Timing a Cloud Pak Migration to Maximise Leverage, Not Just to Meet a Deadline

The single most valuable thing you own in an IBM Cloud Pak conversation is the moment you confirm you are willing to move PVU workloads onto VPC. Spend it once, spend it late, and spend it only when a written commercial ask is already on the table.

Your Willingness to Convert Is the Asset, Not the Conversion Itself

Strip the modernisation narrative away and the commercial position is simple: IBM already knows the technical case for moving PVU workloads into Cloud Pak VPC bundles. What it does not know, and what it cannot manufacture, is whether you will actually do it and when. That uncertainty is the only asset you hold that IBM's account team is directly compensated to remove, because Cloud Pak attach and ARR conversion sit at the centre of their quota. The moment you confirm intent, you have handed over the deliverable and left yourself negotiating the price of something you already promised. The price difference is not cosmetic. Redress Compliance's transition work is consistent on this point: conversions executed as standalone product transactions, outside an ELA envelope, typically land at list or near-list VPC pricing, which runs roughly $1,000 to $2,800 per VPC per year depending on product. A 200 VPC estate at that band is $200,000 to $560,000 annually before any discount. Fold the identical conversion into an ELA renewal and you are negotiating inside the 28 to 40 percent band that VendorBenchmark's 2026 study found at $1M+ ACV, tightening to 30 to 38 percent at $2M+ annual. Same workloads, same ratios, same ILMT data, two entirely different numbers, separated only by when you said yes. So treat this as a sequencing problem, not an architecture problem. Your technical readiness date, the point at which the platform team can genuinely run the workload on Cloud Pak, and your disclosure date, the point at which IBM learns you have decided, are two separate decisions. They should never be the same date. Readiness is an engineering milestone. Disclosure is a bargaining chip, and it is worth the most on the day a written commercial ask is already on the table.

Your technical readiness date and your disclosure date are two separate decisions, and they should never be the same date.

What IBM Does the Moment You Signal Intent

The response sequence is predictable enough that you should be able to call it in advance and tell your own team what is coming. First, the account executive escalates. Within days you will have a Cloud Pak specialist or brand overlay on the call, someone who was not in the renewal conversation last year. That overlay is the tell: their presence means the deal has been reclassified internally from a renewal to a conversion, and conversion deals carry different approval paths and different incentive money. Second, the framing changes. What was a renewal of your existing entitlements becomes a "modernisation programme," and the comparison shifts from what you paid last year to what a greenfield Cloud Pak estate would cost. Third, the 70:1 illustration appears. Seventy PVUs to one VPC looks generous on a slide, and it is genuinely favourable for some components, but it is a headline that hides product-specific ratios: MQ at 4:1, MQ Advanced at 2:1, App Connect Enterprise consuming three VPCs of entitlement for every one deployed. Expect the illustration to be built on your best-converting workloads. The damage is anchoring. Once the discussion is denominated in VPC list price, your historical PVU discount, the one you fought for over multiple cycles, stops being the reference point. IBM will not volunteer the benchmark data it holds on what comparable customers in your industry and size band actually pay, because that asymmetry is one of its main advantages. Alongside all of this, watch for two more moves in the same quarter: "transition incentives" that suddenly become available, usually time-boxed to IBM's quarter rather than yours, and an ILMT or audit thread opening in parallel. Read our analysis of an IBM audit opened six months before renewal and how IBM's fiscal clock sets your price before you respond to either.

Model the Conversion Before You Discuss It: The Ratios That Decide the Number

The 70:1 headline (70 PVUs buys one VPC) is the number IBM leads with because it is the only number that flatters the whole estate at once. It is a simplification, and IBM knows it. The moment you go product by product, the arithmetic splits into winners and losers, and the aggregate lands anywhere between a genuine saving and a 15 to 40 percent increase that you will not see until year two. WebSphere ND converts at parity, so it is neutral and should be treated as trading stock rather than a win. Db2 Advanced converts favourably at half a VPC per core. MQ carries a 4:1 factor and MQ Advanced 2:1 inside Cloud Pak for Integration, and App Connect Enterprise runs 1:3, meaning one VPC of actual ACE usage bills three VPCs against your Cloud Pak entitlement. Power estates holding 120 to 140 PVU per core entitlements are the clearest case for retention: those licences are worth more where they sit than anything IBM will offer for them.

The worked example is the one to run in your own estate before IBM ever sees a slide. A standalone Db2 footprint sized at 1,400 PVU maps to 28 VPCs once the bundle ratio is applied, which reads clean until you total it and find a 15 percent effective increase visible only through line-by-line modelling. Then add the container problem: on platforms where resource limits are not set, measured VPC counts came in 20 to 40 percent above the equivalent PVU footprint, because the metric bills the assignment and an unbounded assignment bills the whole node. Set limits before you measure, not after. IBM will respond to a modelled estate by moving the conversation to bundle value and away from ratios. Hold the line at product level, and treat the timing question the way you would IBM's quarter and year-end clock: the model is finished before the calendar pressure starts, not during it.

The 70:1 headline flatters the whole estate at once, which is precisely why it should never be the number you negotiate against.

The Hidden Cost Lines That Destroy Your Business Case After You Sign

Everything IBM leaves out of the migration deck is something you have no leverage to fix after signature. Bundled OpenShift is the largest of these. Most Cloud Paks include OpenShift capacity at roughly 1:3 (one Cloud Pak VPC carries three VPCs of OpenShift), which reads as free infrastructure until you find the entitlement is restricted to Cloud Pak workloads only. Anything else on that cluster needs separate OpenShift licensing. In roughly half of estates reviewed, OpenShift cost was omitted from the initial business case entirely, which understates the true annual bill by a margin large enough to invert the whole case. Price it before disclosure, then hold IBM to a written statement of exactly which workloads the bundled entitlement covers.

Three further lines belong in the ask, not the review. First, non-production runs at half the cost of production: that is a stated ratio, so claim it explicitly in the order document rather than assuming your architecture qualifies. Second, term structure runs 12 to 36 months with a 12-month minimum, and IBM will push 36 to lock the price it wants; a shorter term is worth paying a small premium for if your ratio modelling is still uncertain. Third, the compliance gap. In this practice's experience the exposure opens 12 to 24 months post-transition, when ILMT measurement obligations and Cloud Pak deployment constraints were never documented at signature and the estate has drifted. Get measurement method, resource-limit assumptions and deployment boundaries into the agreement text. If IBM has already opened an audit ahead of your renewal, those same clauses become the settlement currency. A strong outcome here is a business case that survives contact with year two: OpenShift priced, non-production at 50 percent in writing, a 12 or 24 month term, and measurement language you can defend without argument.

Sequencing the Disclosure Against the Renewal and Uplift Calendar

Start modelling nine to twelve months before the renewal date, and keep the modelling entirely internal. The sequence that produces the best number is boring and disciplined: build the PVU-to-VPC ratio model in private, tell IBM only that a portfolio review is underway, let IBM present its unconstrained first proposal, and say nothing about conversion until a written commercial ask exists on paper. IBM's opening proposal is the reference point you want on the table before you reveal anything, because a first proposal built without knowledge of your conversion appetite is priced against your PVU renewal baseline, not against a Cloud Pak bundle IBM has already decided you need. The moment the account team learns you intend to convert, the renewal stops being a renewal and becomes a transformation sale, and the discount conversation restarts from a bundle list price you have never negotiated. Conversion requested outside an ELA context routinely prices at or near VPC list, which is the single most expensive way to make this move.

Then there is the 1 January uplift, which IBM will position as the reason you must decide now. The 2026 announced increases run roughly 6 percent across Distributed PPA software and SaaS, 6 percent on Power with IBM i SWMA at 10 percent, 6 percent on Storage, 6 to 10 percent on TLS, and 1 to 10 percent on IBM Cloud. Read what that list actually says. It is an increase to published rates on metrics you already hold, and in a Cloud Pak conversation you are being asked to adopt a metric you do not yet own. A price increase applied to a VPC rate you have never bought is not your deadline, it is IBM's calendar, and the sibling piece on answering the January 1 buy-now urgency play covers the scripted responses. The overlapping question is where in IBM's quarter you land, covered separately in the Q4 versus Q2 discount comparison and in the pieces on letting a deal slip past quarter end and on when to open an ELA negotiation at all. Sequence the two calendars so your conversion signal lands inside a quarter IBM needs, not inside a January IBM invented.

Trading the Signal: What a Strong Outcome Looks Like in Numbers

When you finally trade the signal, trade it once and trade it for a written package, not for goodwill. Benchmark data across 312 enterprise ELAs puts IBM's target band at 30 to 38 percent off list at $2M+ annual commitment and 28 to 40 percent at $1M+ ACV, and the 5 to 15 point delta between a standard outcome and an aggressive one is the entire prize. That delta is not a function of spend. Accounts above $10M annual have settled at 20 percent off list because they had nowhere else to go, while accounts in the same tier reached 40 percent-plus by putting a credible AWS or Azure migration case in front of the account team. First offers routinely improve 25 to 40 percent when challenged with modelled data rather than opinion, and unbundling plus right-sizing (container resource limits, non-production separation, component ratio corrections) typically surfaces another 20 to 35 percent against list before discount is even discussed.

The non-price terms carry as much money as the discount line. Conversion ratios must be written into the contract, because a ratio referenced from a policy page can be revised without your consent and revision is where the 15 percent hidden inflation appears. Price protection on renewal VPC rates matters more than the first-year number, since a 38 percent discount that resets to list in year four is a 12 percent discount averaged across the term. A metric-conversion clause allowing partial PVU retention protects Power estates carrying 120 to 140 PVU-per-core entitlements that convert badly. Hold the signal until all seven items appear in the same paper, and treat any partial concession as a reason to use the silence window rather than to close.

A price increase applied to a metric you have not yet adopted is not your deadline, it is IBM's calendar.

When the Audit Arrives First: Converting Compliance Pressure Into Conversion Price

The common inversion is that IBM does not open with a migration pitch. It opens with an ILMT deficiency finding or a formal audit notice, and by the time the conversion conversation starts you are no longer choosing to move, you are being told that moving is the remediation. That reframing is worth real money to IBM, because a conversion presented as a fix carries no obligation to price competitively. Refuse the framing. A compliance exposure and a Cloud Pak conversion are one transaction, not two, and you should say so in writing before you produce a single measurement report. The precedent that matters here is the engagement where an audit firm opened at a **$198M claim and closed at $12M inside the renewal envelope**: the audit was a renewal conversation in disguise, and the settlement priced only because it was absorbed into a deal IBM wanted. Settle first and negotiate the Cloud Pak afterwards and you have paid twice, once in cash and once in the leverage you burned. The sequencing detail matters enough that we treat it separately in the piece on an IBM audit opened six months before renewal. Expect IBM to resist bundling, to insist the audit is a separate compliance process run by an independent firm, and to offer a "goodwill" credit against future VPC purchases instead of a reduced settlement. A credit is not a discount. A strong outcome looks like a single signed instrument covering settlement, conversion ratios, VPC pricing and price protection, with the settlement figure at 5 to 10 percent of the opening claim and the conversion priced at the discount band you would have earned in a clean renewal, not a punitive one.

What to Do First

Everything below happens in the next 30 days, before your next scheduled call with the account team.

• **Freeze all statements of migration intent.** Infrastructure, architecture, procurement and any cloud programme steering deck. One engineer telling an IBM technical specialist that OpenShift is "the direction of travel" hands over the asset for free. Put the freeze in writing to every named contact.
• **Build the line-by-line model before you talk.** Per-product ratios, not the headline 70:1. MQ at 4:1 and MQ Advanced at 2:1 behave nothing like App Connect Enterprise at 1:3. Include bundled OpenShift entitlement and its Cloud Pak-only restriction, and set container resource limits: unbounded assignments have produced VPC counts 20 to 40 percent above the equivalent PVU footprint.
• **Build the alternative to survive questioning.** Retain PVU on Power where 120 to 140 PVU per core entitlements convert unfavourably, retain partially, or price a competitive platform. In our experience the difference between a 20 percent discount and a 40 percent discount in the same spend tier is whether the alternative is documented or merely asserted.
• **Set the disclosure trigger as a document, not a date.** You reveal willingness to convert when a written IBM commercial proposal with VPC unit pricing, ratios and multi-year protection is in your hands. Not at a meeting, not on a QBR, not because a quarter is closing. IBM's fiscal clock is a reason for them to move, not for you.
• **Name one negotiation owner and close the side channels.** Every IBM question routes through that person. Architects answer technical questions only, in writing, after review.

You reveal willingness to convert when a written proposal is in your hands, not when a meeting is in your calendar.

Related advisory: the full topic guide.

Frequently asked questions

When should I tell IBM we are considering a move to Cloud Pak?

Only after IBM has put a written commercial proposal on the table, and ideally inside an ELA renewal window rather than as a standalone product conversation. Conversions negotiated outside an ELA typically price at or near VPC list, roughly $1,000 to $2,800 per VPC per year. The signal is worth 5 to 15 discount points if you spend it as a trade rather than as an announcement.

Is the 70:1 PVU to VPC conversion ratio actually accurate?

It is a headline simplification, not the contract math. Component ratios inside a single Cloud Pak vary sharply: MQ at 4:1, MQ Advanced at 2:1, App Connect Enterprise at 1:3, WebSphere ND at parity, Db2 Advanced at half a VPC per core. Model every product line individually before you accept any IBM conversion illustration.

Does the January 1 IBM price increase mean I should sign before year end?

Not automatically. The 2026 uplift ran around 6 percent across Distributed software, Storage and Power, with IBM i SWMA at 10 percent, but an increase applied to a metric you have not yet adopted is IBM's deadline, not yours. Model the uplift cost against the discount you would forfeit by signing early; the discount delta is usually larger than the increase.

What discount should I expect on a Cloud Pak deal?

IBM ELA structures typically target 30 to 38 percent at $2M+ annual spend and 28 to 40 percent at $1M+ ACV. Accounts with no credible alternative routinely settle at 20 percent off list, while buyers with a documented migration case to AWS, Azure or a retained PVU estate reach 40 percent-plus. First offers commonly improve 25 to 40 percent when challenged with line-by-line modelled data.

Should Power Systems workloads move to Cloud Pak VPC?

Usually not without a very specific case. Power estates carrying 120 to 140 PVU per core entitlements convert unfavourably under VPC, and retaining those entitlements is often worth more than the bundle discount. Treat partial retention as a negotiating position: a contractual right to keep selected PVU entitlements alongside Cloud Pak is a real concession worth asking for.

IBM opened an audit before our Cloud Pak discussion. What now?

Treat the two as one transaction, never as separate ones. Settlements price best inside a deal IBM wants; one documented engagement moved from a $198M opening claim to $12M once folded into the renewal envelope. Do not remediate the exposure by purchasing VPC entitlements at list before the commercial terms of the wider deal are settled.