ArcGIS Pro Virtualization – Hardware and VM Profiles (2026 Edition)

Overview

ArcGIS Pro is a powerful desktop GIS application built for real work — editing data, working with 3D scenes, running analysis, and visualizing results. When you move Pro into a virtualized environment, the goal shouldn’t just be “can it run,” but “does it still feel like a workstation.”

That’s where many designs go wrong.

A lot of virtual desktop environments are built for email, spreadsheets, and web applications. ArcGIS Pro is none of those. It behaves much more like CAD, 3D modeling, or engineering software and places very different demands on CPUs, GPUs, memory, and storage.

This guide explains what actually matters when building infrastructure for ArcGIS Pro users and how to translate those requirements into hardware and VM profiles that perform well in production environments.

Who This Guide Is For

This guide is intended for GIS administrators, IT architects, and virtualization engineers responsible for designing or supporting ArcGIS Pro virtualization environments.

It focuses on practical infrastructure guidance for deploying ArcGIS Pro on GPU-enabled virtual desktop platforms such as VMware, Citrix, and other modern virtualization solutions.

Rather than covering every possible hardware configuration, the goal is to highlight the architectural principles that consistently lead to responsive and reliable ArcGIS Pro user experiences.

Typical ArcGIS Pro Virtualization Architecture

A typical ArcGIS Pro VDI environment places GPU-accelerated virtual machines in the data center while users access them remotely from their desktops or laptops.

In this model, ArcGIS Pro runs inside a virtual machine hosted on GPU-enabled servers managed by a virtualization platform such as VMware, Citrix, or Hyper-V. Users connect to these desktops through remote display technologies provided by platforms like VMware Horizon or Citrix Virtual Apps and Desktops.

From the user’s perspective, ArcGIS Pro behaves like a local workstation application, while the compute resources are centralized and shared in the data center.

Why CPU Choice Is So Important for ArcGIS Pro

One of the most common mistakes in GIS VDI designs is treating CPU selection as a simple “more cores is better” problem.

That approach works well for batch jobs, databases, and analytics. It does not work well for ArcGIS Pro.

ArcGIS Pro spends much of its time drawing maps, rendering scenes, updating symbology, and responding to user input. These operations depend heavily on single-core performance, not just total core count.

With Intel’s newer Xeon 6 processors, the difference between CPU models can be significant. Some processors offer 24 cores running around 3.0 GHz, while others provide 80 or more cores running well below 2.5 GHz. Those large CPUs are excellent for AI and HPC workloads, but they can feel sluggish for interactive desktop applications.

In practice we consistently see this pattern:

A 3.0 GHz 24–32 core CPU typically provides a noticeably better ArcGIS Pro experience than a 2.0 GHz 64-core CPU, even if the larger processor costs more.

Higher clock speeds translate directly into faster application launches, smoother map redraws, more responsive editing, and better 3D navigation.

CPU Classes That Work Well for ArcGIS Pro VDI

If you are evaluating Intel Xeon 6 Performance processors, the sweet spot for ArcGIS Pro virtualization tends to be:

• Xeon 6 6527P — 3.0 GHz, 24 cores
• Xeon 6 6737P — 2.9 GHz, 32 cores
• Xeon 6 6745P — 3.1 GHz, 32 cores with large cache

Once you move into processors with 48, 64, or more cores that run below about 2.5 GHz, you are typically moving out of the interactive workstation range and into CPUs designed primarily for AI and HPC workloads.

While those processors look impressive on spec sheets, they rarely deliver the best user experience for ArcGIS Pro.

What a Good ArcGIS Pro VDI Server Actually Looks Like

Once the right class of CPU is selected, the rest of the server design becomes more straightforward.

A modern ArcGIS Pro virtualization host should essentially resemble a rack-mounted workstation.

A typical 2026 configuration might include a Dell PowerEdge R770 (or a comparable platform) configured with:

• Two Xeon 6 Performance CPUs in the ~3.0 GHz range
• One to two terabytes of DDR5 memory
• Two to four NVIDIA data-center GPUs
• High-performance NVMe or tier-1 networked storage
• 25 Gb or 10 GB networking

Each of these servers replaces dozens of physical GIS workstations, so the platform needs to be designed accordingly.

Why GPUs Matter for ArcGIS Pro

ArcGIS Pro uses the GPU for far more than just 3D scenes. Raster display, symbology, layer drawing, and even basic navigation all benefit from graphics acceleration.

In a virtual environment, that acceleration comes from NVIDIA data center GPUs using NVIDIA vGPU software. From ArcGIS Pro’s perspective, the GPU appears as a local workstation graphics device — it is simply shared among users behind the scenes.

ArcGIS Pro also includes functionality that leverages NVIDIA CUDA for certain processing tasks. CUDA allows the GPU to accelerate computational workloads in addition to graphics rendering. Depending on the workflows being used, this can improve the performance of raster analysis and other GPU-enabled geoprocessing tools.

Because CUDA is NVIDIA-specific, ArcGIS Pro virtualization environments are typically deployed using NVIDIA GPUs that support both graphics and CUDA compute workloads.

The goal is not simply to add GPUs to the environment, but to allocate GPU resources appropriately so each user receives enough graphics and compute capability without wasting capacity.

Light, Medium, and Heavy Users

Most organizations naturally fall into three user categories.

Light users primarily view data, run simple queries, and present maps.

Medium users perform editing tasks, work in both 2D and 3D, and carry out typical day-to-day GIS workflows.

Heavy users perform advanced visualization, complex symbology, spatial analysis, and GPU-accelerated geoprocessing.

A good starting point for VM sizing is:

• Light users: 4 vCPU, 8 GB RAM, 2Q GPU
• Medium users: 8 vCPU, 16 GB RAM, 2Q–4Q GPU
• Heavy users: 12–16 vCPU, 32 GB RAM, 4Q GPU or larger

CPU and GPU resources must be sized together. Assigning a large GPU profile without sufficient CPU resources simply creates a different performance bottleneck.

Example GPU Sizing with NVIDIA vGPU Profiles

Modern NVIDIA GPUs allow their memory to be divided into multiple vGPU profiles, enabling several virtual machines to share the same physical GPU.

For most organizations, 2Q or 4Q profiles provide the best balance between performance and user density.

How Many Users Fit on a Server?

The number of ArcGIS Pro users that fit on a single server depends heavily on the virtual machine configuration, GPU profile sizes, and the mix of user workloads.

On systems similar to a PowerEdge R770 configured with high-clock CPUs and multiple GPUs, organizations commonly support 30+ ArcGIS Pro users per host, with the exact number determined by VM sizing and workload characteristics.

Environments focused on lighter GIS workflows may support higher densities, while environments focused on advanced 3D visualization or complex analysis may require fewer users per host.

The most reliable way to determine capacity is to begin with conservative VM sizing and validate performance with representative workloads before increasing density.

Alternative High-Density GPU Platforms

While balanced platforms like the PowerEdge R770 work well for many environments, some organizations may prefer higher GPU density within a single server.

Systems such as the Dell PowerEdge XE7740 and XE7745 can support up to eight GPUs per host and may be used to build larger GPU virtualization clusters.

These platforms are designed primarily for AI and high-performance computing workloads. At the time of writing, Esri performance engineering has not formally validated ArcGIS Pro VDI deployments on these systems, but properly configured environments should be capable of supporting similar workloads.

Comparing Server Platform Options

CPU Options for XE7740 (Intel Platform)

Recommended processors include:

• Xeon 6 6527P – 3.0 GHz, 24 cores
• Xeon 6 6737P – 2.9 GHz, 32 cores
• Xeon 6 6745P – 3.1 GHz, 32 cores

These CPUs provide strong single-thread performance while maintaining reasonable VM density.

CPU Options for XE7745 (AMD Platform)

Examples include:

• AMD EPYC 9375F – 32 cores, high frequency
• AMD EPYC 9475F – 48 cores
• AMD EPYC 9354 – balanced performance

These CPUs generally provide a better experience for interactive workloads than very large EPYC processors designed primarily for AI clusters.

Real-World ArcGIS Pro VDI Mistakes We See Most Often

Across the GIS and VDI community, several design issues appear frequently.

Using High-Core, Low-Clock CPUs

Large CPUs designed for AI workloads often perform poorly for interactive desktop applications.

Fewer, faster cores typically provide a better ArcGIS Pro experience.

Oversubscribing GPU Resources

Placing too many users on a single GPU can reduce frame rates and cause choppy 3D navigation.

Start with conservative GPU density and increase gradually.

Mismatched CPU and GPU Sizing

Large GPU profiles paired with small CPU allocations often create bottlenecks where the GPU waits for CPU processing.

Treating GIS Desktops Like Office VDI

ArcGIS Pro behaves more like CAD or engineering software than a traditional office application.

Designing GIS desktops like office VDI environments frequently leads to performance problems.

Poorly Optimized 3D Content

Extremely detailed 3D meshes or poorly optimized datasets can overwhelm even powerful infrastructure. Using scene layers and level-of-detail models can significantly improve performance.

A Simple Rule of Thumb for GPU Profiles

Oversized GPU allocations reduce user density without improving performance for most workflows.

Quick Design Guidelines

For organizations planning ArcGIS Pro virtualization deployments, the following principles consistently lead to better performance and user experience:

• Prioritize CPU clock speed over maximum core counts
• Balance CPU, GPU, and memory resources together
• Start with moderate GPU profiles such as 2Q–4Q
• Validate user density using real workloads
• Treat ArcGIS Pro like a workstation application

Final Thoughts

When ArcGIS Pro VDI is designed correctly, it can feel just as responsive as a high-end physical workstation while providing the operational advantages of centralized infrastructure.

The key is understanding how GIS workloads behave: fast CPU cores, balanced GPU resources, properly sized virtual machines, and optimized data.

When those elements are aligned, modern GPU-accelerated virtualization platforms can deliver excellent ArcGIS Pro performance at scale.