Architecture

A new edition, a dead product line, and a paragraph the press did not get

Eleven Broadcom posts in a week. vSphere Express is a 128-core floor rather than a discount, four capabilities quietly stopped being chargeable, Telco Cloud Platform is finished, and one licensing concession was missing from the version handed to the press.

Captured 6 November 2026

Everything below was checked against Broadcom documentation on 6 November 2026, against VCF 9.1.1. Where a figure comes from a reseller, a distributor listing or a journalist rather than from Broadcom, I say so in the line itself, because on this topic the difference matters. Broadcom's own product documentation for vSphere Express, dated 2 November, settles several points the announcement and the coverage left vague, and it is cited where it does. Sources are at the foot.

Broadcom published eleven posts to the Cloud Foundation blog between 30 October and 4 November. Most weeks that would be noise. This one is not, because two of those posts change what you are paying for, one retires a product line, and a third contains a paragraph that was missing from the version the press received.

I have spent the week reading all of it, along with what the trade press and the licensing specialists made of it, because the gap between those two readings is where the useful information usually sits. What follows is the five things worth your time, and the part of each one that nobody put in a headline.

None of this is a recommendation. It is what the documents say, what the people who negotiate these contracts for a living are reporting, and where the two disagree.

1. vSphere Express, and why it is a floor rather than a discount

The headline is that Broadcom has put a small edition back on the price list. vSphere Express is standalone vSphere, built on vSphere 9, capped at 128 cores. It carries vCenter Standard, ESX Standard, vMotion, HA, the vSphere Distributed Switch, vCenter Quick Patch, and TPM 2.0 and vTPM 2.0 support. Broadcom's own portfolio update is the primary source, and it is notably short on detail.

The detail is coming from elsewhere, and it changes the shape of the thing considerably.

A comparison of vSphere Express, vSphere Standard 8, VVF and VCF across what each includes, core cap, term and who can buy — Broadcom has published no price for Express and I am not going to repeat a distributor figure as though it were one. What is documented is the shape: the cap, the term, and what is in the box. Captured 6 November 2026.

It is 128 cores whether you need them or not

Broadcom's own vSphere Express documentation, updated 2 November, states "A maximum of 128 cores per customer" and adds that the "VCF Business Services console prevents any allocation that exceeds 128 cores per customer or site ID". So the cap is enforced in software rather than being a contractual expectation. The same page says "You cannot split the default license into smaller licenses. You cannot share the default license across sites." That means you license all 128 even if you run twelve, you cannot expand it, and you cannot spread it across two sites. C4C put the consequence plainly: "Every estate below 128 cores pays for 128." Count every CPU at a minimum of sixteen cores when you work out whether you fit. If you are at 130 to 200 cores, the uncomfortable arithmetic is that consolidating onto fewer hosts to drop under 128 may be cheaper than the alternative.

You may not be allowed to buy it

Per reseller briefings, the eligible buyer is a new customer, meaning an organisation that has never bought VMware or has not bought in the last three years. Existing customers need Broadcom's authorisation, and an order can be rejected after it is placed. Broadcom has published no eligibility rules at all, which means the only way to find out where you stand is to ask your partner before you build a plan around it.

Cheaper is doing a lot of work in that sentence

Express is being described everywhere as the affordable option, and Broadcom says only that it sits significantly below VVF and VCF. Numbers are circulating from distributor listings and reseller guides. I am not going to repeat them, because none of them came from Broadcom and a figure with my name next to it will outlive the caveat attached to it. What I will say is that a one-year term and a hard cap change the comparison in ways a headline number does not capture, and that the only figure worth planning on is the one your partner puts in writing for your core count. Techzine reached a similar conclusion, and also set Express against alternatives priced per socket rather than per core, which is a comparison that needs normalising to your own host count before it means anything.

What is not in it

Broadcom states it directly: "No add-ons are available for vSphere Express", and "With vSphere Express, you use VCF Operations only for licensing your environment." The licensing overview lists the 9.1 add-ons that Express therefore cannot have, which are vSAN by TiB and Private AI Foundation with NVIDIA by cores. Mandatory components are ESX, vCenter, VCF Operations and a license server. The feature-level exclusions circulating from reseller briefings, which Broadcom has not published, are DRS, memory tiering, SR-IOV, GPU and vGPU, Proactive HA and live patching. If your consolidation ratio depends on DRS, or your refresh plan depends on memory tiering, Express is not the product you are looking for.

There is a documented upgrade path, and a documented way out

This is the part the coverage missed entirely. Broadcom documents the route from vSphere 8 into Express: the target is "9.1.1 and later", with minimum source versions of vCenter Server 8.0 and ESXi 8.0, optionally Aria Operations 8.18.6 and vSphere Replication 9.0.2.2, and a prescribed upgrade order that starts with vSphere Replication and ends with virtual machine compatibility. Express is version 9 or later only, and version 8 license keys are not available for it. The way out is documented too. An existing vSphere Express environment can be converged to VCF or extended to vSphere Foundation, so it is not the dead end the core cap makes it look like. One operational obligation worth putting in a runbook rather than discovering: licence usage reports are required at least once every 180 days to maintain your licences, and after a licence expires you have 90 days to update it.

Nobody agrees on when you can have it

Broadcom's post says Express is "available immediately in all regions." StorageReview, Network World and Techzine all say available now. TechTarget reports it will be available in all regions by the end of January 2027. One of those is wrong and I cannot tell you which from the public record. Do not put a date in a plan based on any of them.

2. Four things stopped being chargeable, and that is the bigger story

If you read only one part of this week, make it this one. Both Network World and SiliconANGLE report that capabilities previously sold as advanced services have moved into core VCF. Broadcom's own post does not frame it that way, which is why it has had far less attention than the new edition.

Four capabilities moving from the Advanced Services column into core VCF, with native disaster recovery also landing in VVF — Reported by Network World and SiliconANGLE on 4 November 2026. Broadcom's own post describes the capabilities without framing the entitlement change. Captured 6 November 2026.

Site Recovery Manager, VPC policy-based connectivity, infrastructure configuration and compliance management, and confidential computing are now described as part of the platform rather than as things you add to it. Native disaster recovery is landing in both VCF and VMware vSphere Foundation, which is the detail most likely to matter to a mid-sized estate that was never going to buy the full stack.

Why this is a renewal conversation and not a feature announcement

If four line items have just become part of the platform, then any quote prepared before this week is pricing something different from what is now being sold. That is not a reason to celebrate or to complain. It is a reason to ask your account team, in writing, which of these four are included in the SKU you are being quoted, in which release, and whether the figure you were given reflects it. Confidential computing is already GA in 9.1.1. The rest are tied to the next release, currently estimated for Q1 FY27, which ends in early February 2027.

3. The paragraph that was not in the press version

Broadcom's post carries a paragraph for customers in the European Economic Area. vSphere Standard version 8 and vSphere Enterprise Plus version 8 remain available for sale there until 31 October 2027, and Extended Support is included in any active subscriptions to those offerings until the earlier of the subscription end date or 31 October 2028.

The Register reported on 5 November that this paragraph was not in the version of the post it received, and was not mentioned in the briefing. Its earlier write-up of vSphere Express was based on the version without it.

Why it matters even if you are not in Europe

General support for vSphere 8 and VCF 5.x ends on 11 October 2027. This concession runs more than a year past that date, for one region only. The Register links it to regulatory pressure without confirming the connection, and notes the open European Commission antitrust investigation prompted by CISPE into whether VMware's channel changes distorted the cloud infrastructure provider market, along with Dutch courts enforcing a VMware support contract for a government agency.

Australia and the UK are not in the EEA

Neither is anywhere else I usually work. Michael Warrilow's reading, quoted in the same piece, is that customers elsewhere could seek similar concessions by working through industry associations, user groups and national regulators. He is also realistic about the value of it: "Ultimately, all it does is buy a little more time to upgrade or migrate." Worth knowing it exists, because the first step in asking for something is being able to point at somebody who already has it.

4. Telco Cloud Platform is finished, and VCF Telco replaces it

This one went almost unremarked outside the telco press, and it is the clearest lifecycle statement of the week. Broadcom's VCF Telco announcement on 2 November says that rather than continuing to develop Telco Cloud Platform as a standalone platform, telco capability now ships inside VCF. Telco Cloud Platform licensing SKUs reached End of Availability on 2 November 2026.

Existing TCP subscriptions remain supported through the transition, with realignment handled at renewal. If you buy the VCF SKU you get the telco features, present and future.

What VCF Telco actually carries

VNFs and CNFs on one platform rather than separate network and IT clouds. A single management domain spanning data centre to edge with zero-touch lifecycle. Private AI Services built in, including support for fully air-gapped deployment for sovereign AI. Two-year Long-Term Support with every minor release of Kubernetes, which is a meaningful commitment if you have lived through Kubernetes upgrade cadence in a regulated environment. Enhanced NVMe memory tiering that removes the hardware RAID dependency, and next-generation compression with global deduplication. Policy alignment to NIST and the UK Telecommunications Security Act, with customer-controlled encryption keys and audit-grade evidence.

The question to ask if you are a TCP customer

Broadcom gives no version number and no availability date for VCF Telco, and says only that telco capabilities keep growing with every release. If your TCP renewal falls inside the next twelve months, the thing to establish now is which of your current TCP capabilities exist in VCF today, which are scheduled, and which have no equivalent. That is a gap analysis, and it is better done before the renewal conversation than during it.

5. VKS 3.8, and the upgrade order that will catch people

Released 22 October, so slightly outside the week, but it lands on anyone moving to 9.1 and the release notes contain two prerequisites that are easy to miss.

VKS 3.8.0+v1.37 brings Kubernetes v1.37, Cluster API v1.14 and CAPV v1.17. It requires a minimum Supervisor Kubernetes version of 1.33, and the Supervisor must already be on VKS 3.5.0 or later.

Two things to check before you touch Supervisor 9.1

First: all VKS clusters must be on ClusterClass 3.6.0 or later before you upgrade to Supervisor 9.1. Second: the builtin-generic ClusterClasses 3.1.0, 3.2.0 and 3.3.0 have been removed, so any custom ClusterClass depending on them needs migrating first. Neither of these fails loudly at the point you would want it to.

The useful new things

Chained upgrades across up to three Kubernetes minor versions in one operation with automated pre-checks, which is the first time this has been practical at fleet scale. Configurable in-place updates through a new inPlaceUpdateStrategy variable, defaulting to MinimizeSpareNodes. Control plane zone selection via failureDomainSelector, which needs VCF Supervisor 9.1.1.0100 or later. CRD-based API server audit configuration. AIDE-based file integrity monitoring on Linux nodes for DoD STIG compliance. VLAN-backed VPC workload support, which needs VCF 9.1.1.

Immutable Photon OS will break somebody's runbook

The new image locks /usr with a read-only filesystem and restricts direct SSH access. If any of your operational procedures depend on SSH to a node, migrate those first rather than discovering it during a change window. Relatedly, usrPartitionMode: ReadOnly is deprecated and rejected for new clusters.

Scale limits that are documented but not prominent

With up to 300 clusters, stay under 600 Helm add-ons. With up to 500 clusters, stay under 100. Exceed either and the add-on manager can run out of memory. There is also a known issue where an unreachable OCI Helm registry delays reconciliation for every other AddonRepositoryInstall resource, which is the kind of dependency worth knowing about before it is load-bearing.

The rest of the week, briefly

VMware Compliance and Configuration Management, a new Advanced Service built on the Salt remote execution engine that Broadcom now sells standalone. CIS, DISA and customer-defined benchmarks, drift detection and automated remediation across VMs and containers.
VMware Live Recovery added to Advanced Services. The on-premises cyber recovery workflow replicates across an air gap into an isolated vSAN clean room, with network isolation and EDR scanning before anything returns to production.
VMware HyperArmor, announced at Explore on Tour Frankfurt on 13 October, added to the catalogue on 4 November. Virtual patching for ESX, blocking of unauthorised software, MFA on the SSH login path. GA expected Q1 FY27. SDxCentral has the detail.
VMware Cloud on AWS moves to 9.1. GA expected during calendar 2027, fleet upgrades running throughout the year, no customer action required, and staying on 1.26 is not an option. All clusters in an SDDC must run the same version.
AI Infrastructure Quickstart, a vCenter plugin that folds cluster deployment, Supervisor activation and VKS provisioning into one workflow. Next release.
vSAN native S3-compatible object storage moves from tech preview to GA in the next release, with self-service provisioning through vSphere Supervisor or VCF Automation.
Demystifying VCF Automation Node Architecture and Load Balancing, a technical deep dive published the same day as the portfolio update and almost entirely overshadowed by it.
Three technical explainers on 30 October, on isolating infrastructure management, confidential computing, and why memory tiering is not simply swapping to NVMe.

What did not happen

No new security advisory. The current one is still VMSA-2026-0006 from July. No new component patch set either: the most recent is 12 October, and the patch release page restates something worth repeating, which is that patch releases are not synchronised across VCF components and do not require a new Bill of Materials. You can apply only the components that apply to you. Any tool or tracker showing 9.1.1 as a single flat version is giving you a version number that does not exist in your estate.

Where I could not get to the bottom of it

Broadcom's own product documentation settles more of this than the announcement or the coverage does, which is worth saying because none of the trade pieces reference it. The mechanics of Express are documented. What is left open is commercial.

The price. Broadcom has published none. The figures circulating come from distributor listings and reseller guides, not from Broadcom, which says only that Express sits significantly below VVF and VCF.
Who is allowed to buy it. The new-customer rule and the post-order rejection both come from reseller briefings. Broadcom documents the technical cap but publishes no eligibility criteria.
When you can have it. Broadcom says immediately in all regions. TechTarget says by the end of January 2027.
And one on the other story. Whether the four capabilities moving into core VCF carry any change to the VCF SKU price. Neither Broadcom nor the trade press addresses it, and it is the question with the most money attached.

What I would actually do with this

If you have a VCF quote dated before 4 November, ask in writing which of the four capabilities it includes and whether the number changes.
If you are under 128 cores and not already a VMware customer, ask your partner about Express in writing, and ask specifically about eligibility and about the availability date.
If you are between 130 and 200 cores, work out what consolidating under 128 would cost you operationally before you price VVF.
If you hold Telco Cloud Platform, start the gap analysis against VCF now rather than at renewal.
If you are heading to Supervisor 9.1, check every VKS cluster is on ClusterClass 3.6.0 or later first.
If you are in the EEA, confirm the vSphere 8 extension applies to your agreement. If you are not, it is still a useful precedent to have in your pocket.

The pattern across all of it is the same one I keep running into. The announcement tells you what is new. The release notes tell you what breaks. The resellers tell you what it costs. You need all three before any of it reaches a plan, and this week none of the three agreed with the other two.

Tools that go with this

Most of the above only matters once you know what you are actually moving. Estate Readiness Triage sorts your clusters into what can move as is and what has to be replaced first, and Upgrade Runway works backwards from your end of general support date to the latest date you could still start.

Free to use, including with customers. Nothing you enter leaves your browser. Everything else on VCF.