Data processing unit

[2] They can be used in place of traditional NICs to relieve the main CPU of complex networking responsibilities and other "infrastructural" duties; although their features vary, they may be used to perform encryption/decryption, serve as a firewall, handle TCP/IP, process HTTP requests, or even function as a hypervisor or storage controller.

By offloading tasks such as packet processing, encryption, and traffic management, DPUs help reduce latency and improve energy efficiency, enabling these AI factories to maintain the high throughput and scalability needed for advanced machine learning operations.

This approach offloads critical but lower-level system duties—such as security, load balancing, and data routing—from the central processor, thus freeing CPUs and GPUs to focus on application logic and AI-specific computations.

Similar to AWS’s Nitro System and NVIDIA’s BlueField DPUs, Microsoft’s DPU focuses on enhancing cloud efficiency while addressing rising energy and security demands.

[8] The introduction of DPUs like Azure Boost reflects a broader shift in the cloud computing industry toward offloading specific functions from general-purpose processors to specialized hardware.