What if your network infrastructure could detect threats that traditional security tools miss entirely? As AI-generated attacks become customized for individual targets and polymorphic malware constantly morphs its signature, endpoint security faces an existential challenge. HPE’s answer involves transforming the network itself into a security solution or platform.
In this interview from RSAC 2026 Conference, David Hughes, who leads HPE’s SASE and security business, reveals why the company deliberately structures its networking division to highlight that SD-WAN represents infrastructure rather than pure security—and why that distinction matters more than most people realize.
Hughes explains how HPE collects telemetry across switches, access points, and gateways throughout their entire customer base to establish behavioral baselines. But the real insight comes when he discusses what happens when an electronic door lock starts behaving differently. Is it a manufacturer update rolling out across the fleet, or has someone breached that specific device? The answer lies in patterns that only become visible when analyzing telemetry at scale.
What this video is about
Our conversation with Hughes gets into a challenge that’s dominating discussions at this year’s RSAC Conference: identity management for AI agents. These non-human entities can emulate human behavior, creating an entirely new category between traditional human and device identities. Hughes shares the emerging approach that links agent identities to human supervisors—but the implications raise questions that security teams are only beginning to grapple with.
You’ll learn why the cloud-based security model that surged during COVID-19 creates serious problems for factory environments where traffic travels east-west. Hairpinning that traffic to the cloud for security decisions introduces latency and costs that aren’t always transparent. Hughes talks about HPE’s hybrid approach and why geopolitical sovereignty concerns are driving enterprises back to on-premises enforcement.
Perhaps most valuable for security leaders, Hughes addresses the tension between networking and security teams. When security demands network changes for threat protection while networking prioritizes performance for new initiatives like agentic AI frameworks, who wins? The answer involves shared visibility, automated policy management, and a fundamental rethinking of how rules get implemented across thousands of devices.
Key topics covered
- Why signature-based security fails against polymorphic and AI-generated threats
- How network telemetry enables behavioral detection at scale
- The emerging framework for AI agent identity management
- Why hybrid enforcement models outperform cloud-only or on-premises-only approaches
- How to bridge the gap between networking and security teams through shared data
- The role of policy abstraction in eliminating device-by-device configuration
- Where HPE focuses its security efforts versus where it partners with specialists
Our conversation with Hughes also offers a perspective on what HPE doesn’t attempt to do. Rather than claiming to cover the entire security spectrum, the company focuses specifically on areas where security and networking overlap: SSE, SASE, SD-WAN, universal ZTNA, hybrid mesh firewall, and data center microsegmentation. For deeper security functions like post-threat forensics, ransomware recovery, and endpoint protection, HPE partners with specialists.
This strategic focus hints towards something important about enterprise security architecture: organizations still need best-of-breed solutions rather than vendors claiming universal capability, even though the individual building blocks can and most likely will continue to grow in size. Watch the full interview to understand how this network-centric approach addresses threats that evade traditional security tools, and why the collaboration between networking and security teams matters more than ever as threat landscapes evolve.