Azure Boost: leading cloud performance innovation

Azure’s proprietary system Azure Boost announced has driven new industry records in storage performance in 2025.  This system enables effective and efficient offloading of virtualization processes onto accelerated hardware enabling Azure Compute offerings to achieve an industry leading performances of 800K IOPS and 16 GB/s throughput for remote storage, and 6.6M IOPS and 36 GB/s throughput for local storage, made possible in conjunction with our Azure Boost SSD technology. Azure Boost capabilities are available to customers in both generally available and upcoming Azure VM SKUs releases. 

Scott Guthrie, executive vice president of Microsoft Cloud + AI, showcases the latest Azure Boost innovations during his Unpacking the tech session at Build 2025.

Azure Boost also takes network optimization to new heights, as showcased in the preview of our new type of VMs: the network-optimized Dnsv6 and Ensv6 series. These VMs leverage Azure’s proprietary hardware acceleration technology to enhance network throughput and reduce latency. The result? An architecture that can handle even the most demanding network scenarios, from video streaming platforms to edge computing applications, with up to 200 Gbps networking bandwidth.

Azure Boost also makes the general availability of the Fxv2 VM series possible, which marks a transformative step for compute-intensive workloads. Designed for CPU performance-demanding applications, these VMs offer significant improvements in CPU performance, memory capacity, and storage performance capabilities, making them ideal for compute-intensive workloads such as databases, data analytics, and electronic design automation (EDA).

FXv2 series VMs provide up to 50% better CPU performance compared to the previous generation, with sizes supporting up to 96 vCPUs and 1,832 GiB of memory. They also feature NVMe support for faster remote storage performance up to 400K IOPS and 11.25 GB/s throughput.

Mark Russinovich, Azure CTO, has demonstrated how Azure Boost can support guest RDMA for NVIDIA GPUDirect. This technology enables GPUs across the network to read and write directly to each other’s memory, bypassing the CPU and system memory (RAM). The result is increased bandwidth (up to 11x), lower latency, and more efficient inter-node GPU communication: Inside Azure innovations with Mark Russinovich.

Azure Boost is designed to cater to organizations looking to push the boundaries of what’s possible in storage and networking.  Whether it’s enabling ultra-low latency for distributed systems or ensuring consistent performance for data-heavy applications, these VMs deliver. The improved performance is particularly appealing to enterprises deploying hybrid and multi-cloud setups, where network speed plays a critical role in overall efficiency.

Why Azure Boost Matters

• Industry leading storage performance


• Enhanced throughput for network-intensive workloads


• Extremely low <1s maintenance time


• Airtight physical and memory security with 100% isolation between Azure Boost and customer workloads

By integrating Azure Boost, Microsoft not only meets the evolving performance demands of modern enterprises but also sets a new standard for cloud-based storage and networking solutions.

Azure Compute Fleet and VMSS Instance Mix: Making scaling easy

The release of Azure’s cutting edge VM technology also comes the innovative solutions to deploy and manage those VMs at scale.  The general availability of Azure Compute Fleet transforms how compute capacity is obtained and managed by enabling businesses to tap into a wider pool of compute resources, making it easier to scale on demand across a variety of SKUs, regions, and pricing models.   This is particularly useful for organizations running workloads with fluctuating demands, such as e-commerce platforms, financial trading systems, and event-driven applications.

Azure Compute Fleet ensures that resources are deployed intelligently to the customer’s workload demands, reducing the need for manual search of cost and availability – optimal VM sizes. By automating compute provisioning, businesses can focus on innovation rather than infrastructure management.

Azure Compute Fleet Benefits

• Extremely flexible scaling to accommodate workload fluctuations


• Cost-effective size allocation


• Granular control Spot VM allocation and replacement strategies

The introduction of Compute Fleet underscores Azure’s commitment to providing flexible, scalable solutions that prioritize both performance and cost-effectiveness.

Similarly to Azure Compute Fleet, the new Instance Mix feature for Virtual Machine Scale Sets with Flexible Orchestration Mode, allows customers to specify multiple VM sizes within a single management group, significantly improving capacity attainment, reducing deployment complexity, and enabling optimization for capacity availability.

This flexibility means customers can optimize your deployments with a capacity-informed allocation strategy that meets your workload’s needs.  Scale set deployments using instance mix will dynamically determine the best VM sizes to deploy when scaling based on underlying available capacity.

A unified vision for the future

The innovations unveiled at Build highlight Azure’s relentless pursuit of excellence in performance and scalability. Whether it’s the storage breakthroughs enabled by Azure Boost or the dynamic capabilities of Compute Fleet, these developments signal a future where businesses can operate without compromise.

As organizations continue to navigate an increasingly complex digital landscape, Azure’s latest offerings provide the tools they need to stay ahead of the curve. By prioritizing performance, scalability, and cost efficiency, Microsoft reaffirms its role as a leader in cloud innovation.

Stay tuned for even more updates as Azure continues to push the boundaries of possibility, empowering businesses and developers to achieve their boldest ambitions.