Cisco President and CPO Jeetu Patel disclosed in an interview that AI orders from hyperscalers reached $9.3 billion over two years; we judge that AI spending is extending from models to networking and security.
What this is
Cisco traditionally provided routers and switches that connect computer networks; it is now applying that capability to GPU (graphics processing unit) clusters and data centers. AI Agents—AI programs that can autonomously invoke tools to complete tasks—also require permissions and access control, so security requirements are extending into the agent invocation process.
Patel argues for making Cisco's 32,000 engineers behave more like a startup team: faster experimentation and cross-team collaboration. He believes AI should help generate new ideas rather than merely cut costs, so the near term requires more engineers, not fewer.
Industry view
The $9.3 billion order book is a signal: AI budgets now cover not only models but also the supporting infrastructure that connects compute and keeps data centers running. Notably, buyers are starting to pay for capabilities beyond the model, with networking and security becoming the new landing points for AI investment.
But orders do not equal recognized revenue, and the interview did not disclose margins. Patel argues data centers are not overbuilt; we do not treat that as a settled conclusion: if demand falls short, equipment utilization and cash collection cycles will still come under pressure. Strong technical demand does not mean every infrastructure investment pays back on time.
Impact on regular people
- For enterprise IT: Budgets may shift further toward cluster interconnect, access control, and agent security; procurement evaluation will expand from model demos to delivery and isolation capabilities.
- For individual careers: Engineering roles may gain more networking, data, and security tasks rather than being uniformly replaced. Individual competitiveness will depend more on problem definition and cross-team collaboration; the value of merely demonstrating tool operation may decline.
- For the consumer market: This kind of spending first affects enterprise procurement and cloud pricing, not necessarily new household-facing products; whether it ultimately translates to better services depends on equipment utilization and market competition.