For 40 years, Cisco has been best known for building routers, switches, and other networking technology that connects computers within offices, data centers, and across the internet. Cisco’s also a software business, known for its cybersecurity products and familiar applications like the conferencing and communications platform Webex. And last year, the company announced the $28 billion acquisition of big data company Splunk, part of Cisco’s growing role in powering data-driven AI technology.
Cisco also last year named Jeetu Patel as executive vice president and chief product officer, with the goal of breaking down barriers within the company as it bets big on providing hardware, software, and security infrastructure for the rapidly growing AI sector.
Patel spoke to Fast Company about his role, what the AI-powered transformation looks like inside Cisco and in the industry as a whole, and why the company is well-positioned to play a key role in the future of artificial intelligence.
Can you tell me a little bit about what this AI transformation looks like?
This is an exciting time at Cisco. We’re a $55-billion+ company, but the goal and the hope is that we’re going to be able to operate like the world’s largest startup, meaning we’ll operate at speed with scale.
Every once in a while, you have this kind of massive shift in the market that creates an unpredictable opportunity none of us would have thought about even 10 years ago. It feels like the foundation we’ve been laying for 40 years is now going to come to life.
I frankly feel like there are going to be two types of companies in the world: companies that are really dexterous with the use of AI and companies that really struggle for relevance, and it’ll be the largest shift we’ve seen from a platform perspective in our lifetimes.
We’ve done a lot of studies with customers, and 97% of them are really excited about the possibilities of AI, but only 1.7% feel prepared that they know how to tackle it. So there’s a big gap between the possibilities and what needs to happen. And when you ask them a follow-on question, what holds you back, they answer three things.
The first is, they’re not sure they have the necessary infrastructure or technical know-how to construct that infrastructure. The second is they feel like there are 1,000 ideas within their company, but adoption slows down because safety and security become big concerns. The third area is that they just don’t feel they’ve got the right level of training and skills internally.
In each of those three areas, Cisco can be meaningfully accretive to their mission, and it’s an amazing opportunity for Cisco to shape this entire next wave.
What will Cisco’s role be in that transition?
We will provide infrastructure for AI. You’re going to have many AI agents talking to each other across the data center, as well as physical AI, with robotics and humanoids. A world of 8 billion people will feel like a world of 80 billion from a throughput perspective, because of the digital workers that will be added to the mix. And I think that order of magnitude differential might be a gross understatement.
That means there’s going to be much more demand for high-performance, low-latency, power-efficient networking so one agent can talk to another and coordinate and come back with an answer. You’ll see much more demand for high scale infrastructure, not just around GPUs but around [data processing units (DPUs)] and different types of compute paradigms.
Five key areas will be networking, security, safety, data, and models, and Cisco will participate in each of them and we will partner with others. We just announced a partnership with Nvidia. We just made an investment in Anthropic. We are investors in Scale.AI and we’re investors in Groq. All of these innovative things that kind of refactor the world will start happening, and Cisco will be at the center of it all.
What puts Cisco in a position to lead in all of these different areas?
Cisco historically has been very oriented on different technologies that operated well in their own siloes. But the true opportunity is not in acting like a holding company but as an integrated platform, tightly integrated but loosely coupled.
The first characteristic of a platform is to lower the marginal cost for existing customers for every new technology from Cisco. It’s good for customers and good for us. But the second thing is to compound the value of things you might already have.
For instance, networking is great, but networking without security is just selling pipes. When you take security and bake it into the fabric of the network, you can deliver a trusted network and trusted communications. Cisco can be a secure networking company that’s AI first and completely change the value proposition for the market.
Our goal is to make sure to tie things together, build amazing products people love and talk to friends and family about, operate in an open ecosystem, and make sure we’re at least 10 times better than others in the market. Nobody switches to you for 10% better, but when it’s 10 times better, it’s irresponsible for a customer to not consider switching. We want to make it irresponsible for the customer to not consider Cisco.
What does being AI first mean in the security context?
To handle attacks that are happening at an extremely sophisticated level, you have to have defenses not at human scale but at machine scale. So you have to natively build AI capabilities into your product, not as an afterthought that’s bolted on.
But it’s not just about using AI in your security stack. It’s about securing AI itself. The characteristic of AI that’s scary to companies is that models are, by definition, non-deterministic, which means they’re by definition unpredictable. If I ask the same question twice, I might get two slightly different answers.
But enterprises bank on predictability. You need to have a common way that safety and security can be addressed regardless of which model you’re using, which application you’re using, and how many agents you have.
The huge opportunity is, can we make AI safe and secure so that when it does hallucinate, we’ve got guardrails for it, and when it has toxicity and harmful content, we have guardrails for it. And when something like DeepSeek comes out, we can figure out how to jailbreak it through an algorithmic red teaming exercise, and figure out guardrails organizations can put around it so if they do end up using it, they’ll know it doesn’t behave in an unpredictable way.
That’s where a company like Cisco comes in, as that common, neutral-party security substrate across every model, every application, every cloud, every agent.
A product that just went into general availability is AI Defense, an AI safety and security product. If you believe in a world that’s going to be multi-model and multi-agent, we would be the common security layer that could provide visibility into every company about what data is being used and what applications of AI are being used by developers and users. We can also provide validation of AI models—are they working the way we intend them to work? What would have taken seven to 10 weeks to validate a model can now be done within a matter of seconds or minutes, because we have the algorithmic process to do it.
For example, in the first 48 hours that DeepSeek came out, we were able to jailbreak that model on 50 different prompts that HarmBench had. But then we were able to say, given that, we can now give you guardrails on what you do.
How are the investments and partnerships you mentioned important to your work in AI?
One of the key principles we talk about is that you have to operate with the broader ecosystem. You cannot be a walled garden in this day and age.
With our partnership with Nvidia, Nvidia now includes Cisco in its reference architecture, and where an enterprise has Nvidia and Cisco, they want the two of them to work together. As companies we’re committed to making that happen, and we’ll work with Nvidia and ensure our switches work well with Nvidia SmartNICs, and their GPU clusters will have CIsco networking to connect clusters with low latency and high performance and power efficiency.
How have you worked to break down silos internally?
I can give you an example. I was just with the Splunk team. That’s a very large acquisition—about $28 billion. People on the all-hands call asked what we think about preserving the Splunk culture.
Culture is basically an agreed-upon set of norms of how we’re going to work together. I think Splunk culture is amazing, and not only should we preserve the Splunk culture, we should actually take elements of it and make it part of the CIsco culture, so Cisco can learn from what the Splunkers have learned.
But I also feel that Splunk has a lot to learn from Cisco, so we should integrate elements of Cisco culture into Splunk culture. And if both Cisco and Splunk don’t do certain things well, we might learn from the market. If another company’s doing it well, we should evolve with the market. Just staying stagnant is actually our biggest enemy. We need to constantly be curious, keep learning ferociously, and also remember to keep unlearning certain patterns.
When I explained this to the team, what the possibilities were, it took very little time for them to get on board, and before you knew it, the tempo of the organization changed. We have amazing people in the company, and we have to make sure we can unlock their creativity while still using Cisco’s size and scale.
The good news is, I’ve done this a few times. When I came in I was running the Webex business, and it’s now a beautifully built product with AI infused into its fabric. It does really well with the team that built Webex in the past but had struggled because the clarity was not there from leadership.
We’ve got an amazing broad portfolio that, if we harmoniously weave together, magic starts to happen.
You’re Cisco’s first chief product officer. What’s your vision for that role?
We want to be an AI-first, product-first company. We have to build amazing products that people love, that they talk to their friends and family about. There’s no marketing engine that’s as good as word of mouth.
We want to make sure we really sweat the details and build amazing products that people love. We want to make sure that we stay slightly dissatisfied, slightly paranoid and superbly humble, keeping our heads down, just obsessed about the customer and how we’re going to get better and better.
The reason this role is consequential is the decision-making velocity. You don’t have to go out and get in a room with a committee to make a decision. There’s very clearly one team, and the goal is to make sure we build a platform that accretes value on top of each other. What would have taken us three years to get done, we were able to do within three months. There was that much of a difference in decision-making velocity.
We have a key set of values and guiding principles that we make sure everyone’s aligned on. The people here believe in the power of AI and what it’s going to be able to do.
It’s less about me and more about taking the friction out of the organization to move at speed and move with scale, with a combination of systems thinking and a level of grit, being able to be scrappy. If you can be scrappy with systems thinking, you get the best of both worlds. Operate like the world’s largest startup, and you can do things no startup can do, and things no large company can do.
No comments