Amazon EVS lets you run VMware Cloud Foundation natively on AWS using dedicated EC2 bare metal instances. It enables lift-and-shift migration of VMware workloads with minimal changes, keeps familiar tools like vCenter, and integrates easily with AWS services. EVS supports scalability, high availability, and phased modernization, making it a practical option for VMware-heavy organizations moving to the cloud. Generally available since August 2025.
As these organizations explore cloud options, a key consideration is how to integrate their existing VMware environments with cloud services without undergoing extensive re-platforming. Amazon Elastic VMware Service (EVS) is designed to address this need by allowing users to run their VMware workloads natively on AWS.
Amazon Elastic VMware Service (Amazon EVS) is an AWS service that runs VMware Cloud Foundation directly inside your own Amazon VPC on dedicated EC2 bare-metal instances. It lets you move VMware workloads to AWS without re-platforming or refactoring applications, keeping vCenter, vSphere, vSAN and NSX exactly as your team already uses them.
AWS made Amazon EVS generally available on 5 August 2025, after introducing it in preview at re:Invent 2024. This guide covers what the service does, where it now runs, what changed after launch, and where it fits in a wider modernization plan.
What Is Amazon EVS?
Amazon EVS gives you a customer-managed software-defined data center on AWS infrastructure. You get full administrative access to vCenter Server and the ESXi hosts, the same as an on-premises deployment, but the hardware underneath is EC2 bare metal inside your own VPC.
The distinction that matters is ownership. EVS is not a managed VMware service where AWS operates the stack for you. You self-manage the environment, or bring in an AWS Partner to run it, and you keep control of the virtualization layer.
Amazon EVS General Availability: What Changed
The service reached general availability on 5 August 2025 in six AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Tokyo), Europe (Frankfurt) and Europe (Ireland).
That footprint has expanded considerably since. AWS added Mumbai, Sydney, Canada (Central) and Paris in November 2025, then US West (N. California), Hyderabad, Malaysia, Calgary, Milan, Mexico (Central) and São Paulo in December 2025.
At launch the service supported VCF 5.2.1 on i4i.metal instances. Several capabilities have arrived since:
- VCF 9.0 and 9.1 support, announced in July 2026 and available in every Region where EVS runs.
- Memory Tiering, a VCF 9.0 feature that uses local NVMe as a second memory tier alongside DRAM, raising cluster density. It runs on both i4i.metal and i7i.metal-24xl.
- NSX Federation, which connects multiple NSX environments under a single Global Manager for multi-site networking and disaster recovery.
- Multiple NSX Tier-0 Gateways, added in January 2026, for finer network segmentation and distributed traffic across edge clusters.
Early adopters named by AWS include Aeromexico, the Alcaldía de Cali, Effectual and Huron Consulting Group.
Why Organizations Use Amazon EVS
1. Migration Without Re-Platforming
The main draw is that workloads move without being rewritten. Because EVS runs the standard VCF stack, teams keep their existing VMware skills, tools and runbooks. IP addresses do not change, and staff do not need retraining on a new virtualization platform.
2. Access to Native AWS Services
Once workloads sit inside your VPC, they are adjacent to the rest of AWS. That opens up analytics, machine learning, managed databases such as Amazon RDS, and storage such as Amazon S3, without re-architecting the VMware estate first. It makes incremental modernization practical rather than all-or-nothing.
3. Elastic Capacity and Resilience
Capacity adjusts on demand rather than through hardware procurement cycles. Multiple Availability Zones can be built into high-availability and disaster-recovery designs, and NSX Federation extends that across sites.
4. Dedicated Hardware and Full Control
Workloads run on single-tenant EC2 bare-metal instances, so you get direct access to the underlying server resources and retain administrative control of the VCF environment inside your own VPC.
5. License Portability and Consumption Choice
VCF license portability means existing Broadcom entitlements can come with you. Consumption is flexible: on-demand, one-year and three-year commitments are available, which matters when modelling the cost of a multi-year migration.
Where Amazon EVS Fits in a Modernization Plan
EVS is most useful as a first move rather than a destination. It gets workloads out of a data centre quickly and onto AWS infrastructure, which buys time to decide what happens next, application by application.
From there, the usual path is selective: some workloads stay on VCF indefinitely, some get refactored onto containers with Amazon EKS, and some are replaced by managed services. That sequencing is the core of most application modernization programmes, and it avoids the risk concentration of a single large rewrite.
The cost question follows closely behind. Running VMware on dedicated bare metal is not automatically cheaper than on-premises, so right-sizing matters, and Memory Tiering changes the sizing maths considerably. Our work on Kubernetes cost optimization with OpenCost covers the same discipline applied to container workloads.
For teams weighing EVS against a fuller re-platform, cloud infrastructure evolution sets out how the two approaches compare over a longer horizon.
Considerations Before Adopting EVS
A few things are worth settling before committing:
- Region availability has expanded well beyond the launch six, but check your specific Region and data-residency requirements.
- VCF version determines which features you can use. Memory Tiering requires VCF 9.0 or later.
- Instance type affects both cost and capability, with i4i.metal and i7i.metal-24xl currently available.
- Operational ownership stays with you or a partner. EVS is not a fully managed service.
- Sizing needs redoing rather than copying from on-premises, particularly if Memory Tiering is in play.
Frequently Asked Questions
What is Amazon EVS?
Amazon EVS (Amazon Elastic VMware Service) is an AWS service that runs VMware Cloud Foundation on dedicated EC2 bare-metal instances inside your own Amazon VPC. It lets organizations migrate VMware workloads to AWS without re-platforming or refactoring, keeping vCenter, vSphere, vSAN and NSX in place. You retain full administrative control of the environment, either self-managed or run by an AWS Partner.
Is Amazon EVS generally available?
Yes. AWS announced general availability on 5 August 2025, following a preview introduced at re:Invent 2024. It launched in six Regions and has since expanded to many more across North America, Europe, Asia Pacific and South America.
Which VCF versions does Amazon EVS support?
Amazon EVS supported VCF 5.2.1 at general availability. AWS announced support for VCF 9.0 and 9.1 in July 2026, available in every Region where EVS runs. VCF 9.0 is also the version that introduced Memory Tiering.
How much does Amazon EVS cost?
AWS does not publish a single figure, because cost depends on instance type, host count, Region and commitment term. Consumption options include on-demand, one-year and three-year commitments, and VCF license portability lets you bring existing Broadcom entitlements rather than buying new ones. Memory Tiering can reduce host count for the same effective memory, which changes the calculation.
Is Amazon EVS a managed service?
No. Amazon EVS gives you a customer-managed VCF environment on AWS infrastructure. AWS provides and maintains the underlying bare-metal hardware and the deployment workflow, while you retain administrative control over vCenter, ESXi and the rest of the stack. Organizations that want it operated for them typically engage an AWS Partner.
What is the difference between Amazon EVS and VMware Cloud on AWS?
The key difference is where the environment lives and who controls it. Amazon EVS runs VCF inside your own Amazon VPC on EC2 bare metal, with you or a partner managing the stack. It is a native AWS service, purchased and operated through AWS, with VCF license portability from Broadcom.
