0
Deep & Emerging Tech·September 30, 2026·1 min read

CScale exits stealth with $145m to stop laser failures stalling AI

Share

CScale raising $145M for optical interconnects to prevent laser failures in AI data centers shows that photonics reliability is now a first-order scaling constraint. Infra teams should be asking vendors not just about bandwidth and latency, but about failure modes, monitoring, and swap economics for optical links at cluster scale.

Deep & Emerging Tech

US government reveals $5.25 billion spending on boosting national grid to help power new AI data centers — money will cover 31 projects across 26 states, but how will it affect energy bills?

A $5.25 billion federal push into wires and sensors instead of new plants confirms that grid capacity and reliability are now the gating factors for AI data center growth. If you’re planning large-scale AI infra, your real bottlenecks are interconnect queues and transmission upgrades, not just GPUs.