New rack to feature 256 liquid-cooled Vera CPUs

Agentic AI and agents were mentioned constantly during the two-hour keynote

Up from $500 billion a year ago

Also establishes new manufacturing facility for modular data centers

OCI-MSA looks to drive open 'plug-and-play ecosystem’ in the notoriously fierce optics market

Dell’Oro Group data shows Celestica, Nvidia, and Arista dominating hyper-competitive market to meet the relentless demand from the likes of Amazon and Meta

Funding from Bosch, MediaTek, and John Doerr to fuel startup’s mission to support millions of GPUs and achieve up to 70% in power savings

Rami Rahim talks Greenlake, AI, and more at MWC 2026

Backlog swells to $7 billion as demand outstrips supply ‘for at least the next several quarters’

GPUs will be deployed across Akamai's distributed cloud infrastructure

Chip giant ships first 2nm custom compute SoC to power Fujitsu's next-gen supercomputer project

Jensen Huang declares ‘compute equals revenues’ as the big green giant continues to defy AI market skeptics

ETX-2i-400G is designed to prevent GPU resources from sitting idle, plus support for QKD and PQC

Powered by Broadcom's Tomahawk 6, open networking vendor's new platforms join a burgeoning market projected to surpass $100 billion by 2030

Specialized storage solution already being used by unnamed edge services neocloud

Benchmarks expose a 'composability' gap that AMD admits it needs to fix

Experts weigh up power, launch cadence, and economic feasibility

Rohit Mittal, a veteran of Google Cloud's custom TPUs and Intel's silicon photonics, joins Nvidia's NVLink Fusion team

Dell'Oro Group report sees Ethernet to prevail over InfiniBand in data center fabrics, even as proprietary technologies like UALink gain traction

E-Tube RF validation to examine an energy-efficient, low-latency alternative to copper

Fabless chip designer bets big on optical interconnects to boost data center capabilities

Hyperscaler leverages a two-tier Ethernet-based topology, custom AI Transport Layer & software tools to deliver a tightly integrated, low-latency platform

Upscale AI wants open protocols like UALlink and Ultra Ethernet to power its SkyHammer line

Investment giant targets operational burdens with software-defined Infrinia AI stack