AMD scored another victory as Microsoft moved its updated Azure Data Explorer platform to EPYC 2-based virtual machines (VMs).

Microsoft's AMD-based Dav4 VM family is now the default SKU for Azure Data Explorer clusters, while the Easv4 and Lsv2 instances — also AMD based — are aimed at customers that need high SSD performance or are working with storage-heavy workloads, respectively.

"These VM families represent unprecedented improvements in return on investment in cost and performance," said Uri Barash, product manager for Azure Data Explorer, in a blog post.

According to Microsoft, AMD's product will bolster CPU performance by more than 30%, and up to 50% in some workloads. Combined with a new Kusto analytics engine, the cloud provider claims a 30-fold improvement in operational efficiency.

Introduced in 2018, Azure Data Explorer is a big data analytics platform built around Microsoft's Kusto Engine, first introduced internally in 2015. Not long after launching the offering, the cloud provider discovered two common scenarios for the service: interactive telemetry data lakes and analytics platforms built on the Data Explorer backend.

With these scenarios in mind, Microsoft rebuilt the engine in a bid to improve performance. Kusto EngineV3 included a complete "redesign of the data storage format, native code generation for portions of the query plan, and automatic selection of strategies based on data shard statistics," Barash explained.

Alongside the new engine, which is now available in public preview, Azure is also adding streaming telemetry and key query language (KQL) support to their service.

Since the launch of its EPYC server chips in 2017, AMD has steadily stolen marketshare long held by Intel. Building on that success last August, AMD launched its second-generation EPYC chips based on the ZEN 2 architecture upping the core count to 64, more than twice that offered by Intel.

Besides Azure, which began offering EPYC 2 this spring, AMD now has more than 60 industry partners actively deploying its server offering, including all major cloud providers — Google, Amazon Web Services (AWS), IBM, and Oracle. Additionally, the company has seen a steady stream of adoption from strategic parters like VMware, Dell Technologies, and Hewlett Packard Enterprise (HPE).

Even Nvidia — AMD's rival in graphics — is using two 64-core EPYC 2 CPUs, alongside the eight massive A100 GPUs in its DGX A100 supercomputing rack, announced in April. However, AMD still faces stiff competition from their long-time rival, Intel, with its third-generation Xeon Scalable chips.

Intel earlier this week announced the addition of secure enclaves across its entire Xeon Scalable line up in a bid to combat AMD's secure encrypted virtualization-encrypted state (SEV-ES), which VMware and Google are now using for confidential computing.

Meanwhile, a wave of Arm server chips from Marvell, Ampere, and Amazon have also seen widespread adoption by cloud providers. Arm — which Nvidia has announced a $40 billion bid to acquire — detailed its upcoming Neoverse N2 spec in September, which boasts 50% higher performance than the previous generation Neoverse N1.

According to a recent report from The Wall Street Journal, AMD could be looking to buy Xilinx in a deal worth $30 billion.

Xilinx would add FPGA to AMD's product offerings, allowing the chipmaker to directly compete with Intel — the only other large FPGA manufacturer — in the data center and at the edge. Xilinx’s FPGAs enable a variety of workloads including SDN, virtualized switching, NFV, and AI inference to be offloaded, freeing up the CPU for other applications.

However, there is no guarantee the rumored acquisition will move forward with the paper reporting that earlier talks stalled before restarting again recently, according to people familiar with the deal.