HPE has unleashed a comprehensive wave of hardware and software innovations designed to help enterprises build, secure, and manage large-scale AI infrastructures. Announced at the HPE Discover event in Las Vegas, the new offerings span data center switches, AI-native network management, unified SASE, and deeper integration with Nvidia for AI agent support. These moves represent a major step forward in HPE’s strategy to become the leading infrastructure provider for AI workloads, especially as enterprises shift toward agentic AI and autonomous operations.
The announcements come at a time when AI workloads are placing unprecedented demands on networking. Traditional data center architectures struggle with the massive bandwidth, low latency, and congestion management required for GPU-to-GPU communication. HPE’s response is a tightly integrated portfolio that combines the strengths of its Juniper acquisition with its own AI expertise, particularly through the Mist AI and Marvis AIOps engine. By unifying management across wired, wireless, and SD-WAN, HPE aims to deliver a “self-driving network” that can proactively detect and resolve issues without human intervention.
New QFX5140 Switch Targets AI Inferencing and Edge
On the hardware side, HPE introduced the HPE Juniper Networking QFX5140 switch, a 1RU, 16T fixed-configuration data center switch designed specifically for AI inferencing and scale-up architectures. The switch supports 24Ă— 400G QSFP112 ports, 8Ă— 800G OSFP800 ports, and 2x SFP28 ports, and includes RDMA over Converged Ethernet (RoCEv2) support. RoCEv2 is critical for AI because it allows direct memory access between GPUs over Ethernet, reducing CPU overhead and latency. Key congestion management features such as Priority Flow Control, Explicit Congestion Notification, and dynamic load balancing ensure efficient GPU-to-GPU communications, a critical requirement for AI workloads.
The QFX5140 fills the mid-tier of the QFX family, slotting between the high-end 102T QFX5240/QFX5250 and the entry-level 100GbE QFX5100. HPE also announced the QFX5252 module for its AMD Helios turnkey AI package, which integrates 72 GPUs per rack with CPUs and open Ethernet networking into a unified platform for AI training and high-volume inferencing. The Helios system is designed for organizations that need a pre-validated, high-performance AI cluster without the complexity of integrating components themselves. Both the QFX5140 and the QFX5252 are now managed through HPE’s Data Center Director platform, which provides a centralized, automated view of all network components. This integration improves visibility into network health, speeds root-cause analysis, and reduces the time spent on manual configuration.
Mist AI Integration Deepens Across Networking Portfolio
A major theme of the announcements was the tighter integration between Juniper’s Mist AI and HPE’s Aruba Central management platform. HPE will integrate Juniper’s natural language Mist AI engine into Aruba Central, and vice versa, powered by the core AIOps engine Marvis. Marvis collects telemetry and user state data from Juniper routers, switches, access points, firewalls, and applications to detect and resolve enterprise networking issues. A key component is Marvis Actions, which uses AI to identify and prioritize network problem remediation. By the end of 2026, Marvis Actions will be extended to Aruba Central, enabling proactive resolution of wired, wireless, and SD-WAN issues.
Furthermore, HPE is melding the Aruba CX switching portfolio with Mist AI, providing CX customers with AI-native visibility, zero-touch provisioning, wired assurance for Layer 2 access, service-level insights, and Marvis-driven support. This expansion brings the “self-driving network” concept from campus environments into the data center. For example, Mist can now use AI/ML to predict potential optics failures before they cause outages. Additionally, a new advanced reasoning AI agent within Mist can autonomously reason across diverse data streams, including millions of TAC cases and a contextual graph database from HPE Networking Data Center Director, to deliver precise root cause analysis in minutes. According to HPE executives, problems that once took hours or days to diagnose can now be resolved in minutes or even proactively before users notice any issue.
The integration of Mist AI into the data center represents a significant step. Historically, Mist was focused on campus and branch environments, but HPE is now applying the same AI-driven operations to the core network. This aligns with the industry trend toward “AIOps” for network management, where machine learning models analyze patterns and automate responses. For enterprises running AI workloads, this level of automation is essential because the scale and complexity of GPU clusters make manual troubleshooting impractical.
Unified SASE Orchestrator Simplifies WAN Security
HPE also unveiled a new SASE Orchestrator package that ties together its SD-WAN and SSE capabilities with cloud security and a unified policy engine. The orchestrator uses AI to manage branch, remote user, and cloud connectivity from a single console. It promises simpler operations via AIOps, faster zero-trust adoption, and improved user experience through intelligent traffic steering and application awareness. Customers can set security policies once and deploy them across many sites, reducing complexity and operational overhead. The SASE platform is built on HPE’s existing Aruba EdgeConnect SD-WAN and the Silver Peak technology, combined with cloud-delivered security from the Zscaler partnership and HPE’s own security services.
With the rise of hybrid work and distributed AI deployments, secure access to data center resources is more important than ever. The SASE Orchestrator allows IT teams to enforce consistent security policies for branch offices, remote users, and cloud applications, all while leveraging AI to optimize traffic flows. For example, the orchestrator can prioritize AI inference traffic over less time-sensitive data, ensuring low latency for GPU access. HPE claims this approach can reduce total cost of ownership by eliminating separate management tools for networking and security.
Nvidia Integration and AI Agent Management
HPE deepened its partnership with Nvidia through the HPE Private Cloud AI platform, a turnkey AI factory co-engineered with Nvidia. The platform now supports Nvidia’s Agent Toolkit software, including Nemotron open models, NemoClaw, and OpenShell secure runtime, to provide an agent operating system that reasons, monitors agent behavior, enforces policies, and reduces deployment risk. Additionally, HPE is bringing Nvidia Confidential Computing to the HPE AI Factory via HPE Services, ensuring data and model privacy during execution for on-premises or sovereign deployments. This is particularly important for regulated industries like finance, healthcare, and government where data sovereignty and confidentiality are paramount.
The Private Cloud AI package delivers a preconfigured hardware and software stack featuring the latest Nvidia AI Enterprise software and blueprints. It is designed to simplify the deployment of AI agents, which are autonomous software entities that perform tasks on behalf of users. As agentic AI becomes more prevalent, enterprises need a secure and manageable environment to deploy, monitor, and update these agents. HPE’s integration with Nvidia’s Agent Toolkit provides a foundation for building and managing AI agents at scale, with built-in governance and security controls.
Zerto and Morpheus Updates for Data Protection and VM Management
HPE announced new capabilities in Zerto Software to identify rogue AI agent actions and enable rewinding to a clean slate using data protection. Private Cloud AI now supports secure local agent registration for approving AI models, skills, and tools under centralized governance. This is critical in the age of generative AI, where agents could potentially make unauthorized changes or access sensitive data. Zerto’s continuous data protection allows IT teams to roll back to a known good state in seconds, minimizing the impact of malicious or erroneous actions.
Meanwhile, HPE Morpheus, a multicloud management platform, now offers a VM Essentials package with a one-year free license for customers migrating from legacy platforms like VMware. HPE also provides Zerto migration licenses during that period and zero-interest financing for cloud ops software over three years to ease the transition. This move is aimed at capturing customers who are dissatisfied with Broadcom’s management of VMware after its acquisition. By offering financial incentives and migration tools, HPE hopes to accelerate adoption of its alternative virtualization and cloud management platform.
The VM Essentials package includes support for virtual machines, containers, and cloud resources, all managed from a single control plane. It is often positioned as a direct alternative to VMware vSphere, with the added benefit of deeper integration with HPE’s hardware and AI infrastructure. The free first-year licenses lower the barrier to entry for organizations considering a switch.
AI Networking Market Context
The announcements come amid explosive growth in AI infrastructure demand. According to Gartner analyst Mike Leibovitz, HPE’s integration strategy leverages the core Mist platform to expand Marvis across the portfolio, innovating while integrating into Aruba networking, data center, and branch environments. Leibovitz noted that agentic NetOps is the most exciting area of innovation in enterprise networking in over 20 years, as organizations move toward hands-off operations. HPE is well positioned with Marvis as the AI engine, but competition from Cisco, Arista, and emerging independent software providers means leadership is still undecided.
HPE’s products are designed for a world where AI workloads require massive bandwidth, low latency, and deterministic performance. The QFX5140 and QFX5252 address the need for high-density 400G and 800G switching, while Mist AI’s predictive analytics and autonomous remediation reduce IT staff burden. The unified SASE platform enables secure access from any edge to AI data centers, and the enhanced Nvidia integration allows enterprises to deploy and manage AI agents with confidence.
With these updates, HPE is positioning itself as a one-stop shop for AI networking and management. The company’s acquisitions of Juniper and Zerto, combined with deep partnerships with Nvidia and AMD, provide a comprehensive stack from silicon to software. As enterprises increasingly adopt agentic AI, HPE’s focus on autonomous network operations and simplified management could give it an edge over rivals. The next few years will likely see intense competition as vendors race to deliver AI-ready infrastructure that scales from the data center to the edge. HPE’s latest barrage of products and integrations marks a significant milestone in that journey, and the market will be watching closely to see how customers respond to the integrated vision.
Source: Network World News