Microsoft has two new chips designed for data centers. The Azure Boost DPU is Microsoft’s first data-specific chip, or data processing unit, and is designed to work with high efficiency and low power. Microsoft anticipates DPU-equipped Azure servers will be able to run storage workloads at four times the performance of those currently in place while reducing power consumption by three times. The Azure Integrated Hardware Security Module (HSM) chip allows cryptographic security keys and encryption keys to be contained in secure modules “without compromising performance or increasing latency.”

The new chips, unveiled this week at Microsoft Ignite, are scheduled for release in the coming months.

Following last year’s launch of the Azure Maia AI accelerators and Cobalt 100 CPUs, these new additions to Microsoft’s custom silicon portfolio mark “another major step in the company’s comprehensive strategy to rethink and optimize every layer of its stack — from silicon to software — to support advanced AI,” further optimizing Azure servers for large-scale AI workloads, reports VentureBeat.

The Azure Integrated HSM is equipped with a dedicated hardware security module that meets the federal government’s FIPS 140-3 Level 3 security standards.

“Unlike traditional HSM architectures that require network round-trips or key extraction, the [Azure Integrated HSM] chip performs encryption, decryption, signing, and verification operations entirely within its dedicated hardware boundary,” explains VentureBeat.

The other new chip, the Azure Boost DPU, is “designed for scale-out, composable workloads on Azure,” delivering “‘efficiency across storage, networking, acceleration, and more for its cloud infrastructure,’ Microsoft notes in a blog post. “Azure Boost DPU integrates high speed Ethernet and PCIe interfaces along with network and storage engines, data accelerators, and security features, into a fully programmable system on chip (SoC).”

TechCrunch notes that “the Azure Boost DPU likely has its origins in Fungible, a DPU fabricator that Microsoft acquired last December” for an estimated $190 million. The company was founded by former engineering staff from Apple and Juniper Networks. Fungible’s team joined Microsoft’s infrastructure engineering division following the acquisition.

The efficiency improvements come as “Goldman Sachs published research estimating that advanced AI workloads are poised to drive a 160 percent increase in data center power demand by 2030, with these facilities consuming 3-4 percent of global power by the end of the decade,” VentureBeat adds.

In addition to its new chips, Microsoft also announced improvements in data center cooling, “an advanced version of its heat exchanger unit — a liquid cooling ‘sidekick’ rack.” Although it did not share the specific gains from this new hardware, it said “it can be retrofitted into Azure data centers to manage heat emissions from large-scale AI systems using AI accelerators and power-hungry GPUs such as those from Nvidia,” according to VB.