Cisco Announces New High-Speed Silicon One G300 Chip to Drive Next-Gen AI Data Centers
Published: 18 Feb 2026
Author: Precedence Research
Cisco systems launched its breakthrough product, Silicon One G300 networking chip, meant to radically accelerate the data flow in large-scale AI data centers. The company announced the chip on February 10, 2026, at its Cisco Live EMEA conference in Amsterdam. This is a part of company’s move into AI infrastructure technology.
The newly launched Silicon One G300 boasts the ability to transfer data at the speed of up to 102.4 terabits per second, which makes this networking chip one of the fastest available in the market. This chip is designed specifically to be used in AI systems. This chip was developed using the most modern 3-nanometer technology and is designed to solve the problem of slow data transfer between a cluster of thousands of GPUs.
Cisco claims that the G300’s architecture enables the network to better absorb the flow of data and reroute it in real time, minimizing delays associated with AI calculations. According to Martin Lund, the executive vice president of Hardware at Cisco, the smart features of the chip enables to compute tasks at 28% higher speed as compared to its traditional counterparts.
Cisco will integrate this new silicon chip into their N9000 and 8000 series ethernet systems, which can be deployed with either air cooling or full liquid cooling to match the power requirements of AI workloads. The system will also be able to deploy cutting edge optics to provide high speed connectivity between compute nodes. G300 launch marks a strategic vision of Cisco to stay competitive with rival networking technologies like Broadcom and NVIDIA.
We focus exclusively on this industry and provide insights that help you make better decisions. Get in touch to learn more@ https://www.precedenceresearch.com/ai-data-centers-market