The Proxmox Movement That VMware Probably Didn’t Expect

For years, Proxmox VE has played a very specific role in the virtualization market. For many organizations, it was primarily an open source and much more affordable alternative to VMware. However, this narrative is starting to fall short.

The arrival of official support for ARM64 in Proxmox VE means much more than simply adding compatibility with a new processor architecture. It is a declaration of intent regarding how data centers will operate in the coming decade, where x86 servers, ARM processors, GPU accelerators, Kubernetes, and AI workloads will coexist within a unified management platform.

The key points of Proxmox’s strategy in 30 seconds

• Proxmox VE now includes official support for ARM64, starting with NVIDIA Grace and NVIDIA Vera platforms.
• This move coincides with the integration of NVIDIA Mission Control for managing AI infrastructures.
• While VMware maintains ESXi-Arm as a Tech Preview, Proxmox already offers official support for production on the first certified platforms.
• The shift is happening amidst a broader migration from VMware, as companies distribute their workloads across multiple options without a single replacement in sight.

Additionally, the announcement comes at a particularly interesting time for the market. Following changes introduced by Broadcom in VMware’s licensing model, thousands of organizations are reevaluating their virtualization strategies. But, contrary to expectations, many are not migrating entirely to one platform.

The market no longer seeks a single VMware replacement

A CloudBolt study, conducted with over 300 IT managers, highlights the degree of fragmentation in the market.

According to the report, 86% of organizations are reducing their reliance on VMware, while only 4% have completed full migration. Most expect the process to take between 18 and 24 months.

What’s more striking is not that data, but where workloads are headed.

This demonstrates that there is no single successor to VMware. Each company is crafting its own strategy by combining different technologies based on its needs.

In this environment, Proxmox continues to grow. The company reports having over 1.5 million installed hosts, more than 200,000 users, and presence in more than 140 countries.

But these numbers may no longer be the most important news.

ARM64 support shifts the conversation

Until now, Proxmox’s main argument was relatively simple: providing enterprise virtualization based on open source software, with high availability, distributed storage via Ceph, integrated backups, and a highly competitive licensing model.

With ARM64, that focus is starting to change.

The first officially supported platforms are NVIDIA Grace and NVIDIA Vera, two architectures designed specifically for AI, high-performance computing (HPC), and large-scale cloud infrastructures.

This doesn’t seem to be an isolated move.

Just a few weeks ago, Proxmox announced integration with NVIDIA Mission Control, the US manufacturer’s platform for managing so-called AI Factories, large infrastructures composed of thousands of GPUs.

Taken together, these strategies begin to paint a different picture of the future.

It’s no longer just about virtualizing servers

Traditional virtualization will remain a core part of any data center, but the infrastructure is changing rapidly.

New deployments are increasingly combining:

• Traditional x86 servers;
• High-performance ARM processors;
• GPU accelerators for AI;
• Kubernetes;
• Containers;
• Distributed storage;
• Software-defined networking.

The goal is no longer merely running virtual machines.

The real challenge is managing much more heterogeneous infrastructures cohesively.

And this is exactly where Proxmox seems to want to position itself.

VMware still holds significant advantages

This doesn’t mean Proxmox has surpassed VMware.

Broadcom’s platform continues to be the reference for many large organizations thanks to its vastly superior ecosystem in numerous aspects.

VMware maintains clear advantages in large enterprise deployments, especially where there are significant investments in enterprise storage, automation, or software-defined networking.

However, Proxmox’s move points toward a different challenge.

The true change is at the heart of future data centers

For a long time, Proxmox seemed to grow because it offered a reasonable alternative for those seeking to leave VMware.

Now it is beginning to develop capabilities for something much more ambitious.

If it manages to natively oversee infrastructures that include x86 processors, ARM architectures, NVIDIA GPUs, Kubernetes, and upcoming AI-accelerated platforms, it will stop competing just in the virtualization market.

It will start competing in the comprehensive AI infrastructure management space.

And that difference is far more significant than it might seem.

Official support for ARM64 doesn’t only mean compatibility with a new family of processors. It’s the first visible step in a strategy aimed at data centers that will emerge over the coming years.

Soon, we might stop talking about Proxmox as “the alternative to VMware.” Instead, we may see it as a platform designed to manage the hybrid infrastructure that will underpin the next generation of AI applications.

Frequently Asked Questions

What does ARM64 support in Proxmox VE mean?

It enables official deployment of Proxmox VE on ARM64-based servers. The first supported platforms are NVIDIA Grace and NVIDIA Vera.

Does Proxmox replace VMware?

Not necessarily. VMware remains the reference in many enterprise environments due to its ecosystem, certifications, and integration with corporate infrastructures. However, Proxmox is now expanding its strategy into new areas related to artificial intelligence.

Where are companies migrating from VMware headed?

According to CloudBolt, most companies are combining multiple options. Public clouds remain the most common destination, followed by Hyper-V, OpenShift Virtualization, Proxmox VE, and XCP-ng.

Why is ARM important for data centers?

Architectures such as NVIDIA Grace are designed to deliver high performance per watt for workloads in AI and HPC, two segments that attract significant new infrastructure investments.

Sources:

• Proxmox VE.
• CloudBolt Industry Survey 2024.