SAN FRANCISCO, California: As competition intensifies around the infrastructure powering artificial intelligence, Cisco Systems has rolled out a new networking chip and router aimed at accelerating data flow inside massive AI data centres, putting it head-to-head with rivals Broadcom and Nvidia in a market fuelled by surging investment.

Cisco said that its new Silicon One G300 switch chip, expected to go on sale in the second half of the year, is designed to help the chips that train and run AI systems communicate more efficiently across hundreds of thousands of network connections. The launch positions Cisco to tap into what it estimates is a US$600 billion AI infrastructure spending boom.

Manufactured using Taiwan Semiconductor Manufacturing Co's 3-nanometer process, the G300 includes new features intended to prevent AI networks from slowing down when hit by sudden surges in data traffic. Martin Lund, executive vice president of Cisco's common hardware group, said the chip incorporates multiple "shock absorber" capabilities to keep traffic moving smoothly.

The company expects the chip to improve performance for certain AI computing tasks by up to 28 percent, partly by automatically rerouting data around network disruptions within microseconds.

"This happens when you have tens of thousands, hundreds of thousands of connections — it happens quite regularly," Lund said. "We focus on the total end-to-end efficiency of the network."

Cisco's push underscores how networking has become a central battleground in AI development, alongside processors and software. As AI models grow larger and more complex, the speed at which vast numbers of chips exchange data has emerged as a critical constraint on performance.

Nvidia highlighted the importance of networking last month when it unveiled its latest AI systems, which include a networking chip that competes directly with Cisco's products. Broadcom is also targeting the same market segment with its Tomahawk series of networking chips.