The Network as the Foundation

At HPE Discover 2026 in Las Vegas, CEO Antonio Neri laid out a comprehensive strategy for an AI architecture designed specifically for the age of autonomous agents. His keynote centered on the idea that every component of the data center—from networking to compute to storage—must be rethought to support workloads driven not just by humans but by AI agents operating alongside them. Neri emphasized that the network is the most critical element, stating that every token, every decision, and every interaction depends on it. The expansion and deep integration of Juniper Networks into the HPE portfolio was a recurring theme. New QFX switches target GPU rack scale-up and cluster scale-out, while the PTX 12,000 routing platform delivers 800 Gbps data center interconnect. The SRX 4700 firewall provides quantum-safe protection at 1.44 Tbps in a single rack unit, and the MX 301 edge router brings the MX platform to inference endpoints using sixth-generation Trio silicon. Neri highlighted the cost of latency: a small delay multiplied across hundreds of thousands of GPUs over weeks of training can mean the difference between training a model in 90 days or 30 days.

Compute for the Agentic Era

While networking connects systems, the compute layer must be optimized for AI. HPE organized its server portfolio into three AI Factory tiers: enterprise, service provider, and sovereign deployments. The new ProLiant DL 394 Gen 12 is purpose-built for agentic AI and long-context workloads. Neri claimed that at the AI Factory at Scale tier, new configurations require one-quarter the number of GPUs for training compared to the prior Blackwell-generation platform, and inference costs drop to one-tenth per million tokens. Private Cloud AI configurations now scale to 256 GPUs with multi-node inference, allowing capacity to grow with demand. A unified gateway offers a single API for accessing both frontier and open-source models, while shared cache reduces first-token latency. This shift is critical as enterprises move from experimentation to production, where cost and speed become paramount.

Storage: Data Readiness for AI

Agents are only as smart as the data they access. The Alletra MPX 10,000 becomes the storage backbone for Private Cloud AI, unifying file and object storage on a single architecture. It adds real-time metadata enrichment and native MCP support, enabling agents to retrieve data across structured and unstructured sources seamlessly. Neri noted that traditionally, building AI data pipelines required months of custom preparation, but the new approach delivers 7 to 12 times faster time to value. The platform carries Nvidia Certified Storage validation, ensuring tight integration with GPU-accelerated workloads. This unified storage layer addresses a key pain point: data fragmentation that cripples agent effectiveness. By providing a single, intelligent data plane, HPE aims to reduce the complexity enterprises face when deploying AI at scale.

Governance for AI Agents

Running on top of networking, compute, and storage are AI agents, which Neri described as reasoning across data, applications, and workflows to automate processes and take action on behalf of users. However, agents are often created by developers outside formal IT oversight, creating governance and scale challenges. HPE’s answer is a governed agent layer built into Private Cloud AI. Enterprises can register agents built in any framework, applying security controls on API calls, identity, and encryption with zero code changes. A three-tier identity model verifies the user, governs the agent, and requires human approval for sensitive actions. This is backed by Nvidia Open Shell for isolated policy-enforced runtimes, NeMo Cloud for governed workflow blueprints, and Zerto for clean-state rollback when agents make errors. The goal is to balance innovation with safety, allowing enterprises to deploy agents without compromising security or compliance.

The Power Challenge

Neri also addressed the elephant in the room: power. He warned that the U.S. faces a 19 gigawatt power gap by 2028, with data centers projected to account for nearly half of U.S. electricity demand through 2031. He framed the AI factory as fundamentally a machine that turns electrons into tokens, and the future will be defined not just by compute but by how efficiently infrastructure can be powered, cooled, and connected. HPE’s announcements at Discover 2026 represent a cohesive effort to build the physical and logical fabric needed for the agentic era, addressing everything from foundational networking to governance and energy constraints. The company’s integration of Juniper assets is a central pillar of this strategy, as is its partnership with Nvidia on agentic governance and storage validation. With the Unleash AI program now covering more than 60 validated partners, HPE is positioning itself as a one-stop partner for enterprises navigating the shift from human-driven workloads to a world where AI agents run alongside end users, requiring new architectures and new ways of thinking about data center operations.

Source: Network World News