On July 6, 2026, AWS made a move that reshapes the hybrid cloud chessboard for Windows-heavy enterprises: Amazon Elastic VMware Service (EVS) now supports VMware Cloud Foundation (VCF) versions 9.0 and 9.1. But the real headline isn’t just the version bump—it’s that customers can provision EC2 bare-metal hosts directly into their own Amazon Virtual Private Cloud (VPC), marrying the control of a self-managed datacenter with the operational ease of a managed service. For Windows administrators staring at decaying on-prem hardware and increasingly complex licensing stacks, that changes the math.
What Actually Changed: EVS Gets a VPC Transplant
Amazon Elastic VMware Service isn’t new. First launched in 2024, EVS offered a way to run VMware workloads on AWS without the overhead of managing the underlying cloud plumbing—think of it as a halfway house between fully self-managed VMware on EC2 and the more lock-in-prone VMware Cloud on AWS. But until now, the service’s biggest limitation was its abstraction: the physical hosts lived inside an AWS-managed network boundary, meaning you couldn’t stitch them seamlessly into your own VPC fabric.
That changes with VCF 9.0 and 9.1 support. The updated EVS deploys EC2 metal instances—likely high-density i3en.metal or i4i.metal types, though AWS hasn’t specified an exclusive list—into a VPC you control. You define the subnets, security groups, routing tables, and even the IP address ranges. Your VMware VMs inherit that networking, so on-premises VPN or Direct Connect links can extend into the cloud without unnatural NAT contortions. For the first time, a VMware vCenter cluster inside EVS can be a proper neighbor to your RDS databases, your legacy .NET application servers on EC2, and even your Active Directory domain controllers—all under a single, familiar network topology.
Under the hood, VCF 9.0 brings vSphere 8.0 Update 2, an updated vSAN, and improved DRS that can rebalance workloads based on memory usage across a heterogenous cluster. But on AWS, these aren’t just feature bullet points—they’re the foundation for a hybrid operating model where you can live-migrate VMs between on-premises and cloud hosts (assuming sufficient bandwidth) using VMware HCX. And because EVS is AWS-managed, the underlying host firmware, hardware health, and VMware software patching remain someone else’s problem.
Windows workloads, which represent the silent majority inside most enterprise VMware clusters, stand to gain the most from this new architecture. Here’s why.
What It Means for You: Windows Licensing, Migration, and the Day-2 Reality
For the Windows admin who’s been told to “migrate to the cloud but don’t break anything,” EVS with VCF 9.0 suddenly becomes a less disruptive lift-and-shift vehicle. Consider these three scenarios:
1. License Mobility Finally Makes Sense
Microsoft’s per-core licensing model for Windows Server (2016 and later) means you need licenses for every physical core on the host where the VM runs. In most public cloud contexts, that obligation is satisfied when you buy a cloud VM with Windows Server licensing included. But large enterprises often own Windows Server Datacenter licenses through Software Assurance, which grants license mobility rights—you can deploy those licenses on authorized dedicated hosts in the cloud.
EVS on EC2 metal qualifies as a “dedicated host” from Microsoft’s perspective, because you’re renting the entire physical server. By deploying EVS inside your own VPC, you not only meet the licensing terms but also can use AWS License Manager to track core allocations and avoid compliance gaps. That’s a big deal for organizations facing a Microsoft audit. For example, an i3en.metal instance with 96 physical cores would require 96 cores of Windows Server Datacenter licensing, plus the same number of cores for System Center if you use it. The math is predictable in a way that shared-tenancy cloud VMs often are not.
2. Hybrid Windows Workloads Stay Coherent
Picture an on-premises vSphere cluster running a .NET Framework web app that depends on a local SQL Server instance and an Active Directory domain for authentication. In a traditional VMware Cloud on AWS migration, you’d have to re-IP everything and deal with forced domain trust changes. With EVS, you extend the on-prem VPC (or a linked AWS VPC) into the cloud, keep the same IP scheme, and simply move the VM. The AD domain controllers, whether left on-prem or also migrated, remain reachable over a private, low-latency connection. No DNS headaches, no Kerberos problems.
That coherence extends to modern Windows containers too. VCF 9.0 integrates with vSphere Pod Service (formerly VMware Tanzu Kubernetes Grid Service), and AWS has its own EKS Anywhere for hybrid clusters. If you’re running Windows container hosts on evicted VMs, the ability to pivot between on-prem and AWS resources under one vCenter umbrella cuts down on cognitive load for your operations team.
3. The Operational Model Your Team Already Knows
One underappreciated fact about large Microsoft shops is that they’ve often built entire careers around vCenter and the VMware toolchain. PowerCLI scripts, vSphere tags, vRealize Operations dashboards—these are the tools of the trade. Dragging them into a “cloud-native” refactor is expensive and risky. EVS with VCF 9.0 gives you the same vCenter Server interface, the same APIs, and the same policy-driven management overlay you use on-prem. AWS handles the hypervisor host layer, but you keep the control plane. That means your existing backup solutions (Veeam, Commvault) and monitoring tools (SCOM, SolarWinds) can plug right in, provided they have network reachability—which the VPC arrangement guarantees.
How We Got Here: Broadcom, the VCF Strategy, and the Hybrid Imperative
To understand why this announcement matters, you have to rewind to 2022. Broadcom’s then-pending acquisition of VMware sent shockwaves through enterprise IT, and the subsequent licensing changes—the shift to per-core subscription licensing for all VMware products—pushed many organizations to explore alternatives. AWS, which already operated VMware Cloud on AWS (a product managed by VMware/Broadcom), saw an opening to offer something more AWS-native.
The result was Amazon Elastic VMware Service, launched in late 2024. Initially, EVS supported VMware Cloud Foundation 5.x and offered a simplified deployment model but kept customers at arm’s length from the VPC. It was a stopgap for those fleeing Broadcom’s new pricing but still needed VMware compatibility. Now, with VCF 9.0/9.1, AWS is sending a clear signal: “You don’t have to choose between VMware’s software stack and AWS’s network fabric.”
This dovetails with Microsoft’s own evolving cloud story. Windows Server 2025 (released October 2024) brought per-core licensing complexity to new heights, especially for those using Azure Hybrid Benefit on-premises. Meanwhile, Azure VMware Solution (AVS) continues to mature, offering a comparable managed VMware experience inside Azure datacenters. But AWS’s aggressive move to let customers own the VPC architecture could tip the scale for shops that have already standardized on AWS for other workloads and don’t want to consume Azure just for VMware.
The broader trend is clear: the cloud providers are no longer treating VMware as an afterthought but as a first-class hybrid citizen because it’s how the world’s largest enterprises run their Windows Server farms.
What to Do Now: Four Steps for Windows-Centric VMware Shops
If you’re reading this on a Monday morning with a coffee in hand and a mountain of Windows Server 2019 VMs to migrate, here’s a concrete action plan.
Step 1: Audit Your Current VCF Version and Windows Licensing Position
Are you already on VMware Cloud Foundation 5.x or later? If so, a direct upgrade path to 9.0 exists via VMware’s SDDC Manager. If you’re on older vSphere, you may need to plan a greenfield deployment. On the Windows side, use the Microsoft Assessment and Planning (MAP) Toolkit to inventory the number of VMs, their host allocations, and whether you’re using Datacenter or Standard edition. This output will feed directly into AWS License Manager if you go the BYOL route.
Step 2: Model a Hybrid Network Architecture
Involve your networking team early. Map out which VLANs/subnets you’d extend to AWS via Direct Connect or VPN. Plan for a stretched vCenter if you need live migration, or use HCX bulk migration if cold cutover is acceptable. Ensure your Active Directory sites and services topology is updated to reflect the new AWS location—otherwise you’ll have slow logons and replication issues.
Step 3: Engage AWS and Microsoft for License Guidance
For Windows Server BYOL, you’ll need to submit a license verification form to Microsoft or work with a licensing specialist (LAR). AWS has license mobility verification processes that integrate with AWS License Manager. Ask AWS specifically about “EVS dedicated host core counts” to match your Windows licensing cores exactly. For example, if you buy five i3en.metal instances with 96 cores each, you need 480 cores of Datacenter licensing—no more, no less. Misalignment here can cost tens of thousands in non-compliance penalties.
Step 4: Test with a Low-Risk Workload, Then Scale
Pick a single Windows Server 2022 VM—maybe a utility server or a secondary domain controller—and migrate it using HCX into an EVS test VPC. Validate application performance, RDP latency, and backup integration. Once comfortable, move a larger slice of the environment, keeping rollback snapshots. AWS’s standard EVS SLA covers the infrastructure; you’re responsible for the guest OS, so configure Windows patching via your existing WSUS or cloud-based patching tools.
Outlook: The Windows-Licensing Chess Game Isn’t Over
AWS isn’t doing this out of charity. By giving EVS customers VPC-level control, it’s making a hard run at the data center workload that has stuck stubbornly on-prem: the monolithic Windows Server application that powers a manufacturing plant or a hospital system. Those workloads need VMware’s reliability and Windows Server’s ecosystem, and now they can get both on AWS without sacrificing networking sovereignty.
But watch for countermoves. Microsoft could tighten the rules around BYOL on non-Azure clouds—something it has done before with terms that favor Azure Dedicated Host over AWS. At the same time, Broadcom’s VMware Cloud Foundation subscription model is still evolving; future price hikes could dampen the economics of running EVS at scale.
For now, Windows admins have a new tool in the hybrid toolkit. The July 6th announcement isn’t just a version update; it’s a statement that the wall between on-premises VMware and cloud VPCs is gone. Use it wisely.