Cisco executives argue that the distinction between product and model companies is disappearing, and that accessing the 55% of enterprise data growth that today’s AI ignores will separate the winners from the losers.
VentureBeat recently sat down with Jeetu Patel, Cisco President and Chief Product Officer and DJ Sampath, Senior Vice President of AI Platform and Software, to gain new insights into a compelling thesis shared by both leaders. They and their teams maintain that every successful product company must become an AI model company to survive the next decade.
When you consider how compressed product lifecycles are becoming, combined with the many advantages of digital twin technology in accelerating time to market for next-generation products, the thesis makes sense.
The conversation revealed why this transformation is inevitable, backed by solid data. The team maintains that 55% of all data growth is machine data that current AI models do not touch. Greg Brockman of OpenAI estimates that we need 10 billion GPUs to give every human being the AI agents they will need, and Cisco’s open source security model, Foundation-Sec-8B, has already had 200,000 downloads on Hugging Face.
Why the model is becoming the product
VentureBeat: You’ve said that in the future every product company will become a model company. Why is this path inevitable and not just a possible path?
Jeetu Patel: In the future there will be no distinction between model companies and product companies. Great product companies will be model companies. The close link between model and product is a closed circuit. To improve the product, you improve the model, not just a UI fix.
These companies that are forming now are a thin wedge on top of a model; His days are numbered. The real moat is the model you build that drives the behavior of the product. This requires being good at two things simultaneously: creating great models in domains where you have great data, and creating great product experiences driven by those models in an iterative cycle where the models adapt and evolve when there are requests for product improvements.
DJ Sampath: This becomes even more critical when you think about the things that happen to agents. Agents will be governed by these models. Your moat will really be how well your model reacts to the changes you need.
Leveraging the growth of machine data is key
VentureBeat: You mentioned that 55% of data growth is machine data, but current models are not trained on it. Why does this represent such a huge opportunity?
Patel: So far, models have been very good at training with publicly available data, generated by humans and freely available on the Internet. But we’re done with how much public data you could track. Where else will you go now? Everything is locked inside the companies.
55% of data growth is machine data, but models are not trained on machine data. Every company says “my data is my moat,” but most don’t have an effective way to condition that data into an organized channel so they can train AI with it and realize its full potential.
Imagine how much log data will be generated when agents work 24/7 and each human has 100 agents. OpenAI’s Greg Brockman said that if you assume every human being has a GPU, you’re three orders of magnitude away from where you need to be; you need 10 billion GPUs. When you think about it that way, if you don’t train your models with machine data effectively, you’re not complete in your ability to realize the full potential of AI.
Sampath: Most models are being trained with public data. The data found within companies is mostly machine data. We are unlocking the data on that machine. We give each company a starting model. Think of it as a starter kit. They will take that model and create applications and agents optimized with their proprietary data within their companies. We will be a model company, but we will also make it incredibly easy for each company to build their own models using the infrastructure we provide.
Why hardware companies have an advantage
VentureBeat: Many see hardware as a liability in the age of software and AI. You argue the opposite. Because?
Patel: Many people look down on hardware. In fact, I think hardware is a great asset, because if you know how to build great hardware, great software, and great AI models and put them all together, that’s when the magic starts to happen.
Think about what we can do by correlating machine data from the logs with our time series model. If there’s a one-degree change in your switch or router, you could predict a system failure in three days, something you couldn’t correlate before. You identify the change, redirect traffic to avoid problems, and resolve the problem. Be much more predictive of infrastructure outages and stability.
Cisco is the critical infrastructure company for AI. This completely changes the level of stability we can generate for our infrastructure. Manufacturing is one of the main industries due to the volume of data that is generated daily. Combined with agent AI and aggregated metadata, it completely changes the competitive nature of manufacturing or asset-intensive industries. With enough data, they can transcend disruptions around tariffs or supply chain variations, taking them out of the commoditization of pricing and availability.
Cisco’s deep commitment to Open Source
VentureBeat: Why make your security models open source when that seems to offer a competitive advantage?
Sampath: The cat is out of the bag; Attackers also have access to open source models. The next step is to equip as many defenders as possible with defense-strengthening models. That’s actually what we did at RSAC 2025 when we launched our open source model. Foundation-Sec-8B.
Funding for open source initiatives has stagnated. There is a growing leak in the open source community, which needs collaborative and sustainable sources of funding. It is a corporate responsibility to make these models available, and it provides access to communities to begin working with AI from a defense perspective.
we have integrated clama widely used open source antivirus tool, with hugging facewhich houses more than 2 million models. Each model is analyzed for malware. We must ensure that the AI supply chain is adequately protected, and we are at the forefront of that.
Patel: We launched not only the open source security model, but also one in Splunk for time series data. These correlate data; time series and security incident data, to be able to find very interesting results. With 200,000 downloads on Hugging Face, we’re seeing resellers start building apps with it.
Taking the pulse of customers after Cisco Live
VentureBeat: Following the Cisco Live product launches, how are customers responding?
Patel: There are three categories. First of all, completely ecstatic customers: ‘We’ve been asking for this for a long time. Alleluia.’
Secondly, those who say ‘I’m going to try this’. DJ shows them a demo with white glove treatment, they do a POC and are shocked to see that it’s even better than what we said in three minutes on stage.
In third place are the skeptics who verify that each advertisement comes out on the exact days. That group used to be much larger three years ago. As it narrows, we have seen significant improvements in our financial results and how the market views us.
We don’t talk about things three years from now, just a six-month period. The payload is so large that we have enough to discuss for six months. Frankly, our biggest challenge is keeping our customers up to date with the speed of innovation we have.
Obsess with customers, not hardware
VentureBeat: How are you migrating your hardware-focused installed base without creating too many disruptions?
Patel: Instead of focusing on ‘hardware versus software’, you start from where the customer is. Your strategy can no longer be a perimeter firewall for network security because the market has moved. It is hyperdistributed. But these days you have firewalls that need efficient management.
We offer you a completely updated line of firewalls. If you want to see what we’ve done with the public cloud, managing egress traffic with Multicloud Defense with zero trust, not just user-to-application, but application-to-application. We have created Hypershield technology. We have built a revolutionary Smart Switch. All managed by the same Security Cloud Control with AI Canvas on top.
We tell our clients they can go at their own pace. Start with firewalls, move to Multicloud Defense, add Hypershield application points with Cilium for greater observability, and add Smart Switches. There is no need to add more complexity because we have a true platform advantage with Security Cloud Control. Instead of saying “forget everything and move on to the new thing,” creating too much cognitive load, we start where the customer is and guide them along the journey.
What’s next: Empower global partners to turn AI into a revenue opportunity
The interview concluded with discussions about the November Partner Summit in San Diego, where Cisco plans major partner activation announcements. As Patel noted, "It takes a sustained and consistent emphasis to get the entire reseller engine going." VentureBeat is convinced that a strong global partner organization is indispensable for any cybersecurity company to achieve its long-term AI vision.
