Committed vCore ran 40 to 70 percent above peak production, and the multi year cap that fixed the rate removed the true down

The multi year deal is sold as protection, and on the rate it is. What it also does is fix the quantity for the same three years, which matters a great deal when the quantity was set 40 to 70 percent above what production actually consumes.

Prepared by Redress Compliance · August 16, 2026 · Salesforce advisory. 20 to 30 MuleSoft engagements benchmarked, 2024 to 2025.

MuleSoft Anypoint pricing is unlike anything else in the Salesforce portfolio, which is why procurement teams routinely underestimate the renewal. The tier choice locks the price for the contract term.

The gap between committed and consumed is the single largest number in a MuleSoft file. Priced at the Platinum production vCore rate, it is straightforward to size.

Then apply the term. A three year deal at a 7 to 12 percent annual uplift charges that excess again each year with the uplift on top, and the ramp structure that delivered the cap is generally the same structure that removed the right to true down. The rate protection and the quantity lock arrive in the same clause, and only one of them is presented as the benefit.

MuleSoft renewals are usually approached as a rate negotiation, which is understandable because the rate is the number on the quote and the uplift is the number that moves it. The finding across the engagements benchmarked points somewhere else entirely. Buyers committed to 40 to 70 percent more vCore capacity than peak production consumed, which means the most expensive decision in the contract was a quantity decision taken before anyone argued about price.

Two things make that gap unusually persistent here. The first is that vCore capacity is bought against a forecast of integration growth, and integration growth is genuinely hard to predict: a platform team that under provisions blocks delivery, so the incentive runs firmly toward over provisioning. The second is that the remedy most buyers reach for makes it permanent. A multi year ramp deal caps the annual uplift at 7 to 12 percent, which is a real protection, and in the deals we reviewed it came without a usage true down. So the buyer trades away the ability to reduce quantity in exchange for certainty on rate, having just committed to a quantity that is 40 to 70 percent too large.

The bundling practice compounds it by removing the feedback signal. Salesforce folds MuleSoft into a Sales Cloud or Service Cloud renewal, and doing so obscured the unit price by 20 to 35 percent in our file. When the unit price is not visible, the over commitment is not visible either, because nobody can see what a vCore costs and therefore nobody prices the unused ones. The carve out is not an accounting nicety. It is what makes every other lever measurable, which is why it belongs first in the sequence rather than last.

The practical order is therefore inverted from how these deals usually run. Produce a documented runtime utilisation report before the quote. Right size vCore against measured peak rather than forecast. Move what belongs on Composer off the vCore meter. Consolidate the connector stack, because premium connectors carry 15 to 40 percent uplifts and accumulate quietly. Carve MuleSoft out of the Salesforce bundle so the unit rate is visible. Only then take the multi year cap, and take it with a true down right attached rather than instead of one. The vCore mechanics sit in the MuleSoft pillar, and the wider library in the Salesforce practice.

• Produce a documented runtime utilisation report first, covering measured peak rather than provisioned capacity, because every later lever prices off that number.
• Right size vCore against measured peak, which is where the 40 to 70 percent gap closes and where the largest single saving sits.
• Test the tier downgrade, since Gold, Platinum, and Titanium differ on features you can enumerate, and the tier locks for the whole term.
• Move low code workloads to Composer, priced per flow rather than per vCore, and stop paying runtime rates for integrations that never needed runtime.
• Consolidate premium connectors before renewal, because each carries a 15 to 40 percent uplift and the stack is rarely reviewed as a whole.
• Carve MuleSoft out of the Salesforce bundle, then take the multi year cap with a true down right rather than accepting the cap as a substitute for one.

Across roughly 20 to 30 MuleSoft engagements benchmarked, core based and capacity based pricing produced the largest gap between what buyers committed and what they used:

Multi year ramp deals locked uplift of 7 to 12 percent per year with no usage true down. That combination, an over sized quantity and a term that prevents reducing it, is the specific shape of a MuleSoft overspend.

MuleSoft was acquired by Salesforce in 2018 and its pricing model never converged with the rest of the portfolio. Treating it as a Salesforce line item rather than as a separate integration platform negotiation is the assumption that costs the most.

Watch the briefing · 5:50Every Salesforce Product Is a Different NegotiationWhy MuleSoft prices unlike anything else in the portfolio and has to be negotiated on its own terms.

How is MuleSoft Anypoint priced?

By platform tier, Gold, Platinum, or Titanium, with the tier locked for the contract term, plus consumption measured in vCores. One vCore is roughly one virtual processor core of CloudHub runtime, currently around $8,000 a year per production vCore at Platinum.

How much capacity do buyers typically over commit?

Between 40 and 70 percent above what peak production actually consumed, across the engagements benchmarked. On a 100 vCore commitment that is 29 to 41 vCores nobody runs, or roughly $232,000 to $328,000 a year at the Platinum production rate.

Why does the over commitment persist?

Because vCore is bought against forecast integration growth, and under provisioning blocks delivery, so the incentive runs toward buying too much. The multi year deal that caps the uplift then removes the ability to true down, which makes the over sizing permanent for the term.

What is the true down problem?

Multi year ramp deals locked uplift at 7 to 12 percent a year with no usage true down. The rate protection and the quantity lock arrive in the same clause, and only the rate protection is presented as the benefit. Ask for the cap and the true down, not the cap instead of it.

Does bundling MuleSoft with Salesforce help?

It helps Salesforce. Folding MuleSoft into a Sales Cloud or Service Cloud renewal obscured the unit price by 20 to 35 percent, and when the unit price is invisible so is the over commitment. The carve out is the first leverage move because it makes every other lever measurable.

What is the connector tax?

Premium connectors carry list rate uplifts of 15 to 40 percent on top of the platform tier. Each one is small at signing, the stack accumulates across the term, and it is rarely reviewed as a whole before the annual uplift applies to all of it.

When should we use Composer instead?

For low code integration that never required CloudHub runtime. Composer is priced per flow rather than per vCore and sits inside Salesforce licensing. Establishing which workloads genuinely need runtime is a sizing exercise that belongs before the tier conversation.

What renewal uplift should we expect?

Seven to twelve percent annually without structured pushback. A multi year deal offers the cap, but it locks the volume, so the cap is only worth taking once the volume is right and a true down right is attached.

Which tier should we be on?

The lowest one whose features you actually use, tested feature by feature rather than assumed. The tier choice locks the price for the whole contract term, so a downgrade tested at renewal is worth more than the same test taken mid term when nothing can change.

What is the first thing to produce?

A documented runtime utilisation report covering measured peak production vCore over a full year. Every other lever, tier fit, right sizing, Composer migration, and the cap decision, prices off that single number.