Cisco Takes on Nvidia with the Launch of Silicon One G300 Chip for AI Data Centers
Cisco Systems launched its Silicon One G300 networking chip on February 10, 2026, marking its entry into the $600 billion AI infrastructure market. This move positions Cisco to compete directly with giants like Nvidia and Broadcom. The chip signifies a strategic shift in the AI industry, where networking, not just computing power, is becoming the critical performance bottleneck.
The G300 boasts 102.4 terabits per second of switching capacity and is designed to connect hundreds of thousands of AI chips in massive data center clusters. Built using Taiwan Semiconductor Manufacturing Co’s 3-nanometer technology, the chip features innovative “shock absorber” capabilities that automatically reroute data around network problems within microseconds.
According to Cisco, the G300 can improve AI job completion times by 28% and increase network utilization by 33%. This directly maximizes the profitability of expensive GPU investments.
The chip will power the new Cisco N9000 and 8000 systems, which are expected to go on sale in the second half of 2026. These systems will feature 64 ports of 1.6 Tbps capacity and 100% liquid-cooled designs that improve energy efficiency by nearly 70%.
“This happens when you have tens of thousands, hundreds of thousands of connections – it happens quite regularly,” said Martin Lund, Cisco’s executive vice president.
The launch comes as Nvidia’s latest systems also included networking chips, highlighting how the AI race is expanding beyond GPUs into the networking infrastructure that connects them.