Nvidia and Cisco have unveiled the USFP optical module, a new high-density interface for AI data centers offering up to 12.8Tbps of total bandwidth.
K2 Cloud has overhauled its infrastructure management by centralizing hardware operations, standardizing rack configurations, and migrating to a leaf-spine Ethernet network architecture.
Yandex has restored core AI development services following drone attacks on its data centers, though some functional limitations and resource quotas remain in effect.
Nvidia and Cisco have unveiled the USFP optical transceiver standard, delivering 400 Gbps per lane and aiming to boost bandwidth density in large-scale AI infrastructure.
AMD has launched the Helion rack system, supporting up to 128 Epyc 9006 processors and 32 768 cores to power AI agent infrastructure.
General Catalyst is building a data center to provide its portfolio startups with direct access to AI servers and GPUs amid industry-wide hardware shortages.
Hitachi has launched a modular wooden data center designed for lower deployment costs, shorter construction times, and improved integration into urban environments.
Intel is pivoting its AI strategy toward inference and agentic workloads, leveraging its Xeon processors and new D-series GPUs to capture enterprise market share.
Oracle is seeking expedited federal approval for the Green Chile pipeline to power its New Mexico data center, as project delays push the expected service date to February 2027.
NTT will construct a dedicated data center in Japan for Anthropic to support the company's AI model operations and regional expansion in the Asia-Pacific market.
Astera Labs has unveiled Orion, a modular rack architecture featuring a 320-lane PCIe switch, scheduled for release in early 2027.
Anthropic has signed its first data center lease in Australia as part of a massive project in Queensland valued at up to AUD 32 billion.
NTT Data Group is in advanced talks to build a dedicated data center for Anthropic in Osaka, with project costs estimated at up to $6.3 billion.
Kioxia has launched the CM9-DLC enterprise SSD, featuring integrated cold plate liquid cooling and capacities up to 32TB for high-performance data center environments.
Intel has launched the D270 AI accelerator, featuring the Xe3P architecture and up to 480 GB of LPDDR5X memory, while omitting traditional graphics hardware to prioritize AI inference.
Intel has launched its GPU D-Series, featuring the Crescent Island accelerator built on the Xe3P architecture to support large-scale AI inference in air-cooled server environments.
Intel has begun sampling its D-Series data center GPUs, focusing on memory capacity and AI inference efficiency for standard server infrastructure.
AMD has introduced the Helion reference design for AI agent infrastructure, featuring support for up to 32,768 Zen 6 CPU cores and integration with Instinct MI350P GPUs.
AMD has unveiled Helion, a 21-inch rack reference design supporting up to 32,768 Zen 6 cores, aimed at scaling infrastructure for autonomous AI agents.
Cisco Systems has introduced a new suite of silicon and optical networking technologies designed to scale AI data center infrastructure and improve data transfer speeds.
