Microsoft Unveils New Infrastructure Chips for AI and Security

By Max A. Cherney

SAN FRANCISCO (Reuters) – Microsoft (NASDAQ:MSFT) has announced the development of two new infrastructure chips for its data centers aimed at enhancing artificial intelligence operations and boosting data security during its Ignite conference on Tuesday.

Microsoft has invested heavily in creating its own silicon for general-purpose applications and AI. Similar to competitors like Amazon (NASDAQ:AMZN) and Google (NASDAQ:GOOGL), Microsoft engineers believe that custom-designed chips offer significant performance and cost advantages.

Creating these bespoke chips is intended to decrease Microsoft's dependence on processors from Intel (NASDAQ:INTC) and Nvidia (NASDAQ:NVDA).

The newly developed chips are meant for deep installation within Microsoft’s data center infrastructure, with one focused on enhancing security and the other on data processing.

The company’s push for tailored data center processors stems from its goal to "optimize every layer of infrastructure" and ensure that Microsoft’s data centers can process information at the rapid pace demanded by AI applications, stated Rani Borkar, corporate vice president of Azure Hardware Systems and Infrastructure.

The new security chip, named the Azure Integrated HSM, will be installed in every new data center server starting next year. This chip is designed to securely manage critical encryption and sensitive security data within a dedicated module.

On the data processing side, the newly introduced Data Processing Unit (DPU) integrates multiple server components into a single chip focused on cloud storage, capable of performing tasks at three times lower power consumption and four times greater performance than existing hardware.

Additionally, Microsoft has rolled out a new liquid cooling system for data center servers to help maintain lower temperatures for neighboring components, supporting large-scale AI systems.