VMware After Broadcom: Hyper-V, Nutanix, Azure

If you're running VMware vSphere and your renewal is in the next 12 months, you've already had the meeting. The Broadcom-era pricing came in 6x, 8x, sometimes 10x what you paid before. Your CFO has a question. Your IT director has a different question. The CIO is in the middle.

This post isn't a Broadcom hit piece — there are plenty of those. It's a senior engineer's honest assessment of the six real destinations a mid-market VMware shop has — five hypervisors plus a public-cloud exit — with the math on when each one wins. We don't resell any of them, so the recommendation in any specific case isn't paying us anything extra either way.

What changed, in two sentences

Broadcom collapsed the VMware product catalog from 168 SKUs into four bundles, eliminated perpetual licenses, and moved everyone to subscription-only with 16-core-per-CPU minimums. The result for mid-market customers — typically 30 to 200 VMs across 4 to 16 hosts — is renewal pricing 800% to 1,500% higher than what they were paying, in deals we've seen and in industry coverage.

The largest enterprises absorbed it. The smallest customers were never economical for Broadcom and got dropped or repriced. The mid-market got squeezed in the middle.

As of mid-2026, we're still seeing post-Broadcom renewal quotes commonly land at 2–5x prior pricing once VCF-only bundling is applied — an observed pattern across the deals crossing our desk, not a Broadcom-published figure. The pressure hasn't eased; if anything, the "transition pricing" cushions on the earliest renewals are now expiring.

Six real exit destinations — five hypervisors plus a public-cloud option

1. Microsoft Hyper-V

The default choice for most Microsoft-stack shops. If you run Windows Server Datacenter, you've already paid for Hyper-V — there's nothing to buy. Failover Clustering, Storage Spaces Direct, Live Migration, Replica, shielded VMs, and Virtualization Based Security are all native and mature.

When Hyper-V wins:

• Most workloads are already Windows-stack
• Your team knows AD, GPO, PowerShell — Hyper-V management is in the same toolset
• You're already on Windows Server Datacenter (or upgrading anyway)
• You want native Azure Arc and hybrid cloud connectivity
• Operational simplicity matters more than feature parity with vSphere

Where it loses: heavy Linux workloads with VMware-specific tooling. Some niche features (vSAN stretched clusters, NSX-T microsegmentation) don't have direct Hyper-V equivalents — you architect around them.

2. Azure Local

If you evaluated this as Azure Stack HCI, it is the same product. Microsoft renamed it to Azure Local — its old documentation URL now redirects to the Azure Local docs — so quotes, notes and vendor decks written before the change will still carry the old name.

Microsoft's hyperconverged infrastructure for on-prem deployment with an Azure-managed control plane. It's Hyper-V plus Storage Spaces Direct plus Software-Defined Networking, all packaged with Azure-native management. It's the closest like-for-like to vSAN that exists.

When Azure Local wins:

• You want vSAN-class hyperconvergence on-prem
• You're already invested in Azure (subscriptions, monitoring, governance)
• You want hybrid Azure connectivity by default
• Edge or branch deployments where centralized Azure-side management is a feature, not a bug

Where it loses: if you don't want Azure dependency at all. The control plane is Azure-managed, which means losing Azure connectivity means losing some management functions (workloads keep running, but you can't manage them as cleanly).

3. Nutanix

The longest-tenured VMware-alternative HCI vendor. Strong feature parity for the workloads where vSAN was the reason you stayed on VMware. Mature multi-hypervisor support (AHV is their default; ESXi and Hyper-V also supported).

When Nutanix wins:

• You were on vSAN and the HCI model is non-negotiable
• You want a vendor with explicit VMware-migration tooling and motion
• You want Prism Central as a single management plane
• VDI workloads (Citrix or Horizon) — Nutanix has a strong story here

Where it loses: Nutanix licensing is its own complexity, and customers who picked it as a "Broadcom revenge" play sometimes find the Nutanix renewal math is also higher than expected. Check the 3-year and 5-year TCO explicitly.

4. Scale Computing

The dark horse of the Broadcom-exit story. Scale Computing has reported a sharp uptick in new customers post-Broadcom, almost all of them VMware refugees in the SMB and mid-market segment. Their pitch is radical operational simplicity — one platform, one license, no separate compute/storage/networking SKUs, no per-core minimums.

When Scale Computing wins:

• You're SMB or smaller mid-market (typically <100 VMs)
• Operational simplicity is the actual goal — not feature breadth
• You have edge or branch sites where minimal management overhead matters
• The TCO math is where Scale wins explicitly — particularly under ~50 hosts

Where it loses: heavy enterprise feature requirements. If you need vSAN stretched clusters with NSX-T microsegmentation and DRS-affinity rules, Scale isn't trying to be that.

5. Proxmox VE

Open-source KVM-based hypervisor with a real management UI and mature clustering. Has gained serious credibility in the VMware-exit conversation because it's free at the software layer (commercial subscription is available and recommended for production support).

When Proxmox wins:

• Heavy Linux workloads with strong in-house Linux expertise
• Cost is the dominant factor and you have ops capacity to handle "less polish"
• Lab, dev, and non-critical workloads
• You're philosophically committed to open source

Where it loses: if your operations team isn't comfortable with Linux + open-source troubleshooting, Proxmox in production becomes a liability. The community is active but you're more on your own than with a commercial vendor.

6. Azure IaaS (full lift-and-shift to public cloud)

Not really a hypervisor decision — you're exiting on-prem virtualization entirely. Right answer for some workloads, wrong answer for others. The cost model flips from CapEx-on-renewal to OpEx-monthly, which changes the math entirely.

When Azure IaaS wins:

• You're already running significant Azure workloads
• You're hardware-refresh-due AND VMware-renewal-due in the same cycle
• Workloads have variable demand (auto-scaling pays off)
• You want to retire on-prem datacenters entirely
• Disaster recovery and business continuity are the driver

Where it loses: steady-state predictable workloads. Always-on production VMs in Azure typically cost more per year than well-utilized on-prem hardware. Right-sizing matters enormously — most VMs are 2–3x over-provisioned in their original sizing.

The decision matrix, simplified

Update — September 2026: a patch you may not be entitled to, a new reason to stay, and an EU case that has not been decided

Three things moved since this post was written. They matter to different readers, so take the one that applies to you.

If you are still on vSphere 7: the August advisory revision is not good news for you

Broadcom revised the advisory discussed above. The change log entry reads, verbatim:

“2026-08-19 VMSA-2026-0006.2 Added ESX/vCenter 7.0 patch guidance for Critical severity issues.”

At first glance that looks like relief — 7.0 was not covered, now it is. Read it more carefully. vSphere 7.0 reached End of General Support on October 2, 2025. Guidance being added for an out-of-support release does not mean a patch you can go and download; it means patch guidance for customers with extended support. Broadcom's own end-of-general-support article does not offer a self-service remediation path for 7.0. It points at the maintenance policy handbook and otherwise recommends moving to 8.x.

So the practical position for an out-of-support vSphere 7 estate with a CVSS 9.8 vCenter flaw is: the fix exists, and your entitlement to it is a commercial question rather than a technical one. That is a materially different conversation from “apply the patch,” and it is worth having before you need it rather than during an incident. If you are on 7.0 without an extended support agreement, the remediation path for a critical vCenter vulnerability runs through procurement.

Broadcom is now selling VCF as somewhere to come back to, not just somewhere to stay

On August 31, 2026 Broadcom announced VMware Private AI Cloud and VMware AI Factory on VMware Cloud Foundation, alongside certified VCF AI ReadyNodes from Cisco, Dell, Lenovo and Supermicro. The positioning line is the interesting part:

“a more secure, scalable, and flexible approach to AI that brings the model to the data, not the data to the model.”

Strip the marketing and that is a repatriation pitch: run AI workloads next to data you already hold on infrastructure you already own, instead of exporting the data to a hyperscaler. It is the first argument for staying on VCF we have seen that is not a price argument — and price is the argument that has been losing. Whether it changes any mid-market decision is a separate question, and for most of the estates we see the answer is no: if you are leaving because of licensing cost and portfolio consolidation, an AI-adjacency story does not address that. But it tells you where Broadcom intends to compete next, and it means the “stay” column in your own analysis should not be assumed static.

Neither announcement states a general-availability date. We looked, in both releases. Treat this as direction of travel, not as something you can buy into on a schedule.

The EU case: what has actually happened, and what has not

There is European Commission activity around VMware licensing, and it is being reported loosely enough that it is worth stating precisely what has and has not occurred.

What happened: the Commission issued a decision requiring Broadcom and VMware to produce documents under case AT.40924. Broadcom challenged that request at the EU General Court (case T-280/26, action brought May 6, 2026) and applied for interim measures to suspend it in the meantime. On August 3, 2026 the court refused the interim-measures application only.

What has not happened: there is no infringement decision, no fine, no finding of abuse of dominance, and no ruling on the merits of the underlying document dispute. The annulment action and the Commission's investigation are both still open and undecided.

Update — September 2026: the migration kit went dark

A different kind of change hit the exit-planning calculus this month, and it does not touch pricing at all. The VMware vSphere Virtual Disk Development Kit — VDDK, the library that lets a tool read a VMware virtual disk from outside the hypervisor rather than going through the guest OS — stopped being available from Broadcom's public download pages sometime around late August 2026.

ShapeBlue, who documented the change, checked the outage from five vantage points before publishing: “Every URL above was checked on 25 August 2026, from Europe and through exit nodes in the United States, Brazil, Finland and Japan.” Their guidance for anyone with an active entitlement: “Customers with active VMware entitlements should check their Broadcom Support Portal access and contact their Broadcom account or support team for an authorised download route.”

That last sentence is doing real work. Broadcom has not announced VDDK is discontinued — we are not claiming otherwise — but the self-service path that most migration teams relied on is gone, and the replacement, if there is one, runs through a support ticket rather than a download page.

Why this is not a footnote — VDDK sits underneath the tooling itself

Network World reported on 11 September 2026 what that dependency actually means in practice: “VDDK is, in fact, a vital part of tools used to migrate away from VMware. For example, it is used by Microsoft Azure Migrate, Red Hat's Migration Toolkit, Nutanix Move, and VMware-to-KVM migration products.” We're citing that VDDK is a dependency of those tools — not making any claim about how each one is affected or what workaround, if any, each has built.

Microsoft's current Azure Migrate appliance page still tells you to go get VDDK from the place that no longer serves it. It says, verbatim: "Install the VDDK: The appliance checks if the VMware vSphere Virtual Disk Development Kit (VDDK) is installed. Download VDDK version 8.0 or 9.0 from the Broadcom Developer portal." Microsoft has not removed that instruction — it is still the documented step, pointing at a portal that, per ShapeBlue, is not serving public downloads.

Microsoft's own fallback is in the same document: "VMware vSphere Virtual Disk Development Kit (VDDK) packages are required for agentless migration. If you already have a supported VDDK package, proceed with agentless migration. If you don't have access to a supported VDDK package, use agent-based migration." That is the honest state of Azure Migrate today — agentless if you already hold the package, agent-based if you do not.

Red Hat's guidance for its Migration Toolkit for Virtualization is more direct about where responsibility sits. From solution 7146995, updated 27 August 2026: "Red Hat is aware that VMware VDDK images are currently unavailable to end-users via standard public downloads." And on why Red Hat itself is not a fallback: "The VDDK image is proprietary software owned, shipped, and deployed by Broadcom. Due to licensing restrictions, Red Hat cannot host, distribute, or provide this image directly to customers." Red Hat's stated remediation path is the same one ShapeBlue points to: "To obtain the necessary VDDK image, you must contact VMware/Broadcom Support directly and request access to the image." Red Hat also names one workaround, hedged: "In the mean time, a workaround using Storage offloading/Xcopy could suffice. However, this requires your backend storage to support such features."

The planning consequence

Agentless replication is the usual plan for an Azure Migrate move — it's faster to stand up and it doesn't require installing anything on the source VMs. Agent-based migration is the fallback path, and the fallback has a real cost: an agent goes on every source VM, which means more change windows, more access requests, and more elapsed time than the agentless plan assumed.

If your migration plan assumes agentless replication and you do not already have a supported VDDK package in hand, that assumption needs re-checking now — before the renewal clock forces the timeline. Two actions worth taking this week: confirm whether you already hold a VDDK 8.0 or 9.0 package from an earlier download, and if you don't, open the request for an authorised download route with Broadcom support early rather than during a cutover window, per both ShapeBlue's and Red Hat's guidance above.

What we recommend doing this week

1. Get an honest VM inventory. Workload type, OS, vCPU/RAM/storage actually used (not allocated), I/O profile, dependencies. Most shops haven't run this exercise in 3+ years and the answers reshape the destination calculus.
2. Get the renewal quote in writing. Including the bundled SKU details and the 3-year math, not just year one. Some VMware partners are quietly providing "transition pricing" that artificially flatters the early years.
3. Run a vendor-neutral assessment. One that doesn't end with the assessor's preferred destination. The right answer for your environment is determined by your workloads and your operating model, not by what the consultancy resells. (We're labor-only specifically because we got tired of the bias problem.)

Common mistakes we see

Picking from a sales deck

Three vendors will pitch you. Each will have a deck explaining why they're the obvious answer. They might all be wrong for your environment. The assessment work — VM-by-VM workload analysis — is non-negotiable. If a vendor wants to skip it, that's the signal.

Underestimating the operations transition

"It's just another hypervisor" is a sentence that costs companies millions. Hyper-V management is not vSphere management. Nutanix Prism is not vCenter. Scale's HC3 is genuinely simpler — and that's a feature for some teams and a culture shock for others. Plan for retraining and documentation explicitly.

Migrating before retiring

VMware migrations are the perfect moment to identify workloads that should just be turned off. Most VMware estates have 15–30% of VMs that are running but doing nothing useful — old test rigs, abandoned dev environments, "just in case" servers. Migrate the ones that matter, decommission the rest. Don't pay to move what shouldn't move.

Buying the destination before the assessment

Easy mistake when the renewal pressure is high. CFO says "we're not paying that" and IT signs a Nutanix deal in two weeks because Nutanix's salesperson called first. Six months in, you discover half the workloads should have gone to Azure IaaS instead. The assessment first, the destination second.

Side-by-side comparison matrix

Related reading

• For a modelled 90-day timeline of a mid-market Broadcom-renewal exit, see After Broadcom: 250-VM Hyper-V migration in 90 days.
• For the head-to-head Hyper-V vs Azure decision at 500-VM scale, see VMware to Azure vs Hyper-V: 500-VM decision guide.
• If hardware refresh is also in scope, the Exchange 2019 EOL post covers the parallel decision tree for email infrastructure on the same hardware fleet.
• Same fiscal year, same hardware refresh: if Server 2019 is aging out alongside the VMware renewal, see Windows Server 2019 EOS: real upgrade options for 2026.
• If Server 2019 is also out of mainstream support, see the Windows Server EOS service.
• For tenant identity work that often follows a virtualization migration, see Entra ID Hybrid.

Sources and further reading

• Broadcom — VMware Cloud Foundation product page
• Microsoft — Azure Local documentation
• Nutanix — VMware alternative resources
• Scale Computing
• Proxmox VE
• ShapeBlue (25 August 2026, updated 26 August) — Broadcom removes VDDK download pages without explanation
• Microsoft Learn — Set up an appliance for servers in a VMware environment
• Red Hat solution 7146995 (updated 27 August 2026) — Unable to download VMware VDDK images for MTV
• Network World (11 September 2026) — Broadcom hampers VMware migration by blocking downloads of key SDK

The 30-second version

Six real destinations, each right in different scenarios. Hyper-V for Microsoft-stack shops. Azure Local for vSAN feature parity with hybrid Azure. Nutanix for VDI-heavy or vSAN-mature environments. Scale Computing for SMB and operational simplicity. Proxmox for Linux-heavy and cost-driven. Azure IaaS when hardware refresh and renewal align.

The right answer for your environment is determined by your workloads, your team, and your operating model — not by which vendor's salesperson called first. A 2-week vendor-neutral assessment surfaces the answer with data.

To start that conversation, the project intake form takes about three minutes. We'll come back with scope.

Pro IT NW does not resell VMware, Hyper-V, Nutanix, Azure Local, Scale Computing, or any other virtualization platform. We charge for the engineering. The recommendation is what fits.