Trending
AI chip startup Etched closes $300m funding round, doubles its valuation to $10.3bn Parks S/A to invest R$500m in 5MW data center in Cachoeirinha, Brazil Infleqtion to deploy quantum computer at Illinois Quantum & Microelectronics Park AMD officially launches Helios rackscale system, with MI455X GPUs, Venice Epyc CPUs, and Pensando The 15 Fastest-Growing Digital Health Startups of 2026 Hong Kong property firm ITC inks memorandum to build 1GW data center near Shanghai Google announced as end user of 8 million sq ft data center in Columbia, Georgia Trump expands Ratepayer Protection Pledge with 200-plus new signatories Developer eyes 930-acre data center in Ransom Township, Pennsylvania Intel posts fastest YoY growth since 2011, with Q2 2026 revenue totaling $16.1bn PJM grid hit by voltage disturbance after data center load abruptly drops offline – report DCD Talks: A new definition of speed to power with Patrick Dillow, Rowan Digital Infrastructure AMD launches Instinct MI400 Series GPUs for AI workloads Sabey starts construction on 120MW data center campus in Oregon Empery Digital invests $20m in Cardinal Data Power

The Switch Is the Bottleneck: Why AI Infrastructure Has a Network Problem

AI tasks are progressing faster than network capacities, resulting in costly chips being underutilized. Mark Rushworth clarifies the reason behind the switch being the bottleneck and suggests possible solutions to resolve the issue.

 

Join the conversation

Your email address will not be published. Required fields are marked *