Mohammedia – As businesses move beyond small AI trials and into production use, Cisco Systems announced yesterday a slate of new products aimed at helping companies run artificial intelligence systems at scale.
The announcements came at Cisco’s AI Summit in San Francisco, where the company’s focus was on the infrastructure needed to operate AI systems reliably across large networks and data centers.
Cisco introduced a new networking chip, the Silicon One P200, which it said is designed to handle the heavy data traffic generated by AI systems. The chip delivers up to 51.2 terabits per second of throughput on a single device.
The design is intended to reduce network slowdowns that can occur when large numbers of processors exchange data at the same time.
The company also announced the Cisco 8223 Router, a fixed-form-factor router built to connect AI workloads across multiple data centers.
The device supports 800-gigabit connections and can transmit data over distances of up to 1,000 kilometers using coherent optical technology.
The router is aimed at organizations that need to distribute computing workloads across locations because of power limits, physical constraints, or regulatory requirements.
On the software side, Cisco announced AgenticOps, a framework intended to automate parts of network operations.
Read also: Web Summit Qatar: Morocco’s Seghrouchni Outlines AI-Backed Digital Transition
The system is designed to allow software agents to detect problems, suggest corrective actions, and, in some cases, carry out changes with human oversight. Cisco said the approach is meant to reduce the amount of manual intervention required to manage complex enterprise networks.
Cisco also introduced AI Canvas, a new interface that allows network and security teams to interact with systems using plain-language queries. The platform can generate real-time views of network conditions and pull data from across networking, security, and application environments, according to the company.
In security, Cisco announced AI Defense, a product focused on protecting AI models from misuse and unintended data exposure. The tool can test AI systems for vulnerabilities before deployment and monitor their behavior once they are in operation.
Cisco did not disclose pricing for the new products. Availability will vary by product, with some offerings expected later in 2026.
Companies have increased spending on AI-related infrastructure over the past year, as they look to move experimental systems into regular business use. Cisco said the new products will integrate with its existing networking and security platforms.