Powering Microsoft’s Azure Cobalt 200 with Arm Neoverse CSS V3: The next generation of Arm-based compute for the AI era 

In the era of artificial intelligence (AI), a clear shift has occurred from commodity off-the-shelf systems to purpose-built infrastructure. Every workload, from traditional web services to scalable data analytics and large-scale model inference now plays a role in the AI-enabled intelligence pipeline. Modern data centers must be architected not as collections of discrete compute types, but as converged systems capable of handling diverse and demanding workloads efficiently. The industry is optimizing compute from the ground up to meet this need, prioritizing performance, scalability, and energy efficiency. 

Azure Cobalt 200: Accelerating the converged AI data center 

Today, Microsoft announced Cobalt 200, the next generation of Arm-based CPU processors powering its Azure cloud infrastructure and the first publicly announced silicon built on the Arm Neoverse Compute Subsystem V3 (CSS V3). Cobalt 200 exemplifies Microsoft’s vision of the converged AI data center, where general-purpose CPUs and specialized accelerators work side by side with bespoke networking, storage, and security offload to train, fine-tune, and deploy cloud and AI workloads at a massive scale. 

A partnership for the future of AI infrastructure

Arm and Microsoft share a vision for a more efficient, scalable, and sustainable cloud — one where every layer of the stack, from silicon to software, is optimized not just for token generation — but for complete end-to-end intelligence per watt. The launch of Cobalt 200 marks another milestone in this collaboration, demonstrating what’s possible when world-class computing architecture meets world-class cloud engineering. 

“The shift to purpose-built systems for AI and cloud-native workloads is transforming how infrastructure is designed,” said Dermot O’Driscoll, VP Products and Solutions, Cloud AI Business Unit, Arm. “Our collaboration with Microsoft on Cobalt 200 demonstrates how Arm Neoverse CSS V3 can be leveraged to accelerate this transformation and enable highly optimized silicon for a wide range of data-driven applications.” 

“Cobalt 200 is the result of our longstanding co-innovation efforts between Microsoft and Arm, demonstrating what is possible when optimizing silicon, server, and software together,” said Selim Bilgin, Corporate Vice President of Silicon Engineering, Microsoft. “We’re excited to be the first implementers of the Neoverse CSS V3, and taking advantage of Arm’s extensible architecture to implement unique capabilities like per-core DVFS, custom compression and cryptographic accelerators, as well as Azure Boost. Together, we are delivering compute infrastructure that improves performance and power efficiency for real cloud workloads.”  

Cobalt 200 follows our successful collaboration with Microsoft on its Cobalt 100 processor, which is delivering Neoverse-powered instances that are outpacing traditional competitors on performance and efficiency across various real-world workloads, including database, networking, and quantitative finance. 

Accelerating custom silicon innovation with Neoverse CSS 

Neoverse CSS V3 is a pre-integrated and silicon-proven platform that enables silicon designers — from hyperscalers to Arm Total Design partners including ASIC design houses, silicon startups or established merchant silicon providers — to design and deploy custom silicon faster than ever before. It is built around Arm’s most performant Neoverse CPU core to date and includes support for Armv9 architectural features such as Confidential Computing Architecture (CCA) and Scalable Vector Extension 2 (SVE2). 

This approach allows partners to focus on innovation where it matters most — optimizations in silicon and at the system level specific to their workload profile — while Arm continues to deliver the world’s leading platform for cloud-to-edge compute, backed by a cohesive cloud-to-edge AI-enabled software ecosystem. Microsoft did exactly this by modeling more than 350,000 configuration parameters based on extensive workload profiling to optimize performance for both customer needs and internal applications like Microsoft Teams. 

Building the future cloud AI data center 

As AI continues to permeate the data center conversation, compute must optimize for every workload — certainly AI training and inference, but also the critical web, storage and data serving and processing workloads that form the backbone of modern commerce and are critical in the AI era to translate tokens into intelligence. This macro-level convergence of general-purpose and AI-optimized compute has become the foundation of next-generation cloud infrastructure. With Cobalt 200 powered by Neoverse CSS V3, Arm and Microsoft are a fundamental part of this computing transformation in shaping more performant, efficient and flexible infrastructure for the future intelligence-optimized data center.