Microsoft has released early performance data showing that its Azure Cobalt 200 VMs deliver between 1.3x and 4.15x greater throughput than Cobalt 100 for electronic design automation tasks. This advancement boosts compute capacity for large-scale simulation, verification, and static timing analysis essential for designing new silicon.
- Up to 4.15x faster EDA job completion compared to previous-generation Cobalt 100.
- Improved per-core performance sustained under high concurrency with enhanced cache and memory bandwidth.
- Seamless migration of existing EDA software with preserved functionality and increased throughput.
Infrastructure signal
Azure Cobalt 200 VMs leverage the Arm Neoverse CSS V3 platform with physical cores optimized for high memory bandwidth and dedicated L2 cache, eliminating overhead from hyperthreading. These hardware design choices directly benefit cloud-scale EDA workloads by enabling predictable and consistent per-core performance even at heavy utilization, a critical factor for managing large regression suites and concurrent heavy compute jobs.
This next-gen custom silicon provides a spectrum of throughput improvements across simulation, verification, and timing analysis tools. Compared to Cobalt 100, Cobalt 200 achieves 1.30x to 4.15x speed-ups, underpinning faster job turnaround and resource efficiency. Deploying these VMs within Azure’s highly scalable cloud infrastructure also allows flexible scaling to meet peak demand without the underutilization typical of on-premises capacity planning.
Developer impact
Silicon design teams can run familiar EDA tools like Cadence Xcelium, Synopsys VCS, and Siemens Calibre without any code modifications on the new Cobalt 200 instances. The cloud environment maintains consistent performance at scale, supporting large verification farms and regression testing workloads that previously required cumbersome fixed infrastructure.
Early testing reveals per-core throughput increases exceeding 40% for RTL simulations and approximately 38% for analog and mixed-signal simulations. These performance gains shorten design and verification cycles, enabling faster iteration and reducing time-to-silicon. Developers benefit from enhanced VM concurrency and stable performance, supporting more efficient job scheduling and maximizing compute investment.
What teams should watch
Cloud architecture and chip design groups should monitor the evolving Azure VM offerings that target compute-intensive workloads, as Cobalt 200 sets new benchmarks for performance and cost efficiency in cloud-based silicon engineering. Teams currently relying on fixed x86 servers will find that migrating to Azure Arm-based instances unlocks better resource utilization and throughput without sacrificing tool compatibility.
Observability into VM load performance and memory bandwidth utilization will be key for optimizing job parallelism and throughput in large verification environments. Tracking workload-specific metrics across simulation, static timing, and physical verification will help teams tune deployment configurations to meet SLAs and control cloud spend effectively amid rising chip complexity.