Microsoft Azure Cobalt 200 VMs reach general availability, bringing Arm-powered performance to cloud and agentic AI workloads
News highlights
- Microsoft Azure Cobalt 200 VMs are now generally available, built on Arm Neoverse CSS V3 to simplify silicon development and enable differentiation
- Cobalt 200 VMs deliver up to 50% better CPU performance than Cobalt 100 VMs, supporting demanding cloud and agentic AI workloads
- Arm-based server market share at leading hyperscalers is expected to exceed 60% by the end of 2026, reflecting growing adoption of Arm-powered cloud infrastructure
AI is fundamentally changing cloud infrastructure requirements. As agentic AI moves into production, cloud providers need greater performance, efficiency and flexibility to scale. Custom silicon allows hyperscalers to design infrastructure around the evolving needs of their customers.
At Microsoft Ignite 2026, Microsoft announced the general availability of Azure Cobalt 200 virtual machines (VMs), bringing its latest Arm-based custom CPU to customers running demanding cloud and agentic AI workloads. Built on Arm Neoverse Compute Subsystems V3 (CSS V3), Cobalt 200 marks the next step in our collaboration to advance Azure’s compute infrastructure.
Arm-based server market share at leading hyperscalers more than tripled from 15% in 2024 to over 50% in 2025, with Arm expecting that share to exceed 60% by the end of 2026. Microsoft’s plans for Cobalt 200 to represent a significant portion of new Azure Compute capacity in 2027 and beyond reinforce that momentum and Arm’s growing role in cloud infrastructure.
Delivering greater value for Azure customers tackling agentic AI workloads
Agentic AI is expanding the role of the cloud CPU. Agents execute code, retrieve data, invoke tools and coordinate tasks across applications, relying on CPUs to support these operations alongside the databases, caches and services that power enterprise workloads. Scaling these systems requires high-performance CPUs that can deliver responsive execution and efficiency at scale.
Cobalt 200 demonstrates the impact of purpose-built Arm-based infrastructure. Microsoft reports up to 50% better CPU performance for Cobalt 200 VMs compared with Cobalt 100, while Arm’s workload testing shows similar improvements across cloud applications. These gains strengthen the compute foundation supporting enterprise services and emerging agentic AI workloads.
Customers including Amadeus, Databricks, InMobi, OpenAI, SAP, and Snowflake are already seeing the benefits of Cobalt 200 across real-world workloads, from faster application response times to improved infrastructure utilization.
Accelerating custom silicon differentiation with Arm Neoverse CSS
Custom silicon gives hyperscalers greater control over performance, efficiency and their infrastructure roadmaps. But differentiation shouldn’t require rebuilding every element of a modern cloud CPU from scratch.
Arm Neoverse CSS V3 provides a proven, pre-integrated compute framework that helps partners reduce development complexity and accelerate time-to-market while retaining the flexibility to customize their silicon.
Microsoft utilizes that framework with capabilities tailored to Azure, including custom compression, cryptographic accelerators and per-core power management. Combined with Microsoft’s systems and software engineering, these capabilities allow Azure to optimize how workloads use compute, move data and consume power.
“With Cobalt 200, we’re building on our collaboration with Arm to deliver the performance and efficiency our customers need as cloud and AI workloads evolve,” said Reynold D’Sa, CVP Silicon, Cloud Hardware, and Infrastructure Engineering, Microsoft. “Arm Neoverse CSS V3 gives us the framework to innovate and differentiate, while allowing us to focus on the silicon and systems capabilities that make Azure unique.”
Enabling the Cobalt 200 ecosystem at scale
Cobalt 200 builds on an established Arm software ecosystem. Our collaboration with Microsoft extends across software optimization, migration support and developer enablement, helping customers adopt Arm-based compute for existing applications and new agentic services.
Together, Arm’s flexible compute framework, Microsoft’s silicon and systems innovation, and a mature software ecosystem are helping accelerate the adoption of Arm-based infrastructure across the cloud.
Any re-use permitted for informational and non-commercial or personal use only.






