Realized VCF pricing clears between $185 and $275 per core per year against a $350 list, and the 90 dollar spread is decided by spend tier, term length, and whether you brought an alternative
Published VCF list prices range from $130 to $1,800 per core, and Broadcom benefits from every hour you spend arguing about which one is real. The number that matters is the realized one: $185 to $275 per core at enterprise scale, with sub-$200 reserved for buyers who bring both a multi-year commitment and a credible migration plan. This article sets the benchmark bands by spend tier so you know, before the first call, whether the quote in front of you is a good deal or a 40 percent overpay.
Prepared by Redress Compliance · August 21, 2026 · Broadcom VMware advisory. VCF renewal and first-subscription engagements, 2024 to 2026.
Executive summary
The list price argument is a trap, and the only defensible anchor is the $350 per core per year figure that three independent sources converge on for 2026.
Competing published figures of $130 to $150, $175 to $240, and $1,200 to $1,800 reflect different unit bases, different support tiers, and in at least one case a per-CPU number mislabeled as per-core, so treat any quote that cites an unnamed price list as an unpriced quote.
Realized VCF lands at $185 to $275 per core per year, meaning the market clears 30 to 55 percent below the $350 list and the whole negotiation is a fight over roughly 90 dollars of core-year.
On a 5,000 core estate that spread is worth $450,000 per year, or $2.25 million across a five-year term, which is more than most buyers spend on the entire evaluation exercise.
Term length is the single largest published discount lever: three years buys 18 to 28 percent, five years buys 28 to 38 percent, and scale above 10,000 cores adds another 5 to 12 points on top.
Stack those and a 15,000 core five-year deal should reach the mid-40s off list, which puts the target squarely at $190 to $210 per core before any competitive pressure is applied.
The 16-core-per-socket floor inflates licensable cores by 10 to 25 percent on older hardware, and consolidation onto denser hosts recovers 15 to 25 percent of the increase before you ask for a single discount point.
The 72-core-per-CPU minimum floated in April 2025 was withdrawn after customer protest, yet buyers are still being quoted against it, so verify the floor in writing before you accept any core count.
The mechanics: what a VCF per-core price is actually made of
The number on the quote is not a price, it is the output of five variables Broadcom controls to differing degrees, and only two of them are genuinely fixed.
Start with the list anchor at $350 per core per year, which is the figure most current advisory sources converge on for VCF and which itself was halved from the $700 opening position Broadcom floated in January 2024.
Layer term length on top: three-year commitments pull 18 to 28 percent off that anchor, five-year commitments pull 28 to 38 percent, and one-year renewals get you nothing but the list band and a worse position next April.
Estate size then adds a scale adder of 5 to 12 points once you cross roughly 10,000 licensable cores, which is where the largest deals separate from the mid-market.
The fourth variable, licensable core count after the 16-core-per-socket floor, is the one buyers consistently underprice: on hosts with 8-core or 10-core CPUs, the floor inflates the billable core count by 10 to 25 percent before a single point of discount is discussed.
The fifth, support tier, is where Broadcom recovers margin quietly, and it is also where the vSAN entitlement lives: VCF includes 1 TiB per core, and anything beyond that meters at roughly $20 to $35 per TiB per month depending on storage class. Four of these five move in negotiation.
The 16-core floor does not, which is precisely why it deserves your engineering attention before your procurement attention.
| Variable | Observed range | Who controls it | Where the leverage is |
|---|---|---|---|
| List anchor | $350 per core per year (halved from $700 opening, Jan 2024) | Broadcom, unpublished | Force the price list name, version, and effective date in writing |
| Term length | 18 to 28% off at 3 years; 28 to 38% off at 5 years | Shared | You concede duration only against a locked uplift cap |
| Estate scale | 5 to 12 additional points above ~10,000 cores | Broadcom's tiering model | Aggregate subsidiaries and regions into one paper before quoting |
| Licensable cores | 16-core-per-socket floor; 10 to 25% inflation on low-core hosts | You, through hardware refresh | Consolidation recovers 15 to 25% before any discount conversation |
| Support tier and vSAN | 1 TiB per core included; $20 to $35 per TiB per month overage | Broadcom's SKU structure | Model overage as a line item, not an operational afterthought |
The reason peer benchmarks mislead is that the denominator is not standardized.
Two buyers can both report "$210 per core" and be paying materially different real prices, because one licensed 4,200 physical cores against a 5,600-core invoice under the socket floor while the other refreshed onto 32-core CPUs and licensed close to physical.
The first buyer's effective cost per unit of compute is roughly a third higher, and no discount percentage will reveal it.
This is why the sequencing matters more than the negotiation skill.
Run the core normalization first, using something like a structured core count and estate assessment, and produce a single number: licensable cores under the 16-core floor on your current fleet, versus licensable cores on a consolidated fleet.
In our experience across these deals, that delta alone is worth 15 to 25 percent of the invoice and Broadcom will never surface it for you, because their quote engine takes your existing host inventory as an input, not as a question.
Only once you have the normalized core count is the per-core rate a meaningful thing to argue about.
Go into the discount conversation with an inflated denominator and you will win 30 percent off a number that was 20 percent too large in the first place, then report the outcome internally as a success.
Reconciling the list price: why published figures range from $130 to $1,800
The spread in published VCF list prices is not analyst confusion, it is a structural feature of a market where Broadcom publishes no price list, faces no disclosure obligation, and benefits from every hour you spend litigating which figure is real. Lay the six out honestly.
The consensus figure is $350 per core per year, cited independently by multiple advisories including European sources reporting as of mid-2026. A closely adjacent band of $350 to $400 is quoted for one-year terms, which is the correct read: short terms are priced at or above the anchor.
Then the divergence starts. One publisher puts VCF list at $175 to $240 depending on support tier, which is much closer to realized enterprise pricing than to any list.
Another puts it at $130 to $150 and explicitly frames that as a ceiling rather than a floor, which is a description of a negotiated outcome, not a list. And then the outlier: $1,200 to $1,800 per core per year for VCF Standard, published by the same source that elsewhere quoted $175 to $240.
That is almost certainly a per-CPU or per-bundle figure being reported in per-core units. Do not average it into anything.
The practical consequence is that you cannot benchmark against a list you cannot verify, so stop trying to and change what you demand instead.
Before you respond to any quote, make Broadcom or the reseller state in writing the name of the price list, its version, its effective date, and the unit basis (per physical core, per licensable core after the socket floor, per socket, per bundle).
Reps will resist this because a named, dated price list becomes a comparison point at renewal and across your own subsidiaries. Expect the response to be a verbal assurance that "this is standard pricing" plus a discount percentage with no stated denominator.
Treat that as a refusal and escalate: the same discipline applies here as in any Broadcom renewal where the costed exit does more work than the discount argument.
The most useful data point in the entire list price debate is not any of the six figures, it is the trajectory. Broadcom opened at $700 per core per year in January 2024 and settled the anchor at $350 inside roughly a year.
A vendor that halves its own list within twelve months is telling you exactly how much elasticity sits in the number, and it is not five percent. A strong outcome, given that history, is not a percentage off an unverified list.
It is a realized rate in the $185 to $220 range for a genuine enterprise estate on a multi-year term, with the price list and unit basis named in the contract so the same argument does not have to be relitigated in three years.
How to negotiate Broadcom VMware in 2026
How to negotiate a Broadcom VMware deal in 2026: VCF bundle economics, the core minimum mechanics, subscription conversion exposure, and the levers.
Get the white paper →Spend tiers and the discount bands each one actually reaches
The discount you can reach is a function of how much of Broadcom's quarterly number you represent, not how hard you push. Below roughly 1,000 licensable cores you are a transactional account, the reseller owns the relationship, and Broadcom's field team never touches the file.
The entry math is brutal at that end: at the 72-core practical floor, VCF runs about $25,200 per year against roughly $13,680 for VVF at the same core count, and the VCF bundle is engineered for estates north of 200 cores.
Buying VCF at 72 cores means paying for NSX and the full Aria suite you will not deploy.
From 1,000 cores upward the curve steepens quickly, then flattens hard above 15,000 because Broadcom has already given you everything the standard discount matrix authorizes and the remaining points require an exception approved above the regional VP.
The bands below reflect realized enterprise transactions in the $185 to $275 per core range published by independent advisories, cross-checked against what we see in deals across the table.
| Tier (licensable cores) | Realistic discount off $350 list | Target per-core rate | What unlocks the top of the band |
|---|---|---|---|
| Under 1,000 | 10 to 22% | $273 to $315 | Reseller margin give only. Assume no Broadcom exception. |
| 1,000 to 5,000 | 22 to 38% | $217 to $273 | Three-year term plus a written vSphere Standard fallback position. |
| 5,000 to 15,000 | 35 to 50% | $175 to $228 | Five-year term (28 to 38 point band) plus a costed migration plan. |
| 15,000+ | 45 to 55% | $158 to $193 | Scale adds 5 to 12 points, but only against a named alternative. |
Two structural rules cut across every tier. First, sub-$200 per core is not a volume outcome. It requires both a multi-year commitment and a credible alternative, and the published evidence and our deal experience agree on this.
A 20,000-core buyer with no exit plan and a one-year term will land above $220. A 6,000-core buyer with a costed Nutanix or Proxmox migration for 30 percent of the estate and a five-year signature will land below $195. The alternative is worth more discount than the volume is.
Second, the term length band does most of the arithmetic before scale contributes anything: three-year terms cut 18 to 28 percent, five-year terms cut 28 to 38 percent, and scale layers 5 to 12 points on top.
If you are being offered 40 percent on a one-year term, the number is not generous, it is a placeholder designed to reset your anchor upward at renewal.
The flattening above 15,000 cores catches large estates off guard. Buyers assume the curve keeps bending and open at 65 percent. It does not, and asking for it burns two weeks and your credibility with the deal desk.
The last five points at that scale come from three specific places: multi-year prepayment (Broadcom will trade roughly 3 to 5 points for cash in year one), a public reference or case study commitment, and a co-term that pulls a separate Broadcom estate (Symantec, CA, Tanzu) into the same paper.
Push on those three and you get movement. Push on volume alone and the deal desk will hold, because your volume is already priced in.
Before you argue about any of this, resolve your core count. The 16-core-per-socket floor adds 10 to 25 percent on low-density hosts, and consolidation onto denser servers has recovered 15 to 25 percent of the increase before a single discount point is negotiated. Run that exercise first.
Our note on pricing the cores before the discount covers the mechanics. A 30 percent discount on an inflated core count is worse than a 20 percent discount on an accurate one, and Broadcom will happily let you win the percentage argument while losing the invoice.
Discount off list versus discount off first quote: two denominators, one trap
The published discount figures circulating in the market measure two different things and Broadcom's team will quote whichever one flatters the offer in front of you.
"28 to 35 percent off first proposal" and "30 to 55 percent off list" are not comparable measurements, and the gap between them is where a good-looking concession hides a bad outcome. Watch the arithmetic.
A first quote already sitting 20 percent below the $350 list, then discounted a further 30 percent, lands at $196 per core, or 44 percent off list. A first quote issued at full list and discounted 35 percent lands at $228, or 35 percent off list.
The second buyer got the bigger headline percentage and paid $32 more per core. At 8,000 cores that is $256,000 a year, or $1.28 million across a five-year term, surrendered because the buyer negotiated against the wrong denominator.
| Path | Opening quote | Concession quoted | Net per core | True discount off $350 list |
|---|---|---|---|---|
| Quote below list, then discount | $280 (20% off list) | 30% off quote | $196 | 44% |
| Quote at list, then discount | $350 | 35% off list | $228 | 35% |
| Quote above list on inflated cores | $350 on 12% excess cores | 40% off list | $210 effective | 28% effective |
The third row is the one that costs real money and the one buyers almost never catch. Broadcom does not need to inflate the rate when it can inflate the base.
A 40 percent discount applied to a core count padded by the 16-core socket floor, an unconsolidated host estate.
Or a quietly reinstated 72-core minimum produces a worse effective price than a 30 percent discount on an accurate count. Percentage discounts are the only currency the vendor is willing to be generous with, because the vendor controls the denominator on both sides.
The fix is procedural and costs you nothing. Require the reseller to state list price, first quote, and net price on the same page of the same document, with the licensable core count printed beside each.
Then convert every subsequent concession into an absolute dollar-per-core figure and a total annual contract value before you respond to it.
When the account team says "we can find you another eight points," the only correct reply is "eight points off which number, and what does that make the per-core rate." Track the negotiation in dollars per core per year from the first call.
The same discipline that governs the costed exit that moves the quote further than the discount does.
Broadcom's deal desk is fluent in percentages and comfortable there. It is markedly less comfortable when every offer is restated as a single number you can benchmark against the $185 to $275 realized band, because that number is the only one that survives contact with your budget.
The analysis: the catalog collapse is the price increase, and the discount conversation hides it
Ask yourself why Broadcom's sellers are so comfortable talking about discount percentage. They will walk you through 32, 38, sometimes 42 points off list with a straight face, and they will do it early in the cycle rather than holding it back. That comfort is diagnostic.
Discount percentage is the only metric in this negotiation that has not deteriorated for buyers since the acquisition closed, which makes it the ideal terrain for the vendor to fight on. The unit price of a core moved.
The composition of what a core buys moved far more, and in the opposite direction from your interests. A seller who keeps the conversation anchored on percentage off a list number that Broadcom sets unilaterally, and does not publish, has already picked the battlefield and won the first engagement.
The structural change is the price increase. Roughly 8,000 SKUs and 168 bundles collapsed into four subscription offerings. That is not a simplification exercise, it is the systematic removal of every alternative you used to price against.
When NSX sold standalone at approximately $1,200 per CPU per year, you had a decision: buy it or do not. Now it is inside the VCF per-core rate whether or not a single NSX segment is provisioned in your estate. Aria arrives the same way.
So does the 1 TiB per core of vSAN entitlement, which is generous only if you use it and is dead weight in the price if your storage sits on an external array you already own. Broadcom did not raise the price of network virtualization.
It made network virtualization non-optional and then invited you to negotiate the discount on the combined figure.
The downgrade paths are closing too, which is what turns a bad bundle into a captive one.
VVF has been unavailable in parts of EMEA since December 2025, so buyers in those markets who wanted the smaller footprint at roughly $80 to $110 realized per core no longer have a SKU to point at. vSphere Standard survives at approximately $50 per core per year, but as several buyers have reported.
Sales motion actively steers away from it.
A product that exists on the price list and cannot be bought without a fight is not a substitute, it is a rhetorical device. The four-SKU catalog is narrow by design, and the narrowness is worth more to Broadcom than any single percentage point of discount it concedes.
Run the arithmetic that the discount conversation suppresses. A buyer who wins 40 percent off a $350 list lands at $210 per core, comfortably inside the realized band and defensible against any peer benchmark.
If 60 percent of that bundle is capability the estate will never deploy, the effective cost of the capability actually consumed is $525 per core. That buyer congratulates himself on a strong deal while paying a 50 percent premium over the pre-acquisition equivalent for the same delivered function.
The percentage is real. The saving is not. This is why the correct denominator is cost per workload or cost per VM, not cost per core, and why the first spreadsheet you build should divide the annual commitment by the number of production VMs it actually supports.
The strategic consequence follows directly. If the increase is structural, grinding the percentage cannot reverse it, because the percentage is applied to a number Broadcom controls and to a bundle Broadcom composed.
The only move that reaches the structure is a credible threat to leave it: a costed migration for a defined workload segment, with target platform, timeline, and a per-VM cost that Broadcom's team can verify. That is what re-prices the bundle rather than the rate.
Buyers who arrive with a costed exit plan for a named subset of the estate consistently move quotes further than buyers who arrive with three competitive percentage benchmarks, because one changes the vendor's assumption about renewal certainty and the other does not.
Expect the counter. Broadcom's response to an exit plan is to question its credibility, offer to fund a proof of concept that consumes your quarter, or extend term length in exchange for the deeper band.
Take the term extension only if the per-core number lands below $200 and the exit rights on the migrating segment survive the signature. Otherwise you have traded the one piece of leverage that reaches the structure for a discount on a bundle you already established you do not need.
Core minimums: the multiplier that invalidates every peer benchmark
Every per-core benchmark you receive from a peer, an advisor, or this article is worthless until both sides normalize for the gap between physical cores and licensable cores. The 16-core-per-socket floor is the mechanism.
A two-socket host running 8-core CPUs licenses as 32 cores, exactly double its physical count, and the buyer paying $210 per licensable core on that host is actually paying $420 per physical core of compute delivered.
Peer benchmarks conducted at 32-core or 48-core hosts do not carry that penalty at all. On estates built around small-core CPUs, published analysis puts the inflation at 10 to 25 percent, and in the worst configurations it approaches 100 percent.
Two buyers can quote each other identical per-core numbers and be 40 percent apart on total cost.
Then there is the 72-core minimum, which is still doing damage after being withdrawn. Broadcom announced it on April 10, 2025 through a partner memo, faced sustained protest, and retracted it. The 16-core-per-CPU floor remains the standard.
The problem is that resellers who were briefed on the increase and never briefed on the reversal continue quoting against it, and in our experience across recent VCF renewals, buyers who do not challenge the assumption simply pay it.
If your quote references a 72-core floor, the seller is either working from stale material or testing whether you are. Both cases resolve the same way: demand the applicable minimum in writing, with the effective date and the document it derives from, before any pricing discussion continues.
The pre-quote move that pays for itself is consolidation. Redensifying onto higher-core-count hosts before you accept a quote recovers 15 to 25 percent of the increase, and it does so before a single discount point is negotiated.
Run the core count exercise against your actual host inventory first, because the number of licensable cores you present is the one input Broadcom cannot discount away.
| Host configuration | Physical cores | Licensable cores at 16-core floor | Inflation |
|---|---|---|---|
| 2 sockets x 8 cores | 16 | 32 | 100% |
| 2 sockets x 12 cores | 24 | 32 | 33% |
| 2 sockets x 16 cores | 32 | 32 | 0% |
| 2 sockets x 32 cores | 64 | 64 | 0% |
| 4 sockets x 10 cores | 40 | 64 | 60% |
The table shows why the 16-core floor is not a rounding rule, it is a tax on estate design. Every host at or above 16 cores per socket pays nothing.
Every host below it pays a penalty that scales inversely with core density, and the four-socket 10-core configuration common in older refresh cycles carries a 60 percent inflation on licensable count before any per-core price is applied.
The practical implication for the negotiation: consolidation is not an IT project you run after signing, it is a pricing input you present before quoting. Present the post-consolidation core count as your requirement, not the current one.
Broadcom prices what you declare, and a buyer who declares 640 licensable cores instead of 900 has removed 29 percent of the deal value before the discount conversation starts, which is more than most buyers win in the discount conversation itself.
What Broadcom does when you push, and what to do about each move
Broadcom's negotiation playbook is narrow and repeatable, which is good news: four moves cover roughly everything you will see across a VCF renewal. The first is denominator substitution.
You ask for 35 percent off list, the rep comes back with "we've already taken you to 30 percent," and the 30 points are measured against the first quote, not against the published $350 per core per year reference.
That reframing is worth real money: 30 percent off a quote that already sits at $310 lands you at $217, while 30 percent off $350 lands at $245, and the rep will happily let you believe those are the same concession. Force the arithmetic into writing.
Every version of the quote should carry a per-core list reference, the total core count including the 16-core-per-socket floor effect, and the resulting effective rate. If the paper does not show the denominator, the concession does not exist.
The second move is substituting term for rate. When you push on price, Broadcom will offer five years at the three-year number and present duration as generosity. It is not.
Published bands put three-year terms at 18 to 28 percent off and five-year terms at 28 to 38 percent, which means the incremental two years are worth roughly 10 points on their own. Price the three-year and the five-year separately, in the same document, and make the vendor quote both.
If the five-year rate is not at least 8 to 10 points below the three-year rate, you are paying for lock-in and receiving nothing for it. In our experience across these renewals, the rep who refuses to quote both terms is signaling that the term concession was already banked into the opening number.
The third move is manufactured floor pressure. Buyers are still being quoted the 72-core minimum order that Broadcom announced in April 2025 and then withdrew after customer protest. The 16-core-per-CPU floor is real and permanent; the 72-core order minimum is not current policy.
If a quote or a partner (Arrow and the other distributors carried the original memo) is sizing your deal against 72 cores per CPU, that is a sizing error worth 10 to 25 percent of the contract value and you should demand a corrected quote before discussing rate at all.
The same applies to support-tier reclassification: moving you from one support level to another while holding the headline discount constant is a price increase wearing a discount label.
The fourth move is the clock. Broadcom knows that an expired subscription under the VCF 9 portal model means time-bound license files stop renewing, and it will let the calendar do the work that the discount conversation could not. The counter is timing, not rhetoric.
Open the renewal 9 to 12 months out, get a written quote in hand by month 8, and hold it while Broadcom moves through its own quarter-end and fiscal-year-end. The buyer with a signed alternative path and eight months of runway is negotiating; the buyer with six weeks is transacting.
Everything in the Broadcom VMware licensing playbook depends on that sequencing.
Where the last 10 points come from: the alternative you can actually cost
The gap between a 30 percent outcome and a 45 percent outcome is not talent at the table. It is whether someone in your organization has done the work of pricing an exit.
Published benchmarks are consistent on this: buyers with a credible Nutanix, Azure Stack HCI, or public cloud migration plan pull 15 to 25 percent off the initial proposal, and the 45 percent-plus results come from stacking three things: right-sizing the core count first.
Holding a costed alternative, and committing to five years.
None of those three is sufficient alone. Right-sizing without an alternative just gets you a smaller bill at the same rate. An alternative without a term commitment gets you a discount Broadcom can claw back at the next renewal.
Credible has a specific evidentiary meaning here, and Broadcom's team reads the difference instantly.
A costed alternative names the target platform, identifies a specific workload tranche with core counts attached, carries a quoted migration cost from an integrator or the target vendor, and appears on a timeline that someone at board or steering-committee level has seen.
An uncosted threat, meaning "we are evaluating alternatives," is priced at zero because it is worth zero. The rep has heard it from every account in the region and has an internal probability attached to it. Move the first tranche, even a small one, and the probability changes permanently.
The economics work because Broadcom's own catalog collapse removed your cheap substitutes. With standalone NSX gone and VVF withdrawn in parts of EMEA since December 2025, there is no downgrade path inside the portfolio, so the only remaining substitute is outside it.
That is precisely why a costed exit moves the number more than any rate argument. Start with the core-count and workload assessment before you price the alternative, because you cannot size a migration you have not measured.
The asymmetry worth understanding: Broadcom does not need you to leave for your alternative to work, and you do not need to want to leave.
What moves the price is that the rep has to explain internally why an account with a named target, a sized tranche, and a signed integrator quote should be modeled as a renewal rather than a churn risk. The cost of building that evidence.
Typically 40 to 80 hours of internal effort plus a scoping engagement, is trivial against 10 to 15 points on a seven-figure VCF contract.
One caution. Do not lead with the alternative. Establish the corrected core count and the list-price denominator first, let Broadcom put its opening number on paper, and introduce the migration plan only when the rate conversation stalls.
Deployed early, the alternative becomes background noise the vendor discounts against; deployed at the stall point, it is the reason the second quote arrives 12 points lower.
Metering and true-up: why the portal changes what your benchmark protects
The move from 25-character keys to time-bound license files at vcf.broadcom.com is not an administrative detail, it is a change in who holds the consumption record.
Under VCF 9.0, core tracking lives in Broadcom's portal, overage triggers automatic true-up flagging, and the Broadcom Licensing Assistant reports actual core consumption back to the vendor. That means your negotiated per-core rate now only governs the cores you committed to on day one.
Every core you add after signature is priced in a conversation you enter with no data advantage and no deal-cycle pressure working for you.
I have watched buyers land a clean $205 per core headline and then pay list, $350, on 400 cores of unplanned growth eighteen months later, which pulls the blended rate above $240 and quietly erases the win.
The fix is contractual, not technical: lock the true-up rate at the same per-core number as the base commitment, for the full term, and buy a growth headroom band (typically 15 to 20 percent of committed cores) that can be consumed at the committed rate without a new negotiation.
Broadcom will counter by offering true-up "at then-current list less your discount percentage," which is a trap, because the percentage floats against a list price Broadcom controls unilaterally. Insist on a fixed dollar figure.
The second uncapped meter is storage. VCF entitles 1 TiB per core against VVF's 0.25, and the gap is what makes VCF defensible for storage-heavy estates, but published overage lands at roughly $20 to $35 per TiB per month depending on storage class.
On a 2,000-core estate, 400 TiB of unentitled capacity is between $96,000 and $168,000 a year against a per-core line you already negotiated hard.
Model your TiB-per-core ratio before you agree to the bundle, cap the overage rate in dollars, and use the core and capacity assessment work to establish the real number rather than the architecture team's estimate.
The evidence base: what the published numbers support and where they break down
Be honest with yourself about the sourcing, because Broadcom's sales team will test it. There is no public transaction database for VCF, no independent registry of closed deals, and no regulatory obligation on Broadcom to publish what it charges. Broadcom publishes no official price list at all.
Every advisory figure in circulation, including the ones in this article, is directional: derived from engagement patterns, partner quotes, and advisory collateral rather than audited transactions. That asymmetry is deliberate and it is the vendor's most durable asset in the room.
The correct response is not to pretend to certainty you do not have, it is to be more specific about your own estate than Broadcom is about its pricing.
Published advisory analysis puts post-discount VCF here, with VVF clearing at $80 to $110, against the $350 consensus list.
Three-year terms cut 18 to 28 percent, with scale adding a further 5 to 12 points on large estates.
| Figure in circulation | What the sources support | How to treat it in the room |
|---|---|---|
| $350 per core list | Cited independently by three publishers, including European advisors, as the 2026 baseline reduced from $700 at acquisition | Use as the anchor; it is the most corroborated number available |
| $130 to $150 and $175 to $240 list | Single-publisher claims, mutually inconsistent, likely realized prices mislabeled as list | Do not cite; Broadcom will discredit your whole position on it |
| $1,200 to $1,800 per core, VCF Standard | Contradicted by the same publisher's own $175 to $240 figure | Probable per-CPU or per-bundle unit mismatch; treat as an artifact |
| 72-core minimum order | Announced April 2025, then withdrawn after customer protest; 16-core-per-CPU floor remains | Reject if quoted; buyers are still being priced against a retracted floor |
| 30 to 55 percent clearing discount | Consistent across engagement patterns, widening with term and scale | Defensible as a range, not as a single point |
The recurring patterns matter more than any single figure. Unprepared buyers, meaning those without a core-count audit, a term decision, and a costed alternative, land in the upper quartile of the realized band, near $275, while buyers arriving with all three clear below $200.
The two levers that move the number are the same ones every time: term length and a credible migration path, and neither is a discussion Broadcom starts.
Note also that the 16-core floor inflates licensable cores by 10 to 25 percent on low-density hosts, which means a peer's per-core benchmark is meaningless unless you know their host density. When you present a benchmark, never name the source.
Say instead: "Comparable estates at our scale and term are clearing between $195 and $215, and our board has approved spend at the lower end." The broader Broadcom negotiation framework and the companion piece on proving a benchmark cover how to hold that line when Broadcom demands your evidence.
- 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
Your first five moves
- Normalize the licensable core count and consolidate before you request anything, because the 16-core-per-socket floor inflates small-core hosts by 10 to 25 percent and consolidation onto denser nodes has recovered 15 to 25 percent of the increase before a single discount point is negotiated, so run this with your infrastructure lead 60 days ahead of the first vendor call and use a core-count baseline you can defend line by line.
- Demand the price list name, effective date, and unit basis in writing before you accept any percentage, since published VCF list figures range from $130 to $1,800 per core and at least one of those is a per-CPU or per-bundle mismatch, and a 55 percent discount off an inflated internal sheet can land above a 30 percent discount off the $350 figure, so put the request in the RFI and refuse to model anything until it comes back.
- Set your tier target as an absolute dollar-per-core number and never as a percentage, anchoring on the $185 to $275 realized band with sub-$200 treated as the multi-year plus credible-alternative outcome, then write the target into the negotiation mandate so a rep who moves the denominator cannot move you (in our experience the buyers who hold an absolute number close 8 to 15 points better than those holding a percentage).
- Cost one migration tranche to a named alternative before the second meeting, not a strategy deck but a priced workload set with hardware, staff hours, and a date, because the last 10 points of discount come from a quantified exit and Broadcom prices the difference between a buyer who is annoyed and a buyer who has a signed statement of work waiting.
- Open 9 to 12 months ahead of expiry and lock true-up and growth pricing in the same signature, since the vcf.broadcom.com portal now reports actual core consumption back to Broadcom and flags overage automatically, which means every core added mid-term is transacted at whatever rate you failed to fix, and a co-termed growth rate at the negotiated per-core number typically protects 5 to 10 percent of three-year spend.
Four of these five moves happen before Broadcom quotes you anything. That is the point.
Once a number is on the table, the conversation becomes a percentage argument on the vendor's chosen denominator, and the seller wins that argument roughly every time because they know which price list they used and you do not.
The move that changes the quote most is the costed migration tranche, and it is the one buyers skip because it costs real money to produce. Budget 40 to 80 hours of internal and advisory time.
Against a 2,000-core estate, ten points of per-core movement is worth roughly $700,000 over three years, which makes that assessment the highest-return work in the entire cycle.
Frequently asked questions
What is the actual list price for VMware Cloud Foundation per core in 2026?
The most widely corroborated figure is $350 per core per year, down from the $700 per core opening position Broadcom set in January 2024.
Some advisory sources publish materially lower figures of $130 to $240 per core, and at least one publishes $1,200 to $1,800 for VCF Standard, which appears to be a per-CPU or per-bundle number rather than per-core.
Broadcom does not publish an official price list, so the practical step is to require your reseller to name the price list, its effective date, and its unit basis in writing.
What is a good negotiated VCF price per core?
At enterprise scale, realized VCF pricing lands between $185 and $275 per core per year, which is 30 to 55 percent below the $350 list. Anything below $200 per core generally requires both a multi-year commitment and a credible, costed alternative platform.
Above 15,000 cores with a five-year term, $190 to $210 is a defensible target.
How much discount does a longer term actually buy on VCF?
Published bands put three-year terms at 18 to 28 percent below list and five-year terms at 28 to 38 percent, which is the deepest standard band available. Estates above 10,000 cores add a further 5 to 12 points.
Price each term separately and make Broadcom quote all three, otherwise term extension gets absorbed as a free concession rather than a purchased one.
Is the 72-core-per-CPU minimum still in force?
No. Broadcom announced a move from a 16-core to a 72-core minimum per CPU effective April 10, 2025, then withdrew it after customer and partner protest. The 16-core-per-socket floor remains the standard.
Buyers are still being quoted against the retracted 72-core figure, so demand the applicable minimum and its effective date in writing before you accept a core count.
How much does the 16-core minimum inflate my licensable core count?
On estates running low-core-count CPUs, the 16-core-per-socket floor typically adds 10 to 25 percent to licensable cores. A two-socket host with eight-core CPUs licenses as 32 cores, double its physical count.
Consolidating onto denser hosts before you request a quote recovers 15 to 25 percent of the increase before any discount is negotiated, which is usually the highest-yield single move available.
Should I benchmark my VCF deal per core or per VM?
Per core is how Broadcom prices, so you need it for the negotiation, but per VM or per workload is how you find out whether the bundle is worth buying. A 40 percent discount on a bundle where NSX and Aria go unused is a worse outcome than a 25 percent discount on a right-sized configuration.
Run both numbers and use the per-workload figure to decide the scope question before you argue about the rate.
What is the difference between discount off list and discount off first quote?
They use different denominators and produce very different headline numbers from the same deal. A quote that already sits 20 percent below list, then discounted 30 percent, is 44 percent off list. A quote at list discounted 35 percent is only 35 percent off.
Broadcom will cite whichever framing flatters the offer, so convert every concession into an absolute per-core dollar figure and require list, quoted, and net prices to appear on the same page.
