Daisy Chittilapilly, president, Cisco India & South Asia, says enterprises are entering a new phase of AI adoption, where infrastructure, networking and security are becoming critical to moving beyond pilots. As AI agents take on more autonomous tasks, networks face rising traffic and new security risks. Chittilapilly, in an interview with Sudhir Chowdhary, discusses India’s readiness for AI deployment, the sectors leading adoption and what businesses need to build to scale AI securely. Excerpts:
AI is moving from pilots to production. What are the biggest infrastructure bottlenecks enterprises must overcome to scale AI?
It’s a three-layered problem. First is infrastructure – power, compute and networking, each reinforcing the other, with the gaps becoming visible only when pilots move to production. India is expected to need 8GW of data centre capacity by 2030, with around $70 billion in infrastructure investment expected by 2026. The question is whether organisations are simply adding capacity or building AI-ready architectures that can scale.
Then comes data. Pilots run on clean data; production doesn’t. Fragmented data quickly translates into outputs you can’t trust. The third, and most underestimated, layer is security and trust. In India, only 32% of organisations are integrating AI into their security frameworks, while just 45% feel equipped to control and secure their agents, according to the Cisco AI Readiness Index. Security can’t be bolted on later. It has to be built in – from silicon to application. The winners will be those that bring power, compute, networking, data and security together as one AI-ready foundation.
Cisco has spoken about an AI-driven networking supercycle. What is changing in enterprise networks, and how significant is this opportunity?
When people think about AI infrastructure, they naturally think about GPUs and compute. That’s only part of the story – the network is just as central, and it’s the part most organisations overlook until it becomes the bottleneck. Today, we’ve moved from chat to agentic AI, and agents don’t behave like software waiting for a human click – they reason, act, and loop back continuously, at machine speed.
That makes agentic AI network-bound in a way nothing before it has been: a single agentic task generates 450% more traffic than a human doing equivalent work, roughly 70% of it inference, as per a recent Cisco report. In India, two-thirds of networking leaders are planning major GenAI deployments next year, but you can’t run tomorrow’s AI on yesterday’s network. That’s why we built Cisco Cloud Control – one platform, one login, bringing networking, security and observability together so humans and agents work off the same data. This isn’t an upgrade. It’s a rebuild.
As agentic AI becomes mainstream, what new security risks concern you the most, and how should enterprises prepare?
The AI risk today isn’t a wrong answer – it’s a wrong action. Agents access systems, move data, make decisions autonomously, so the first challenge is visibility: you can’t secure what you can’t see, and Shadow AI only widens that blind spot. The second is trust – confidence that AI is behaving as intended while being protected from external attacks. It’s protecting AI from the world, and the world from AI.
Our AI Readiness Index found that in India, only 38% of organisations have integrated AI into their security and identity systems – that gap is one of the biggest barriers to scaling AI. At Cisco, we believe security has to be built into the AI lifecycle from the beginning – not added later. Whether it’s AI Defense, DefenseClaw or our broader Secure AI strategy, our focus is on helping customers discover AI assets, understand their risks and put the right guardrails in place. Trust isn’t what slows AI adoption. It’s what enables organisations to scale AI with confidence.
India has talent and digital public infrastructure. What will it take for the country to become a global AI deployment and innovation hub?
India has the ingredients – world-class talent, a thriving startup ecosystem, strong digital public infrastructure and government backing. The next step is bringing them together to build AI at scale.
That means investing in AI-ready infrastructure, resilient networks, data centres and cybersecurity; enabling responsible innovation through trusted governance; and building an AI-ready workforce. India’s demographic advantage gives it a strong base to do that.
Industry has a role too. Cisco’s Networking Academy has trained over 3 million students, while its new Technology Hub on AI and Cybersecurity with IIT Delhi aims to advance research, skills and the wider ecosystem.
Which sectors in India are moving fastest from AI experimentation to enterprise-wide deployment, and what is driving that shift?
BFSI and government are ahead – you see it in Bhashini, the National Health Mission’s diagnostics work, and banking’s fraud protection and risk management. Customer service is also transforming across every sector because that’s where human and AI agents are genuinely working side by side today.
Manufacturing is the one to watch next, as factories get more connected and AI accelerates IT-OT convergence, unlocking machine data for predictive maintenance and computer vision-led quality inspection.
What connects all of these is that organisations have stopped adopting AI for its own sake. They’re investing where it improves productivity and business outcomes. That shift, from experimentation to enterprise value, is the real milestone.
The AI infrastructure market is becoming intensely competitive. Where does Cisco see its biggest opportunity to differentiate?
As organisations move beyond experimentation, they need an integrated foundation – networking, compute, security, storage, data and AI operations working as one – and that’s where Cisco is uniquely positioned.
Our differentiation starts with connectivity: we’re the only networking provider across the connectivity spectrum, not just inside the data centre but data centre-to-data centre and data centre-to-enterprise. The bigger piece is security, embedded at every layer of the AI stack, from our own silicon to application software – a full-stack position, extended now through Secure AI Factory with Nvidia.
That’s the differentiated value we bring – not a point solution, but the integrated architecture that AI workloads can count on.
Looking ahead, what will separate AI leaders from laggards – better models, stronger infrastructure, trusted security, or organisational readiness?
All of the above. Most enterprises have similar choices of AI models. What will set them apart is the foundation underneath – secure, resilient infrastructure built from edge to cloud, with integrated security and end-to-end visibility. Organisations that treat AI as separate from infrastructure and governance will struggle. The leaders will be those that build the foundation before they need it, ready to operate at machine speed.