Cisco Systems (NASDAQ:CSCO) on Tuesday introduced a new chip and router designed to accelerate data movement inside large-scale data centres, marking a push to take on competitors such as Broadcom and Nvidia as spending on AI infrastructure surges toward an estimated $600 billion.
The newly launched Silicon One G300 switch chip is slated for commercial availability in the second half of the year. It is built to improve communication between the chips used to train and deploy artificial intelligence systems, supporting networks that can span hundreds of thousands of connections.
Cisco said the chip will be produced using Taiwan Semiconductor Manufacturing Co’s advanced 3-nanometer manufacturing process. One of its standout features is a new “shock absorber” function, designed to prevent AI chip networks from slowing when confronted with sudden spikes in data traffic.
Martin Lund, executive vice president of Cisco’s common hardware group, said the company expects the technology to allow certain AI computing workloads to complete 28% faster. He attributed the gains in part to the chip’s ability to automatically divert data around network issues 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.”