Cisco Launches P200 Chip to Link AI Data Centers Across Long Distances
Cisco has unveiled its new P200 chip and accompanying router, designed to connect AI data centers over vast distances. The solution consolidates what used to require 92 chips into one and reduces power use by about 65%, enabling AI workloads to be distributed across geographically dispersed facilities.
DAte
Oct 8, 2025
Category
AI & Infrastructure
Reading Time
5–6 Minutes
Reuters reports that Cisco launched the P200 chip and a new router aimed at interconnecting AI data centers across large geographic spans. The P200 replaces a complex system of 92 separate chips with a single component and enables the router to use about 65% less power than comparable routing solutions.
The design is especially targeted at linking AI systems distributed across multiple data center sites—“the training job is so large, I need multiple data centers to connect together,” said Martin Lund, Cisco executive VP. Microsoft and Alibaba are among its early customers.
Key Highlights
P200 chip + router launched by Cisco to connect AI data centers over distance
Consolidates 92 chips into one, reducing complexity
Router design uses ~65% less power than previous equivalents
Enables synchronized AI workloads across distant data centers
Early adoption by Microsoft, Alibaba, etc.
Why This Matters
Scalable AI architecture: As AI training scales, connecting dispersed data centers efficiently becomes vital.
Power & efficiency gains: Reducing router power usage directly impacts operational costs and energy footprint.
Infrastructure synergy: Networking is as critical as compute—this shows how the layers beneath AI models must evolve.
Competitive edge: Cisco’s move challenges rivals in networking/hardware that aim to serve AI infrastructure.
Source
Reuters – Full Article
Author
