June 20 (Reuters) – Cisco Systems (CSCO.O) on Tuesday launched a networking chip for AI supercomputers that will compete with products from Broadcom (AVGO.O) and Marvell Technologies (MRVL.O).
Cisco said the chip is being tested by five of the six major cloud providers, but declined to name the companies. According to Bofa Global Research, major cloud players include Amazon Web Services, Microsoft Azure, and Google Cloud, which dominate the cloud computing market.
With the growing popularity of AI applications such as ChatGPT, which have networks of specialized chips called Graphics Processing Units (GPUs), the communication speed of these individual chips has become very important.
Cisco is a leading supplier of networking equipment such as Ethernet switches that connect devices such as computers, laptops, routers, servers and printers to local area networks.
The latest generation of Ethernet switches, called G200 and G202, have double the performance of the previous generation and can connect up to 32,000 GPUs, the company said.
“The G200 and G202 will be the most powerful networking chips on the market, facilitating AI/ML workloads and enabling the most power efficient networks,” said Rakesh Chopra, Cisco Fellow and former Principal Engineer. ‘ said.
Cisco says the chip is more power efficient, has 40% fewer switches, has less latency, and could be useful in performing AI and machine learning tasks.
Broadcom announced the Jericho3-AI chip in April, which can connect up to 32,000 GPU chips.
Reporting by Chavi Mehta and Steven Nellis of Bangalore.Editing: Anil DiSilva
Our standards: Thomson Reuters Trust Principles.
