HPE Expands Oracle Networking Deal for Gigawatt-Scale AI Infrastructure

HPE and Oracle have expanded their collaboration on networking infrastructure for AI data centers. The companies announced on September 2, 2026 that HPE Juniper Networking equipment will be deployed across Oracle Cloud Infrastructure's global footprint under a multi-year agreement, supporting the buildout of computing clusters at gigawatt scale.

What the deal covers

The expanded agreement covers the deployment of four HPE Juniper networking product lines: the PTX and MX routing platforms and the QFX and EX switching platforms. The latest QFX switches are designed to deliver high connection density, dynamic load balancing and advanced congestion management, capabilities required by large AI backend networks built on the RoCEv2 standard, RDMA over Converged Ethernet.

The partnership also includes multi-year network support services and deployment financing mechanisms. The companies said they will jointly develop intelligent telemetry tools designed to give Oracle deeper visibility into network behavior and earlier detection of growing queues, packet loss or component degradation before they affect GPU utilization and network availability.

A decade of joint engineering

HPE says the expanded agreement builds on more than a decade of joint engineering work between Oracle and Juniper Networks, the company HPE acquired. That prior collaboration is what allowed the current deployment to scale quickly to meet the needs of Oracle Cloud Infrastructure, which has been rapidly expanding its compute capacity in recent quarters for training and running large language models.

Oracle is building AI infrastructure at extraordinary scale, and HPE is uniquely positioned to support OCI with one of the industry's most comprehensive AI data center networking portfolios - Rami Rahim, Executive Vice President and General Manager of Networking, HPE

Building AI infrastructure at Oracle's scale requires an integrated ecosystem where every layer of the technology stack works together - Mahesh Thiagarajan, Executive Vice President, Oracle Cloud Infrastructure

Financial results in the background

The announcement coincided with HPE's third-quarter fiscal results, in which the company reported record sales driven by demand for AI infrastructure. Networking segment revenue grew 75 percent to $2.9 billion, data center networking revenue more than doubled, and routing revenue climbed 270 percent. On that basis, HPE raised its full fiscal year 2026 revenue growth forecast to a range of 34-37 percent.

On Oracle's side, the scale of investment is even more striking. The company plans capital expenditures of $55.7 billion in the current fiscal year, up from a previously projected $21.2 billion, and has already announced roughly $70 billion in net capital expenditures for fiscal year 2027. Much of that spending is going toward expanding data centers that support AI-related cloud contracts, including with OpenAI.

What it means for infrastructure markets

Praveen Jain, Senior Vice President and General Manager of Data Center and AI Networking at HPE, highlighted the importance of latency in this type of installation. Every microsecond matters for AI models and agents, and high-performance networking combined with deeper infrastructure visibility is meant to reduce downtime for costly GPU clusters.

The deal has no direct bearing on the Polish market, but it illustrates the scale of spending that global cloud providers are now directing toward the networking layer of AI data centers, an area often overlooked in discussions focused solely on GPUs. Rising orders for networking equipment also indirectly ripple through the global component supply chain, including demand for copper and fiber optics used in server-to-server connections.

The transaction also has a financial component: HPE issued Oracle warrants entitling it to purchase HPE common stock, tying the two companies' interests together beyond a standard vendor-customer relationship. Arrangements like this are becoming more common in the AI infrastructure industry, as hardware vendors and cloud operators look for ways to share the risk of enormous capital expenditures.