HPE Discover Las Vegas 2026 delivered real product launches, not just roadmaps: new switching silicon, Morpheus 9, AI agent governance, and a direct challenge to Broadcom's VMware base.
Hewlett Packard Enterprise just wrapped up its biggest event of the year with a message that cuts against most enterprise AI marketing: before you worry about which model to run, fix your network. At HPE Discover Las Vegas 2026, held June 15-18 at The Venetian Convention and Expo Center, CEO Antonio Neri turned what could have been a routine product refresh into a declaration that the $14 billion Juniper Networks acquisition is now paying off in concrete, shippable products.
For enterprise IT leaders tired of abstract AI roadmaps, this one matters. Real products shipped, real timelines were named, and a genuine commercial fight with Broadcom over the VMware refugee market was declared.
The Juniper Payoff: Self-Driving Networks and New Switching Silicon
Neri told attendees: "There is always one core element of your infrastructure that touches everything. That core element is the network." That framing drove every hardware announcement.
Key launches included Marvis AI extending to Aruba Central, new QFX data center switches supporting the UALink open standard, and GreenLake Intelligence Mesh for governing AI agents. HPE is aligning HPE Aruba Central and HPE Mist AI platforms with shared agentic capabilities, common hardware, and consistent AI-native operations.
On the switching side, two new products stand out:
- HPE Juniper Networking QFX5252: A scale-up switch for AMD Helios that supports UALink over Ethernet, runs SONiC with AI-native operations, and is fully liquid-cooled.
- HPE Juniper Networking QFX5140: Built for inference clusters, delivering 16 Tbps total capacity in 1RU, optimized for AI-native operation with HPE Marvis.
The use of UALink, an open industry standard, is a deliberate alternative to NVIDIA's proprietary NVLink. It gives buyers more flexibility in how they wire AI factories and reduces lock-in to a single GPU vendor's interconnect ecosystem.
HPE also introduced a unified AI-native SASE platform built on HPE Networking EdgeConnect, converging SD-WAN and cloud-delivered security in one management console. It includes integrated Security Service Edge, embedded zero trust network access, and SASE copilot capabilities.
Morpheus 9 and the VMware Migration Play
The announcement with the most immediate commercial teeth was Morpheus 9. Morpheus VM Essentials now claims over 2,000 customers and more than a million installed cores, backed by a claimed 90 percent cost reduction through socket-based pricing.
Morpheus 9 adds:
- Morpheus Central on GreenLake for a single operational pane across all installations
- Software-defined networking based on Juniper technology with built-in micro-segmentation
- Stretched clustering across sites with automatic failover
- Zerto and Veeam integration
- An MCP server for agent-driven operations
The commercial incentive to move is aggressive. New VM Essentials customers can receive up to one free year of VM Essentials licenses, a year of HPE Zerto for one dollar to support non-disruptive migration, and 0% interest on software through HPE Financial Services. For any organization still sorting out its post-Broadcom VMware costs, this is a direct invitation to switch. Morpheus 9 makes HPE the most aggressive VMware alternative on the show floor, a battle Cisco and Extreme are not directly fighting.
GreenLake Intelligence, AI Agent Governance, and the ServiceNow Partnership
HPE is expanding GreenLake Intelligence, an initiative that embeds generative AI and AI agents into infrastructure operations. The platform analyzes telemetry across networking, cloud, storage, and compute to identify issues, recommend actions, and automate tasks. It is being integrated into Aruba Central and other GreenLake services for network operations, VM migrations, capacity planning, and troubleshooting.
A new ServiceNow partnership connects GreenLake Intelligence telemetry to ServiceNow's autonomous AI workforce, creating a pipeline from full-stack infrastructure observability to autonomous service delivery. The rollout is planned through 2026 and 2027.
HPE also addressed a real and growing operational hazard: agentic AI. An agent that makes repeated unauthorized API calls, modifies infrastructure it was not cleared to touch, or consumes token budget exponentially is a failure mode traditional enterprise IT was never designed to contain. HPE's response includes updates to its Private Cloud AI platform adding agent registration, identity controls, and policy enforcement, plus integrations with NVIDIA software to help enterprises manage AI systems operating across corporate data and applications.
New OpsRamp copilots let IT teams monitor which models are consuming token budget, at what rate, and at what cost. That capability became operationally critical after HPE's own engineers experienced runaway token consumption as their systems processed billions of signals from customers.
What Is Shipping Now vs. What Is Coming
Enterprise buyers should read the delivery schedule carefully.
Available now or rolling out through Q2 and Q3 2026:
- OpsRamp Operations Copilot for GreenLake Intelligence
- GreenLake Marketplace with direct customer-to-ISV transactability
- HPE CloudOps Software for cloud service providers
- Aruba CX switches managed via HPE Mist
- HPE Morpheus software updates
Scheduled but not yet available:
- NVIDIA Confidential Computing as a default across the AI Factory, targeted for Q4 2026
- HPE Private Cloud AI air-gapped configuration for classified and sovereign deployments, expected general availability in Q3 2026
- Consolidated HPE Partner Ready Vantage program unifying former Juniper and Aruba partner networks, effective November 1, 2026
What IT Teams Should Do Now
- If you are still on VMware: Get a comparative cost analysis covering Morpheus 9 and the zero-cost first-year migration offer. The economics are moving fast and the window is not permanent.
- If you are deploying AI agents in production: Put agent identity controls and token budget monitoring on your immediate checklist. Neri stated plainly: "IT will be responsible for thousands of agents that are part of your enterprise workforce." That is not an exaggeration.
- If you are buying data center switching: Evaluate the QFX5140 and QFX5252 against your current refresh plans. UALink support matters if you want to avoid GPU vendor interconnect lock-in.
- If you are on Aruba Central: The Marvis AI actions integration is rolling out now. Prioritize the upgrade to get automated network remediation without a separate management platform.
Neri argued that networking remains one of the industry's most important unsolved challenges, stating: "The networking layer of the stack is going to be the next opportunity." Based on what shipped in Las Vegas this week, that argument has real product behind it for the first time.
How 247techify Can Help
At 247techify, we help businesses assess, plan, and execute infrastructure decisions exactly like these, from VMware migration cost analysis and HPE GreenLake deployments to AI agent governance frameworks and modern switching refreshes. If the announcements from HPE Discover 2026 raised questions about your own IT roadmap, our team is ready to work through the specifics with you. Get in touch at https://www.247techify.com/ and let's talk.
