A Japanese carrier has just signed ten years of VMware. The story will reach some renewal meeting this week in the shape of "see? not everyone is leaving", and it will be used to close a conversation that should not close that way. The product annex in the release lists seven lines, and three of them cannot be installed anywhere. They are the three you will not be able to buy.
What DOCOMO signed, line by line
Broadcom's release went out over GlobeNewswire on 20 October 2026, datelined Tokyo the 19th, and states it plainly: "This is Broadcom Japan's first 10-year comprehensive agreement covering licenses, technical assistance and maintenance support for VMware products, including VCF." This is not a new relationship either: "Since 2019, VMware, now Broadcom, has worked closely with DOCOMO to build and operate its private cloud with VMware Cloud Foundation (VCF) at its core." The same span Broadcom's Paul Turner counts in the release: "Over the past seven years, we have built a strong technical collaboration."
The release closes with a one-sentence annex. It says "include", not "consists of" — the body also covers vSphere Kubernetes Service, which is absent from the annex — so these are the seven lines Broadcom chose to list, not the whole contract. Listed they are, though, and in this order: "VMware Cloud Foundation, VMware Avi Load Balancer, VMware vDefend, VMware Tanzu Platform, Support Account Manager (SAM), Technical Account Manager (TAM) and VCF Forte Support."
Four of these seven are software: the platform, the load balancer, the microsegmentation layer and the container stack. The other three — SAM, TAM and VCF Forte Support — have no installer. In Broadcom's own documentation a Technical Account Manager is a designated resource who advises and escalates on your behalf: someone who knows how your platform is built before you write to them, and who is not the one who fixes it. What carries the fixing is the third line, and the release spells it out: VCF Forte Support "offers Broadcom's highest-level service-level objectives". None of the three has an installer, and all three are signed for ten years.
The financial terms were not disclosed. That matters, because every comparison made in meetings this week will be made against a price nobody knows.
The column that is not in your catalogue
If you run six or eight sockets and renew once a year, what you buy is the left-hand column. The software. The people column does not appear in your budget because it is not in your catalogue, and neither party will bring it up on the renewal call.
That difference shows up the day something breaks. Opening a ticket in a portal and messaging someone who already knows your platform are two different experiences billed as though they were the same, and the gap between them appears in no comparison table. We took the small print of these contracts apart in what you actually buy with a 24/7 support contract, where the advertised number turned out to be a first response almost every time, not a fix.
Read backwards, that list is a dependency inventory
The four software lines are not four separate products somebody picked independently. They are four stacked layers: the platform underneath, traffic balancing above it, microsegmentation running through everything, and containers on top. Seven years from 2019 to 2026 is long enough for each of these layers to accumulate configuration nobody ever wrote down.
When we sit down to assess a migration, the first thing we ask for is not the virtual machine inventory. Moving a VM is known work and estimates well. What really moves the schedule is what hangs off the surrounding layers: load balancer rules somebody wrote in 2021 that nobody has looked at since, microsegmentation policies that exist only inside that console, a container registry that turned out to be tied to the platform. We worked out where the savings from that move come from — and where they do not — in where the savings from leaving VMware come from.
A pilot does not solve this, and it is one of the mistakes that has cost us the most months: stand up three machines on the new platform, confirm they boot, declare the migration validated. The surrounding layers never show up in a pilot because a pilot does not use them. We wrote that one out in full in piloting an alternative is not migrating.
When we do tell clients to stay
We have migrated companies from VMware to Proxmox, and we have also told companies to stay on VMware when staying made sense. The specific scenarios are written up in when NOT to migrate from VMware to Proxmox, and what we saw two years after the acquisition — that most companies renegotiate rather than leave — is in the VMware exodus that wasn't. What this week's signature adds is a test we did not have so readily to hand:
If the real cost of leaving lies in rebuilding the load balancing and the microsegmentation, and that rebuild exceeds what you save on licences over whatever horizon you have, the numbers do not work and staying is the right answer. The hypervisor is the cheap part to move; the layers above it are the expensive ones, and they are exactly what DOCOMO has just signed for another ten years. What we never accept as a reason to stay is inertia: if "it is what it is" is the only justification that surfaces in the meeting, what is missing is not a decision, it is the arithmetic.
Your own seven lines
The exercise is to write your own version of that sentence. Seven lines, one for each thing your platform depends on, with three questions beside each. Half an hour, one sheet of paper, and next year's renewal gets argued from evidence:
Two rules for reading it. If all seven of your lines turn out to be software, what you hold is a licence, and licences get renegotiated from a considerably weaker position than a contract with people in it. And the last column on the load balancer row is what sets the real schedule: it is the cell we almost never find filled in, and without it a migration does not yet have a plan, it has an intention.
Ten years against twelve months
The difference between DOCOMO and most of the companies that call us is not size, however much it looks like it. It is position. A carrier with seven years of platform behind it and ten ahead sits down to negotiate a bespoke agreement with assigned engineers; a hundred-and-twenty-person company gets a renewal email in November with a deadline attached. Both are buying the same logo and they are not buying the same thing.
The useful question this week is not "do I stay or do I go?" but "how many years of horizon do I have?" If the answer is twelve months, the strategy cannot be to wait and see, because in twelve months you will be exactly where you are today with one year less room to prepare the alternative.
Who assigns you an engineer
The people row is the one your catalogue does not sell, and the only one of the seven you can contract outside the vendor. We cover it with 24×7 IT support: an on-call rota with names on it and a phone that rings at four in the morning. We run Proxmox VE with Ceph storage in production across several datacentres, and when we take on a VMware-to-Proxmox migration we start from the inventory of the surrounding layers, not the virtual-machine inventory. And if the numbers say stay, we say stay, because consulting does not change our invoice either way: we sell neither VMware nor Proxmox licences.
If you want to run the seven-line exercise with someone who has done it before, that table is the first conversation: get in touch.
Sources. Broadcom press release, "Broadcom to Advance NTT DOCOMO's Private Cloud with VMware Cloud Foundation", distributed over GlobeNewswire on 20 October 2026, datelined Tokyo the 19th: the sentence on Broadcom Japan's first ten-year comprehensive agreement (licenses, technical assistance and maintenance support), the relationship since 2019 with VCF "at its core", and the annex sentence listing seven lines, introduced in the release with «include» rather than as a closed list, including Support Account Manager (SAM), Technical Account Manager (TAM) and VCF Forte Support. Statements taken from the same release: Hiroshi Ishimaru, General Manager of the Service Design Department at NTT DOCOMO, and Paul Turner, Chief Product Officer of the VMware Cloud Foundation Division at Broadcom. The value of the agreement was not disclosed. The description of what a Technical Account Manager is, the reading of the list as a dependency inventory, the three situations in which we recommend staying and the seven-line table are our own judgement and method, not vendor claims. We are not VMware or Proxmox resellers and we sell neither company's licences.