Creator pool will expand as AI democratises coding: Cisco CPO

Networking technology leader Cisco Systems aims to ship at least a half-dozen software products which are fully developed with artificial intelligence (AI) in 2026, said its president and chief product officer Jeetu Patel.

The US-headquartered company is shifting toward “spec-driven development”, where engineers write specifications and AI generates 100% code, he told ET. However, this doesn’t mean that it is eliminating engineers. AI will expand the pool of creators, he said, adding that instead of zero software engineers left, the future could see eight billion software engineers as code generation is democratised.

Patel said strong AI infrastructure demand, not short-term market volatility, will define the next decade for AI. The AI cycle is different from the dot-com bubble, when capital supply exceeded demand, according to him.


“In the dot-com bust… you were doing a forward bet on demand by pre-building supply. That is not the case right now,” he said. “Today, all the capacity that’s being built is being consumed right away… we are still short on compute.”

However, he acknowledged the signs of market froth. “Just because there’s going to be a seismic platform shift does not mean that there are no overvalued companies today,” he said.

Cisco, which was once the most-valued company in the internet era, is reinventing rapidly to new realities in the AI era. It posted the highest quarterly revenue to date of $15.3 billion for the second quarter of 2025-26, with double-digit growth and raised future growth outlook. However, memory shortages and consequential price increases have created short-term margin pressures, Patel said.

When asked about hyperscalers spending massively on AI infrastructure, Patel dismissed that they are overbuilding capacity. “In fact, in the long run, they’re underbuilding it… Autonomous agents will dramatically multiply infrastructure needs. Unlike chatbots, which were productivity tools, agents are like digital co-workers. If each human is augmented by 10, maybe 100, maybe 1,000 agents working 7/24, the infrastructure requirement could scale into the trillions of digital workers,” he said.

Addressing the recent turbulence on global software stocks, Patel said that companies which fail to embed AI may become irrelevant. “I don’t believe that the future is going to be just two or three companies in the world… Independent software companies will persist, provided they innovate on top of system-of-record data,” he said.

Patel outlined how AI scaling laws are driving unprecedented networking needs. Cisco recently launched its Silicon One G300 networking chip to connect massive graphics processing unit (GPU) clusters, competing with the likes of Nvidia and Broadcom. As model companies diversify beyond single GPU clusters and even single data centres, Cisco is building both “scale-out” and “scale-across” capabilities to enable data centres 800 kilometres apart to function like one coherent computer, he said.