HPE has expanded its AI-native networking portfolio and outlined a new strategy for self-driving networks, marking the company’s first major integration milestone following its acquisition of Juniper Networks. Announced at HPE Discover Barcelona 2025, the updates introduce unified AIOps capabilities, new high-performance switches and routers, and deeper interoperability across HPE’s compute, storage, networking and cloud platforms.
The company said the combined HPE Aruba Networking and HPE Juniper Networking portfolios now share a consistent AIOps feature set, supported by a common agentic-AI and microservices framework. This approach aims to deliver autonomous network operations and simplify management across hybrid environments, while enabling AI workloads to run at greater scale and performance.
Rami Rahim, HPE’s executive vice president and general manager for networking, said modern networks must be designed specifically for AI-driven environments where device density, traffic complexity and security risks continue to rise. He said the portfolio expansion positions HPE to deliver high-performing, self-driving networks that adapt automatically to operational needs.
Unified AIOps across Aruba Central and Juniper Mist
HPE has begun merging capabilities from Aruba Networking Central and Juniper Networking Mist Operations platforms. New features include:
- Juniper Mist’s Large Experience Model (LEM) — trained on billions of application telemetry points — now available within Aruba Central.
- Aruba’s Agentic Mesh anomaly-detection and reasoning engine now available within Mist.
- Shared NOC-level organisational insights across both platforms.
- New WiFi-7 access points operable across both management systems.
Aruba Central On-Premises 3.0 also gains additional AIOps functions, ranging from generative-AI-driven remediation to intelligent client insights and simplified configuration workflows.
New switches and routers for AI-scale infrastructure
Addressing the growing demand for high-bandwidth networking to support AI training and inference, HPE introduced:
- HPE Juniper Networking QFX5250 — the first OEM switch built on Broadcom’s Tomahawk 6 silicon, delivering 102.4 Tbps bandwidth.
- HPE Juniper Networking MX301 edge router — a 1RU, 1.6 Tbps multiservice router designed for AI inference, metro routing, and mobile backhaul.
Both products target high-density GPU environments and distributed AI workloads where low-latency transport and energy efficiency are central requirements.
Expanded partnerships with NVIDIA and AMD
HPE also strengthened its collaboration with NVIDIA and AMD, introducing new networking solutions for AI factories and long-haul data-centre interconnects. Highlights include:
- NVIDIA Spectrum-X–based Ethernet networking for AI clusters.
- Integration of Juniper MX and PTX routing platforms for large-scale DCI and edge-to-core connectivity.
- Support for AMD’s Helios rack-scale AI architecture, featuring a purpose-built HPE Juniper Networking scale-up switch capable of delivering 260 TB/s scale-up bandwidth.
AIOps for hybrid operations
HPE is also expanding its hybrid AIOps capabilities across GreenLake and HPE OpsRamp Software, aiming to provide full-stack observability spanning compute, storage, networking and cloud. Enhancements include:
- Integration of Juniper Apstra for data-centre assurance and predictive analytics.
- New Compute Ops Management features, including self-service root-cause analysis.
- Agentic Root Causing and Model Context Protocol (MCP) support to connect third-party AI agents without custom code.
Financing for AI-native network modernisation
HPE Financial Services has introduced zero-per-cent financing for AIOps networking software, including Juniper Mist, and a leasing program offering the equivalent of a 10 per cent savings for customers modernising AI-capable network infrastructure.
The first of the new hardware, the QFX5250 switch, will be available in Q1 2026, with additional products and integrations to roll out through 2026.
