Hewlett Packard Enterprise just made its biggest networking play since acquiring Juniper Networks in 2025. At HPE Discover Las Vegas on June 16, 2026, the company unveiled a sweeping expansion of its self-driving networking portfolio, one that stretches from the network edge all the way into AI factories.

The core move: HPE is integrating its Networking CX switches directly into the HPE Mist platform. Think of it as finally connecting two nervous systems into one brain. Where campus switches, data center switches, and AI networking gear previously lived in separate management silos, they now fall under a single AI-native control plane.

What HPE actually announced

The headline feature is the rollout of fully agentic AIOps capabilities. In English: the network can now detect problems, diagnose root causes, and fix issues in real time without a human touching a keyboard. HPE calls these “self-healing” features, powered by its Marvis virtual network assistant, which is now embedded in HPE Aruba Central.

Marvis isn’t new. But its scope is. Previously limited to campus and branch environments, the AI assistant now extends its autonomous troubleshooting across data center and AI factory deployments. The system watches for anomalies, correlates events across network layers, and takes remediation actions on its own.

HPE also introduced two new switches purpose-built for AI workloads. The HPE Juniper Networking QFX5140 and QFX5252 are optimized for AMD Helios, targeting AI inference and scale-up architectures. These aren’t general-purpose switches being marketed with an AI sticker. They’re designed from the ground up to handle the massive east-west traffic patterns that AI training and inference clusters generate.

On the security front, HPE launched a unified SASE platform that converges SD-WAN and security into a single offering. The goal is straightforward: give enterprises one place to manage both network connectivity and zero-trust security policies, rather than stitching together point solutions from multiple vendors.

The Juniper integration story

Here’s the thing. None of this happens without the Juniper Networks acquisition that closed in 2025. That deal gave HPE two critical assets: Juniper’s Mist AI engine and its QFX switch lineup, both of which are now central to this announcement.

The integration of CX switches into the Mist platform represents the most visible evidence that HPE is actually merging the two companies’ technology stacks rather than just running them in parallel. That’s a meaningful distinction. Plenty of acquisitions in networking end with the acquired product line slowly withering on the vine. HPE appears to be doing the opposite, making Mist the unified management brain while preserving and expanding the hardware portfolios from both sides.

HPE had already released autonomous networking capabilities in HPE Mist and Aruba Central back in May 2026. The June announcement builds on that foundation, extending the agentic features deeper into data center and AI factory environments where the stakes, and the complexity, are considerably higher.

The combined Aruba-Juniper portfolio now covers campus networking, branch connectivity, data center switching, AI factory infrastructure, and security. For enterprise IT teams, the pitch is consolidation: one vendor, one management platform, one AI engine handling everything from a conference room Wi-Fi issue to a GPU cluster network bottleneck.

What this means for enterprises and investors

Look, the enterprise networking market is in the middle of a generational shift. Every major vendor is racing to add AI capabilities to their networking stacks, but the approaches differ wildly. Cisco is building around its own AI networking portfolio. Arista is doubling down on data center and cloud networking. HPE is betting that the winner will be whoever can deliver a truly unified platform that spans every environment where packets flow.

The agentic AIOps angle is particularly worth watching. Network operations teams are expensive and perpetually understaffed. A system that genuinely reduces the number of tickets requiring human intervention doesn’t just save labor costs. It reduces downtime, which in AI factory environments can mean idle GPU clusters burning through electricity while producing nothing.

HPE’s emphasis on reduced total cost of ownership through advanced GPU utilization speaks directly to the economics that AI infrastructure buyers care about most. When a network issue causes even a small percentage of GPUs in a training cluster to sit idle, the financial impact compounds quickly. Self-healing networks that resolve issues in seconds rather than hours change that math considerably.

The zero-trust security integration also addresses a growing pain point. As enterprises distribute AI workloads across edge, campus, and data center environments, the attack surface expands dramatically. A unified SASE platform that enforces consistent security policies across all those environments is table stakes for any organization serious about protecting its AI infrastructure.

For investors tracking the AI infrastructure buildout, HPE’s moves signal that the competitive landscape in networking is consolidating around a few vendors capable of delivering end-to-end solutions. The company’s ability to successfully integrate Juniper’s technology, rather than just its revenue, will likely determine whether HPE can sustain its position as organizations scale AI deployments from pilot projects into production workloads. The QFX switches optimized for AMD Helios suggest HPE is also making targeted bets on specific chip architectures, a strategy that could pay off if those platforms gain traction but introduces concentration risk if the market shifts toward alternative silicon.