Cisco made it clear at Cisco Live 2025 that their focus on artificial intelligence is not just about software and models but — above all — infrastructure. Networks, data centers, security, observability, and strategic partnerships become the core of a strategy aimed at preparing companies and service providers for an influx of increasingly demanding AI workloads.
Only 28% of companies feel prepared for AI
According to the latest Cisco AI Readiness Index, only 28% of organizations believe their infrastructure is ready to support AI workloads. On the other end, 13% of companies labeled as “pacesetters” — the most advanced in AI readiness — are already seeing increased profitability: 91% report growing benefits thanks to adopting these technologies.
The message from Cisco is clear: those who don’t adapt their infrastructure will fall behind. Jeetu Patel, the company’s President and Chief Product Officer, summarized it forcefully: there will be two types of companies — those that excel at AI and those struggling to remain relevant.
The comeback of private data centers in the AI era
Although public cloud has dominated over the past decade, Cisco re-emphasizes the role of private data centers. In the words of Chuck Robbins, President and CEO, “the private data center is back,” and Cisco’s mission is to help organizations transform them to run AI workloads “anywhere, with seamless operations, observability, and security.”
To this end, the company has launched a range of solutions covering:
- High-speed, low-latency networking, with a focus on switches designed for GPU clusters.
- Security capabilities embedded in the network itself, to protect data and AI models.
- Unified observability platforms, supported by integrating Splunk and Cisco ThousandEyes, enabling visibility “from the client to the cloud” and proactively detecting issues before they impact the business.
The new Cisco Nexus 9300 switches with integrated DPUs exemplify this approach: offloading networking and security tasks, freeing up CPU resources, and optimizing traffic within data centers built for AI. Alongside, solutions like Secure AI Factory (developed with NVIDIA) and Cisco AI PODs aim to simplify deploying “AI factories” ready for production.
Distributed data centers acting as a single entity
The next frontier is linking multiple data centers to function as a single large compute system. Generative AI and large-scale models are exceeding the capacity of a single data center, necessitating the connection of various locations separated by hundreds of kilometers.
For this scenario, Cisco has introduced the Cisco 8223 router — the first fixed Ethernet router with 51.2 Tbps designed specifically for traffic between AI data centers. Based on the Cisco Silicon One P200 chip, this device allows for efficient network scaling and, according to Cisco, consumes about 65% less energy than previous generations in its 3RU configuration and 51.2 Tbps capacity.
The goal is for multiple data centers, even if separated by long distances, to behave as a single logical compute block, reducing complexity and latency impact. The combination of this router, Cisco’s optical solutions, and observability platforms aims to provide a comprehensive solution for distributed AI architectures.
Security, observability, and trust: pillars of enterprise AI
Cisco emphasizes that deploying GPUs and bandwidth alone isn’t enough: without trust, AI won’t deliver expected value. This trust hinges on three main pillars:
- Resilience: infrastructures designed to withstand failures and recover quickly from issues, minimizing downtime.
- Integrated security: tools like AI Defense and the Secure AI Factory with NVIDIA platform protect the entire AI lifecycle — from model development to deployment — including software, workloads, and infrastructure.
- Real-time observability: integrating Splunk and Cisco ThousandEyes offers end-to-end visibility, enabling anomaly detection, swift diagnosis, and automated remediation whenever possible.
Cisco believes that this combination of security, resilience, and visibility will enable organizations to “capture the value of AI” without compromising system integrity or customer and user trust.
A strategy supported by partners and ecosystem
The company also highlights the importance of its ecosystem. Partners such as NVIDIA and Microsoft, along with strategic investments in AI startups like Mistral, bolster Cisco’s approach: it’s not just about selling network hardware but about delivering a complete platform to deploy AI reliably, interoperably, and future-proof.
Additionally, Cisco aims to simplify IT teams’ workflows with increasingly intuitive interfaces, many supported by their own AI and the rise of agentic AI, which automates complex operations and accelerates incident resolution.
Looking Ahead: AI, Quantum, and Beyond
During Cisco Live 2025, Chuck Robbins summarized the company’s outlook: organizations that want to be prepared for AI, quantum computing, and “what comes next” need to build infrastructures rooted in security, integrity, and a deep respect for people. All of this, he emphasizes, is “funded by trust,” and Cisco aims to position itself as a strategic partner to turn that vision into reality.
In a landscape where many organizations remain in testing or experimentation phases with AI, the message is clear: without a solid foundation in networking, data centers, security, and observability, the promise of artificial intelligence may remain just that — a promise. Cisco sees itself as a key player in turning that promise into tangible results.
via: newsroom.cisco