The development of AI may be hampered by infrastructure limitations, a growing context gap, and a lack of trust: President of Cisco

Cisco President Jeetu Patel said Artificial Intelligence (AI) is transforming software development and innovation, but major challenges could slow its progress. He was speaking at the India AI Impact Summit 2026 on Friday.

Patel said AI has “completely changed and flipped” how modern software is built. However, he identified three main constraints that could limit AI’s growth.

The first challenge is infrastructure. According to Patel, the world does not currently have enough power, computing capacity, or network bandwidth to support the rapid growth of AI. He described infrastructure as “oxygen for AI” and pointed to issues such as limited memory capacity and the slow expansion of data centres.

The second issue is what he called the “context gap.” He compared AI agents working without enough background information to an emergency room doctor treating a patient without access to medical history. Without proper context, AI systems may still make decisions, but those decisions might not be accurate or appropriate. To solve this, Patel stressed the need to connect AI models with enterprise data and machine-generated data such as logs, metrics, events and traces. He noted that 55% of global data growth will come from machine data, especially as AI agents operate continuously.

The third constraint is trust. Patel said the biggest risk is no longer AI giving wrong answers, but AI taking wrong actions. He warned about threats like jailbreaking, prompt injection, tool misuse and data poisoning, and called for stronger safeguards and security systems.

Patel also revealed that Cisco has launched its first product that was entirely built and coded using AI, without a human writing any code. He said this could dramatically speed up innovation, turning the traditional exponential growth curve into something that feels almost vertical. However, he cautioned that while technology is advancing quickly, institutions and society are slower to adapt.

He highlighted a major mindset shift, saying AI should not be seen only as a productivity tool but as an “augmented teammate” that works on behalf of humans. Organisations, he said, must move from a “human-in-every-loop” model to one where “AI is in every loop.”

Patel emphasised that governance must also evolve. Instead of being limited to written policies, governance should become real-time and embedded directly into AI systems as they operate. He described this as “runtime implementation.”

Explaining Cisco’s strategy, Patel said the company is building AI-ready networks, context-enrichment systems and stronger security frameworks, along with full observability — from GPU usage to model performance, applications and AI agents.

Calling AI a major opportunity for India, Patel highlighted the country’s young workforce and strong digital public infrastructure, including Aadhaar and Unified Payments Interface. He said AI works best at scale, and India’s large population and data ecosystem give it a strong advantage.

Patel concluded by saying AI has the potential to help solve global challenges such as disease, poverty and gaps in education, but stressed that governments, companies and institutions must work together to ensure AI remains safe and secure.