Cato Networks is pushing its SASE platform in a new direction, one that reflects how quickly AI is reshaping both enterprise workflows and security risks.
The company has introduced two updates: a GPU-powered infrastructure layer called Cato Neural Edge and a new Cato AI Security capability focused on governing how AI is used across the business. The platform updates are about where network security is heading as AI becomes part of everyday operations.
Why AI Is Forcing a Rethink in Security
Security tools were built for a world where activity was easier to define. Files were moved, users logged in, and applications were accessed. But that model started to break when AI became part of everyday workflows. Employees now interact with copilots and chat interfaces as part of their daily work. Applications are making decisions through embedded models and more and more autonomous agents are taking action across systems without direct human input.
Such interactions are harder to inspect because they are contextual and conversational. The risk is no longer just malware or unauthorized access. Sensitive data can be exposed through prompts, AI-generated outputs can leak intellectual property, and models can behave in ways that are difficult to predict.
At the same time, security teams are relying more on AI to detect threats and correlate signals across environments. That creates a two-sided challenge. Organizations need to secure AI while also depending on AI to do the securing. This is where infrastructure starts to matter in a way it didn’t before.
Why Cato Is Embedding GPUs Into Its Network
Cato’s Neural Edge introduces NVIDIA GPUs directly into its global backbone, across more than 80 points of presence. Instead of sending traffic out to external environments for AI analysis, inspection happens inline, within the network itself.
That design choice addresses a practical problem. AI-driven inspection requires significant computing power. Many vendors handle this by offloading analysis to cloud-based GPU environments. That approach works, but it introduces latency, inconsistency, and a separation between where decisions are made and where enforcement happens.
Brian Anderson, global field CTO at Cato Networks, explained to ChannelE2E that the issue is not whether AI is used, but where it runs.
“Many vendors use AI for detection, but the key architectural question is where the AI runs. In many architectures, traffic inspection happens in one place, and AI analysis happens somewhere else. This is often in a hyperscaler GPU environment. That separation introduces additional latency variability, and it breaks the tight loop between analysis and enforcement.”
By embedding GPUs directly into its PoPs, Cato is trying to keep that loop intact.
“With Cato Neural Edge, we embed NVIDIA GPUs directly inside the PoPs of Cato’s global private backbone. This allows AI models to run inline where inspection and enforcement already happen, enabling real-time semantic analysis without sending traffic to external GPU environments. The advantage is that we’re ensuring deterministic performance and consistent enforcement at global scale.”
For users, this could mean more predictable performance and fewer blind spots. For security teams, it reduces the complexity of stitching together multiple systems to get a complete picture. The shift here is subtle but important. Network infrastructure is no longer just transporting traffic. It is becoming an execution layer for AI.
What Cato AI Security Adds on Top
Alongside the infrastructure change, Cato is introducing a new layer focused on governing AI usage itself. Cato AI Security brings together controls for employee use of third-party AI tools, internally built AI applications, and autonomous agents operating across workflows. This capability builds on technology Cato gained through its acquisition of Aim Security, which focused specifically on AI governance and protection.
As Anderson explains, “Cato Networks already had the platform foundation. What Aim Security brought was technology focused on governing employee use of AI tools, securing homegrown AI applications, and enforcing guardrails for autonomous AI agents.”
That functionality is now integrated into the broader platform experience.
“AI security has now been converged into the Cato SASE Platform, which means that customers can manage the solution through the same console alongside other capabilities including SD-WAN, SSE, and UTZNA.”
That integration matters because AI security is starting to fragment. Many organizations are experimenting with separate tools to monitor AI usage, scan prompts, or enforce policies. Each new tool adds another layer to manage. Cato’s approach folds AI governance into existing controls. It treats AI activity as another type of traffic and behavior to monitor, rather than something that requires a separate system.
What This Means for Enterprises
The immediate takeaway is that AI security is moving closer to the network layer.
Instead of being handled by standalone tools or add-ons, it is becoming part of how traffic is inspected, policies are enforced, and activity is analyzed in real time. This could reduce the trade-offs organizations often face between performance and deep inspection, while also making governance more centralized.
It also means infrastructure decisions will have a more direct impact on security outcomes. Where compute happens now affects how quickly threats are detected and how consistently policies are enforced.
As AI adoption grows, organizations may need to rethink how their security stack is structured, especially if current approaches rely on external processing or disconnected systems.
What This Means for MSPs and MSSPs
For service providers, this shift changes how AI security can be positioned and delivered. Cato is framing AI Security as a starting point for customer engagement rather than a competing product.
“We see Cato AI Security as a new entry point into the Cato SASE Platform, not a competing motion. Enterprises need to address AI risk immediately, and offering Cato AI Security as a standalone capability allows partners to engage customers around AI governance without requiring a full SASE platform from day one,” Anderson says.
That approach creates a different kind of sales motion.
“For partners, that actually expands the opportunity. Once Cato AI Security is deployed, it creates a natural path to expand into other capabilities like SSE, SD-WAN, or UZTNA. Rather than creating channel conflict, Cato AI Security gives partners a new selling motion that can accelerate platform consolidation over time.”
In practical terms, this allows partners to start with AI governance and expand into broader network and security services over time, rather than leading with a full platform transformation from the outset.
Cato’s update points to a broader change in how security platforms are evolving. AI is influencing how infrastructure is built, where compute happens, and how security decisions are made in real time. Embedding GPUs into the network is one response to that shift. Integrating AI governance into core platforms is another. If AI is becoming part of everyday enterprise activity, then the systems that secure that activity need to operate at the same speed, scale, and context.
