A significant shift is underway in artificial intelligence, moving beyond the initial fascination with chatbots towards systems that act autonomously. This evolution of AI, according to Jeetu Patel, president and chief product officer, Cisco, is fundamentally altering the requirements for computing infrastructure, compelling new strategies for data centre design, workforce integration and security.
“The next era of AI is here,” Patel said, framing this transition as distinct from the period ignited by tools like ChatGPT. “We are moving from two and a half years ago, when ChatGPT made its seminal debut in November 2020. We had this concept of chatbots that intelligently answered our questions, and it felt almost magical. Today, we are transitioning from that era to one where agents will autonomously perform tasks and jobs on behalf of humans. This shift is going to have profound implications on the world.”
This move towards autonomous AI agents promises not just smarter assistants, but a fundamental reconfiguration of the workforce itself. In his address at the company’s annual event Cisco Live, Patel envisions a future defined by collaboration between humans and machines.
“Right now, 100% of the workforce we engage with is human,” he said. “But in the near future, we will see significant augmentation to this workforce—agents, applications, robots and humanoids will all work alongside us, enhancing the overall productivity of the workforce.” The potential impact on global capacity is staggering, with Patel projecting that “Eight billion humans will soon feel like 80 billion, thanks to these enhancements.”
However, supporting this augmented, AI-driven workforce places unprecedented demands on the underlying technological infrastructure. Patel explained several critical constraints emerging: “Moreover, AI is going to be network-constrained, compute-constrained and power-constrained.” This multifaceted pressure necessitates a departure from traditional cloud computing models.
“It won’t just be about public clouds anymore,” Patel said. “A combination of public and private cloud infrastructures will be necessary, especially for inferencing workloads, as organisations will want to have their own private cloud data centres built out for AI.” This hybrid approach becomes crucial as the nature of AI workloads evolves; unlike the sporadic bursts of activity characteristic of early chatbots, autonomous agents generate sustained, continuous demand for inferencing power.
Amidst this technological upheaval, Patel stressed that trust remains the non-negotiable foundation for adoption. “If we don’t trust AI, we won’t use it,” he stated bluntly. He framed the strategic imperative for businesses in stark terms, suggesting that “In the future, there will be only two types of companies: those that are extremely adept at using AI and those that struggle to remain relevant.”
For Cisco, responding to these converging demands means focusing on three core pillars: building secure, high-performance Infrastructure for connectivity, safety and security specifically designed for AI-ready data centres; future-proofing workplaces across diverse environments like campuses, branches and vehicles; and ensuring the resilience of the Infrastructure.
“We need to make sure that this infrastructure is resilient, meaning if there is an outage, you can recover quickly,” Patel elaborated. The networking giant leverages what it sees as key structural advantages: a platform advantage integrating its broad technology portfolio to reduce complexity, a silicon advantage utilising custom, programmable ASICs, and embedding “AI as a foundational principle” across its operations. “AI is not an afterthought for us; it’s a foundational element,” Patel said.
Central to supporting the new wave of AI is a corresponding surge in data centre construction. “We are witnessing one of the largest expansions of data centres in history,” Patel said. “It’s an exponential curve, and it’s getting faster every day.” Cisco is actively involved in major global build-outs, citing partnerships with entities like the Government of Saudi Arabia and G42 Stargate in the UAE.
“These partnerships will continue, and we are committed to participating in these massive data centre build-outs, which will need to be network-constrained, secure, and resilient,” he said.
To address the specific technical challenges, the San Jose-headquartered firm unveiled several upgrades. Patel announced flexible server capacity with GPUs, allowing customers to “scale up your servers as your needs grow, starting with two GPUs and scaling to eight or more.”
He introduced hyper fabric as “fundamentally reimagining the data centre lifecycle,” and intersight as “a world-class management plane for managing UCS servers.”
Recognising the unique vulnerabilities of AI systems, the networking major is also launching AI Security, described by Patel as “a revolutionary product designed to secure AI itself, across all models, agents, clouds and applications.” For organisations seeking integrated solutions, AI Pods offer a “full-stack solution in a reference architecture, combining compute, security, storage, and networking in a single box.”
Further simplifying operations, Patel announced the unification of the Nexus Dashboard, combining Nexus OS and ACI fabrics. “I am thrilled to announce the unification of the Nexus Dashboard… This is a major milestone, and it is something you have been asking for for a long time. This unified solution will ship in July,” he said.
He also highlighted the deepening partnership with Nvidia, involving integration of Nvidia’s technologies into Cisco’s offerings, including the Cisco AI secure AI factory with Nvidia and embedding Cisco switches within the NVIDIA Spectrum X architecture. Cisco’s AI defense capability will now extend to securing models built with Nvidia’s Nemo framework.
