Y CombinatorHow Jeetu Patel Runs Cisco Like the World’s Largest Startup
Y Combinator
Cisco’s AI posture 1:01
Jeetu Patel says Cisco should act as the picks and shovels company for the AI era. He wants the company to provide the infrastructure that lets AI run securely and safely at scale. He also says the work depended on the team setting direction, hiring well, and choosing the right markets.
Founder mode at scale 3:00
He describes his style as founder mode, where you act like an owner and are willing to poke into every part of the business. He says that works best in fast-changing times, but it has to be balanced with people who know how to run a large company. At Cisco, he uses a rule of thirds across the leadership team: scale operators, outside systems thinkers, and leaders from acquired companies.
AI means more demand 6:01
Patel says Cisco is in wartime because the shift to AI infrastructure is creating a much larger opportunity than before. He argues that the company has to move with both speed and scale, like the world’s largest startup. He points to low current agent use, heavy infrastructure demand already, and the need for safety, security, and observability because agents can go rogue or be influenced in harmful ways.
Forcing AI adoption 9:00
Inside Cisco, he says the goal of AI is not just efficiency but new insight and new kinds of problem solving. He expects the company to need more engineers, not fewer, as AI creates new human bottlenecks. To drive adoption, he told employees that if they do not use AI, they will lose their jobs, and he gave them unlimited tokens to start experimenting. After early resistance, he saw code refactoring speed up sharply and made AI a top-down priority with regular leadership reviews.
Stack planning cycles 14:30
Hardware and silicon move on much longer clocks than software. Silicon can take five years of planning, because the team has to line up IP, manufacturing, and production well ahead of time. Hardware often needs 18 to 24 months, software and operating systems about 12 to 18 months, while models, agents, and apps may change on a three-month or even weekly cycle.
Data centers and agents 17:00
The data center buildout is driven by GPUs, but the network has had to grow from inside one server to across servers, racks, and even separate sites hundreds of miles apart. Cisco focuses on scale-out and scale-across networking, plus silicon and photonics. The business went from zero to $9.3 billion in orders in two years, and the speaker argues there is no real overbuild because AI demand is consuming infrastructure almost as fast as it is built.
Trust and security 25:00
AI safety is framed as the key issue now. The bigger question is whether you can trust systems enough to delegate work and personal tasks to them. Cisco’s answer is an agentic security platform with visibility into models, identity for non-human agents, inventory of running agents, model validation, red teaming, and runtime guardrails. Its AI Defense product already covers the chat era and now extends to agents.
Early struggle and ambition 29:30
He says school left him with an inferiority complex. He stuttered, stayed quiet, and could not even say hello on the phone. Waiting on tables changed that. It showed him he had to entertain people if he wanted to make money.
Choosing the right path 30:00
He describes buying a company while still an intern and borrowing 250,000 dollars to do it. He ran that business for 17 years, but later saw it was a poor fit for his real goals. He stayed too long because of ego, then learned more in his first year at EMC than in the 17 before it.
Learning at scale 33:00
At EMC and later at Box, he learned the value of hard lessons, good mentors, and product work. He says Cisco gave him the chance to shape the company’s direction, and he asked for freedom to stay product-obsessed. That move led to the biggest growth of his career.
Ambition and time 35:30
He tells younger founders to think big, put in the time, and not confuse ambition with ability. Business is a team sport, he says, and relationships matter. He also stresses compounding: get a little better every day, think in 20-year windows, and spend time on work that still feels right when you look back at 80.
Use your platform 39:30
He ends with a story from the Taj Mahal, where a guide spoke many languages to meet visitors where they were. That, he says, showed how much a platform matters. If you have education, geography, or tech on your side, do not waste it. In AI especially, this is a rare time to make an impact, and you should keep learning with curiosity.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
