Cisco Systems has launched the P200, a networking chip built to link AI data centers spread across hundreds or even thousands of miles. The chip will anchor Cisco’s new router lineup and has already gained adoption from Microsoft Azure and Alibaba Cloud.
AI firms are increasingly dispersing their data centers to areas with cheaper power, such as Texas and Louisiana, driving demand for ultra-fast interconnections. “AI training jobs now require multiple data centers acting as one,” said Martin Lund, Cisco’s executive vice president.
Replacing 92 separate chips, the P200 reduces power consumption by 65%, improving both energy efficiency and data transfer speeds. Its advanced buffering system helps maintain synchronization between massive AI clusters — a long-standing technical bottleneck for the industry.
Microsoft’s Dave Maltz said the P200 “offers more options for faster, scalable cloud networks,” while analysts view it as Cisco’s strongest challenge yet to Broadcom’s dominance in data networking for AI.
Cisco’s move underscores the infrastructure race behind artificial intelligence, where innovation in hardware is becoming as critical as breakthroughs in software.
