HomeWinBuzzer NewsMicrosoft Unveils Azure Boost DPU to Power Cloud Efficiency

Microsoft Unveils Azure Boost DPU to Power Cloud Efficiency

Microsoft enters the competitive DPU market with its custom Azure Boost DPU, targeting energy-efficient cloud solutions and challenging rivals like NVIDIA and Intel.

-

Microsoft has introduced the Azure Boost DPU, a custom data processing chip designed to handle data-centric tasks on Azure’s infrastructure by consolidating networking, storage, and processing functions.

DPUs, Data Processing Units, offload and accelerate tasks that would traditionally burden a CPU, such as networking, storage management, security, and data movement. This allows the CPU to focus on core application processing while the DPU manages infrastructure-related workloads.

Microsoft´s Azure Boost DPU is engineered to handle data-centric workloads more efficiently than traditional systems, promising four times the performance and three times lower power consumption. With the custom chip for its Azure Boost system, Microsoft wants to meeting the growing demands of artificial intelligence and large-scale cloud operations.

Azure Boost is a infrastructure system Azure, designed to offload virtualization processes traditionally handled by the hypervisor and host operating system onto purpose-built hardware and software.

The Azure Boost DPU will be integrated into Microsoft’s data centers starting in 2025, alongside the newly announced Azure Integrated Hardware Security Module (HSM), which enhances cryptographic security. Together, these advancements signal a strategic push by Microsoft to optimize energy efficiency, performance, and security in hyperscale environments.
 

Rising Competition in the DPU Market

Microsoft’s entry into the DPU market pits it against established players like NVIDIA, AMD, and Intel, each offering specialized solutions tailored for modern data center needs.

  • NVIDIA BlueField: Known for its advanced capabilities, the BlueField-3 DPU handles up to 400 Gbps of network throughput and incorporates ARM cores with accelerators for encryption and packet processing. NVIDIA maintains strong partnerships with VMware and other cloud providers (nvidia.com).

  • AMD Pensando: After acquiring Pensando Systems in 2022, AMD introduced DPUs such as the Salina and Pollara models. These chips are optimized for AI-driven tasks, providing up to 400 Gbps throughput (amd.com).

  • Intel IPU: Intel’s Infrastructure Processing Unit offloads networking and storage tasks to reduce CPU overhead in cloud environments. The IPU focuses on boosting data center efficiency through task-specific optimizations (intel.com).

The Strategic Importance of the Azure Boost DPU

Microsoft’s acquisition of Fungible in late 2023 laid the groundwork for the Azure Boost DPU. Fungible’s expertise in data-centric silicon enabled Microsoft to design a chip that integrates core functions into a single package.

This aligns with its collaboration with Intel Foundry, which is slated to use its advanced 18A process node for Microsoft’s future chips starting in 2024.

Related:

Energy Efficiency and Sustainability Goals

As hyperscale data centers grapple with rising energy consumption, Microsoft has increaseing energy costs due to AI workloads. The Azure Boost DPU is engineered to alleviate these challenges by significantly lowering energy requirements.

In addition, Microsoft is retrofitting its data centers with liquid cooling technology, improving thermal efficiency and supporting high-performance components like NVIDIA GB200 GPUs.

The Azure Integrated HSM adds a layer of hardware-based security to Azure’s infrastructure. By securely storing cryptographic keys within its module, the HSM protects sensitive data without impacting latency.

Microsoft’s move mirrors a broader industry trend of hyperscalers investing in proprietary silicon to maintain control over supply chains and optimize infrastructure for AI workloads. Amazon’s Trainium and Inferentia chips, developed through Annapurna Labs, illustrate this push. Similarly, Alphabet and Meta have introduced custom chips, like Google’s Trillium TPUs and Meta’s MTIA, to handle their AI operations more efficiently.

With competition from NVIDIA, AMD, and Intel, Microsoft’s Azure Boost DPU must prove its value in terms of performance and compatibility. The company’s focus on integrating its custom silicon with broader Azure services could help it carve out a strong position in the DPU market.

SourceMicrosoft
Markus Kasanmascheff
Markus Kasanmascheff
Markus has been covering the tech industry for more than 15 years. He is holding a Master´s degree in International Economics and is the founder and managing editor of Winbuzzer.com.

Recent News

0 0 votes
Article Rating
Subscribe
Notify of
guest
0 Comments
Newest
Oldest Most Voted
Inline Feedbacks
View all comments
0
We would love to hear your opinion! Please comment below.x
()
x
Mastodon