Every renewal conversation I have been in over the last two years starts the same way. Someone forwards a PDF, the subject line has three exclamation marks, and the first sentence of the meeting is a number. Then forty minutes disappear into whether that number is fair, which is a question nobody in the room can answer yet.

The total is the output of about six inputs. If you argue about the output you are arguing about someone else’s arithmetic. So here is the order I actually read these documents in, and roughly what each line is worth.

1. The core count, against the hardware you own

Start at the bottom of the quote, in the quantity column. Per-core subscription replaced per-socket, and every physical core in scope gets licensed, with a minimum of sixteen cores per physical CPU. That minimum is where the first silent money lives.

If your estate is full of 8-core or 12-core CPUs — and if you bought for clock speed, or you bought before 2020, it probably is — you are paying for cores the silicon does not have. On low-core-count estates this inflates the bill by roughly twenty to forty percent over the cores actually in use.

Pull the real inventory before the meeting. Not the CMDB. The actual socket and core counts per host, from the hypervisor. I have yet to see one of these quotes built on a perfectly accurate host list, and the error is never in the customer’s favour.

2. The bundle, against what you run

Second line to read: which bundle. The portfolio collapsed into bundles with VMware Cloud Foundation as the flagship, packaging compute, storage, networking and management into one per-core subscription. For a shop that bought vSphere Enterprise Plus and nothing else, that is the whole story of the increase: you are now paying for storage and networking you may not use.

The useful question is not “is VCF expensive”. It is “what would we have to adopt for this to be worth it”. If the answer is NSX and vSAN and Aria, and you have no team and no project for any of them, you are not buying a platform. You are buying a hypervisor with four unused products stapled to it.

3. The minimum order

New subscriptions carry a minimum purchase of seventy-two cores per order line. This one matters enormously if you are small and not at all if you are large. A three-host cluster of modest CPUs still buys seventy-two cores. Every DR site, every remote branch, every “just two hosts in the factory” is a separate line with its own floor.

I have seen more money lost to scattered small clusters than to the main data centre. Count your edges before you count your core.

4. The term

One, three or five years. The instinct is to take the longest term for the biggest discount. The instinct is sometimes wrong, because a five-year term is also five years during which you cannot use the one lever that actually works.

5. The lever

Which brings me to the only part of this that reliably moves a quote: a credible alternative. Not a slide with three logos on it. Credible means you can say, with a straight face and a document behind it, that you have sized the replacement, you know which workloads move first, you know what it costs to migrate, and you know the date you would start.

Opening quotes get revised when the seller believes the exit is real. They do not get revised because you said the price was high. Everyone says the price is high.

The difficulty is that building that credibility takes months, and most people start the renewal conversation with eight weeks left. Nine to twelve months is the honest lead time if you want the exit analysis to be more than theatre.

6. Only now, the total

By the time you get here you can say something specific. Not “this is a 300% increase” but “this is a 300% increase, of which roughly a fifth is the core minimum applied to CPUs we are replacing anyway, a third is bundle content we have no adoption plan for, and the rest is list movement”. That is a conversation. The other one is a complaint.

What I tell people who ask for one sentence

Fix your inputs before somebody else fixes them for you. The quote is a function, not a verdict, and almost every variable in it is something you can measure yourself in an afternoon with an inventory export and a spreadsheet.

The hardest part is not the arithmetic. It is convincing your own organisation to start the work a year before the invoice, when the invoice is not yet a crisis and therefore not yet interesting to anyone with a budget.