In recent years, the adoption of technologies based on Artificial Intelligence, machine learning, and desktop virtualization has accelerated dramatically. However, behind the innovation visible to end users lies a fundamental requirement: high-performance IT infrastructures capable of efficiently handling complex workloads. While the cloud often appears to be the go-to solution, it is not always the only — or the most suitable — option for every need.

In certain scenarios, organizations are increasingly demanding local solutions that deliver high performance, low latency, and strict data control. This need is often driven by concerns related to security, regulatory compliance, and specific architectural constraints that require workloads to run directly in on-premises environments.

This is where Azure Local comes into play — a Microsoft offering that redefines the concept of hybrid infrastructure. Combining the power of Azure cloud with the flexibility of local deployment, Azure Local enables full utilization of GPUs directly within your datacenter, empowering AI and Virtual Desktop Infrastructure (VDI) scenarios with high performance and full operational control.

Why Use On-Premises GPUs?

Let’s start with a key point: enterprise-grade GPUs — those built for datacenters and heavy workloads—have become essential for handling complex tasks such as:

• Training and inference of AI models
• Real-time video and visual processing
• Virtualization of graphic-intensive desktops and compute-heavy environments

But what happens when these workloads need to run in environments where:

• Internet connectivity is absent, unstable, or unsuitable for continuous traffic
• The data being processed is too sensitive or regulated to be moved to the public cloud
• Applications require immediate response times without delays from round trips to Azure

In these cases, having on-premises GPUs fully integrated into your local infrastructure is not just a strategic choice — it’s often a necessity. This is where Azure Local steps in, enabling organizations to harness the power of enterprise GPUs right in their datacenter, with the simplicity and scalability of the Azure experience.

What is Azure Local?

Azure Local is essentially the extension of Microsoft’s cloud services directly into your own datacenter. It delivers a selection of Azure services, the same APIs, and the same management model—but with the option to run everything locally, wherever it’s needed: on-premises or at the edge.

With Azure Local, you can deploy applications, virtual desktops, and AI models within your own infrastructure while retaining complete data control — benefiting from the flexibility, scalability, and operational consistency of the cloud. No need to move sensitive data. No compromise on the Azure experience. Just the resources you need, right where you need them.

Azure Local + GPUs: A Powerful Combination

One of Azure Local’s most compelling features is its native GPU support, allowing you to tackle AI workloads and VDI environments with high performance and operational efficiency. You can choose between two usage modes:

1. DDA – Discrete Device Assignment: The GPU is exclusively assigned to a single virtual machine. This is the most powerful mode, ideal for scenarios requiring maximum compute power, such as AI model training, deep learning, or advanced rendering.
2. GPU-P – GPU Partitioning: In this mode, the GPU is divided into multiple virtual partitions, each assignable to a VM. Perfect for maximizing efficiency and supporting multi-user environments like VDI.

Both modes are fully compatible with NVIDIA drivers and support major compute and graphics libraries, including CUDA, OpenGL, and DirectX.

Which GPUs Are Supported?

Supported models currently include:

• NVIDIA A2 and A16 – Supported in both DDA and GPU-P modes
• NVIDIA A10, A40, L4, L40, L40S – Supported in GPU-P mode

All are centrally manageable through Azure Arc, ensuring full control and visibility — even in the most distributed environments.

What About Virtual Desktops?

This is where things get even more interesting.

With Azure Virtual Desktop on Azure Local, you can deliver modern, high-performance, and secure desktop experiences directly within your on-premises environments. This means bringing the benefits of cloud-native VDI to where it truly matters, with session hosts physically close to end users.

The result? A significantly improved user experience, thanks to:

• Ultra-low latency, ideal for on-site users or limited connectivity environments
• Optimized performance for graphics applications and compute-intensive workloads
• Data that remains on-premises, ensuring security and compliance
• Full compatibility with Windows 11 and 10 in multi-session mode
• Native integration with traditional Active Directory and Microsoft Entra ID

All orchestrated via the Azure portal, with the same provisioning, monitoring, and management tools—simplifying administration and ensuring operational consistency between cloud and datacenter.

AI Where It Truly Matters

When it comes to Artificial Intelligence, Azure Local is a game changer. You can now train, deploy, and manage AI models directly on-premises or at the edge, without relying on the cloud for every step of the process.

How? With two key technologies:

• Edge RAG (Retrieval-Augmented Generation): Enhances generative models by integrating your local data—without ever moving it out of your environment. An ideal solution for highly secure and confidential use cases such as healthcare, government, or regulated industries.
• Azure Machine Learning with Azure Arc: A unified platform for managing the entire lifecycle of AI models — from training to deployment — whether in the cloud or on-premises, using the same tools, APIs, and capabilities.

The result? A hybrid, secure, scalable, and fully localized AI ecosystem, designed to bring intelligence right where it’s needed: close to your data, your users, and your business-critical processes.

But Is It Complicated to Set Up?

Absolutely not. One of Microsoft’s main goals has been to simplify the configuration and management experience of Azure Local — even in GPU scenarios.

To get started, you simply need:

• GPU-compatible physical hosts (over 100 validated models available)
• VMs configured according to recommended technical requirements for DDA or GPU-P
• Connection to Azure Arc for centralized, consistent, and secure management

Once the environment is up and running, you can operate just as you would in the Azure cloud — but with your own data, your own network, and full infrastructure control. No added complexity—just greater operational flexibility.

Conclusion

In a constantly evolving tech landscape, where performance, security, and compliance demands are increasingly strict, Azure Local stands out as a true game changer. The ability to bring enterprise GPUs directly into local datacenters—while preserving the experience, scalability, and consistency of Azure cloud — empowers organizations to effectively tackle AI, VDI, and high-performance workloads.

Whether it’s about achieving ultra-low latency, protecting sensitive data, or operating in limited-connectivity environments, Azure Local offers a modern and tangible solution, with a flexible, manageable, and most importantly, accessible hybrid approach.

This is not simply about “bringing the cloud on-premises.” It’s about redefining how IT infrastructure supports core business processes, enabling advanced scenarios without compromise in control, performance, or security.

Ultimately, Azure Local is an excellent choice for those looking to bring innovation exactly where it’s needed most: close to the data, the users, and the everyday operational needs.