions