Hewlett Packard Enterprise (HPE) is betting heavily on artificial intelligence, and according to CEO Antonio Neri, the enterprise landscape is changing fundamentally. AI agents now coexist with human end users in infrastructure, reshaping how workloads traverse networks and what compute and storage systems must deliver.
Speaking at the HPE Discover 2026 conference in Las Vegas, Neri used his opening keynote to detail the company's response across its entire stack. The announcements are broad, covering networking, compute, storage, agentic operations, and cloud management.
Networking: The Foundation for AI Traffic
Neri emphasized that every byte, token, and decision crosses the network. To support that reality, HPE has extended AI connectivity from GPU racks to the inference edge. New QFX switches handle scale-up within a rack and scale-out across GPU clusters. The PTX 12,000 routing platform provides data center interconnect with 800G routing capabilities. For security, the SRX 4700 quantum-safe firewall delivers 1.44 Tbps throughput in a single rack unit. At the edge, the MX 301 router brings the MX platform to inference workloads using Juniper's sixth-generation Trio silicon.
Neri illustrated the cost of latency in large-scale AI training: 'Multiply a small delay across hundreds of thousands of GPUs over weeks of training in your network can mean the difference between training a new model in 90 days or 30 days.'
Compute: Scaling for Agentic Workloads
HPE organizes its compute portfolio into three AI Factory tiers: enterprise, service provider, and sovereign deployments. The new ProLiant DL 394 Gen 12 server is purpose-built for agentic AI and long-context workloads. At the highest tier, new configurations reportedly deliver AI training with one-quarter the GPUs required by the prior Blackwell-generation platform, and inference at one-tenth the cost per million tokens.
Private Cloud AI configurations now scale to 256 GPUs with multi-node inference. A unified gateway provides a single API for access to both frontier and open-source models. Shared cache reduces the cost per first token, making inference more efficient.
Storage: Unified Data for Smarter Agents
The Alletra MPX 10,000 becomes the storage layer 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. According to HPE, this results in 7 to 12 times faster time to value compared to custom-built environments.
Neri noted that agents are only as intelligent as the data used to train them. Traditionally, data required custom preparation for every use case, but the new approach eliminates months of building AI data pipelines.
Agentic Operations: Governance at Scale
AI agents are proliferating across enterprises, often outside formal IT oversight. 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.
Additional agentic operations features include Nvidia Open Shell for isolated policy-enforced agent runtimes, NeMo Cloud for governed workflow blueprints, and Zerto for clean-state rollback when agents make errors. This ensures that despite the autonomy of agents, enterprises maintain control and compliance.
Cloud: Centralized Management
HPE CloudOps consolidates virtualization, data protection, and cloud management into a single hybrid operating layer. The Unleash AI program now covers more than 60 validated partners, providing a broad ecosystem for AI deployment.
The Power Constraint
Neri warned of a critical bottleneck: power. 'Every model, every workload, every agent depends on power, because at its core, an AI factory is doing one thing: turning electrons into tokens.' He noted 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. The future of AI, Neri argued, will be defined not just by compute, but by how efficiently we can power, cool, and connect infrastructure.
Source: Network World News