# HPE Discover 2026 Keynote Coverage > Source: > Published: 2026-06-16 18:55:18+00:00 I am in Las Vegas today for HPE Discover 2026. There are going to be a ton of announcements, so we are just going to try covering the highlights. This keynote comes as [Juniper] becomes more fully integrated into the organization, so we expect AI and networking to be major focuses. Yesterday, I was sitting in the HPE industry analyst track with the company’s executives and some customers. As an example, it was interesting to hear how HPE was helping the Dallas Cowboys get jump-started on AI. It reminded me of the [Inside the Secret Data Center at the Heart of the San Francisco 49ers Levi’s Stadium](https://www.servethehome.com/inside-the-secret-data-center-at-the-heart-of-the-san-francisco-49ers-levis-stadium-intel/) tour we did. I am typing this live, so please excuse typos. HPE’s keynotes are like a stream of announcements every few sentences, so this is rough. Also, as a fun bit, I got a great seat dead center to Antonio, but the announcements are happening on the side screens. Oops. ## HPE Discover 2026 Keynote Very telling in this keynote, the first feature is around the HPE self-driving network. While the rest of the industry seems focused on compute, HPE kicked off with networking. Next, HPE is giving a high-level overview of some of the other topics it will discuss today. Those include: - HPE ProLiant Gen12 HPE ProLiant Compute DL394 Gen12 powered by NVIDIA is being featured. - Recently announced HPE ProLiant Compute EL2000 (a neat Intel Xeon 6 edge box) - HPE Nonstop Updates with Intel Xeon 6 - HPE Alletra Storage MP X10000 for AI native file and object storage on - HPE Alletra B10000 block storage - As virtualization costs increased, HPE CloudOps is doing more to bring Morpheus to a broader ecosystem and cover more of the legacy VMware footprint - Cyber resilience with the HPE Silicon Root of Trust - Security-first networking - HPE GreenLake cloud and GreenLake intelligence. HPE is using agentic AI to help optimize and run operations, like troubleshooting issues in an environment. On a personal note, I firmly believe this is where infrastructure is going which you may have seen in our [BIG AI Cluster Little Power the 8x NVIDIA GB10 Cluster](https://www.servethehome.com/big-cluster-little-power-the-8x-nvidia-gb10-cluster-marvell-cisco-ubiquiti-qnap-arm/) - HPE Financial Services HPE is doing a customer cameo with Vultr. HPE is working with Vultr on the next-generation AI build-out and HPE Networking. On the topic of networking, that appears to be the first area that HPE is going to deep-dive into. ## HPE Networking Updates The first announcement is that HPE Juniper Networking is now part of HPE AI Data Center Solutions. HPE is now talking about its AMD Helios AI rack solution. The HPE Juniper Networking QFS5252 is a scale-up switch for AMD Helios. It supports UALink over Ethernet (UALoE), SONiC with AI-Native Operations, and it is liquid-cooled. They did not say it in the keynote, but this is a dual Broadcom Tomahawk 6 switch. HPE Juniper Networking QFX5250 for scale-out networking as a liquid-cooled 102.5Tbps switch. HPE Juniper PTX routing family is designed to route between data centers. It supports 800Gbps routing and ZR/ZR+ coherent optics. If you want to learn more about 800G ZR+ optics, we did a detailed piece: [Going 800Gbps at up to 1000km with the Marvell COLORZ 800](https://www.servethehome.com/going-800gbps-at-up-to-1000km-with-the-marvell-colorz-iii-800g-zr-osfp/). The HPE Networking SRX4700 is a quantum-safe 1.4Tbps firewall in 1U. This is really neat. I think this is a firewall we could scale to test using our CyPerf test platform since that is now running at over 2Tbps. HPE is also talking about its HPE AI Grid with NVIDIA. This uses NVIDIA Spectrum-X, Bluefield-3, and Connect-X, with HPE Juniper Networking and ProLiant compute. Now up is the HPE Juniper Networking MX301 router for the Inference Edge. This is a 1.6Tbps in 1U full duplex 400G connectivity, and AIOps management. This router is based on HPE’s 6th-generation Trio silicon. HPE Juniper Networking QFS5140 Inference Switch 16Tbps in 1RU optimized for Inference. 16Tbps is just an interesting capacity for a switch, since we usually see 12.8T and 25.6T switches as the popular points around there. HPE Networking CX Switches managed by HPE Mist. This is a big one because the customer question will be, “Is this part of a larger integration plan?” HPE Marvis actions in HPE Aruba Central, bringing conversational AI to networking. ## HPE Agentic Enterprise Managing a shadow workforce is challenging because it requires AI agent governance, data, and scale. You can now register agents and provide security. HPE is supporting NVIDIA OpenShell and NemoClaw. HPE is discussing the HPE Private Cloud AI in terms of building AI-ready data pipelines. This is an integration with the NVIDIA AI platform and using the HPE Alletra Storage MP X10000 for the foundation. The X10000 has real-time data enrichment and native MCP integration. Someone is standing in the way, so I cannot get good photos of this. There are new inference capabilities with multi-node inference and a unified gateway. HPE is also discussing shared KV cache capabilities. Interestingly, I did not hear Antonio say NVIDIA CMX here, but that must be part of the solution, no? There is also an HPE Unleash AI partner ecosystem for helping companies find validated solutions. HPE also has different AI solutions for different classes of customers - HPE Private Cloud AI – For enterprises - HPE AI Factory at Scale – Neoclouds and hyper-scale - HPE AI Factory Sovereign – For governments, defense, and similar segments Now HPE is talking about NVIDIA Vera-based systems. HPE is making confidential computing now standard across its HPE AI Factory with the NVIDIA portfolio. Next, we are talking about power. HPE is now doing customer vignettes with Siemens Energy. While I thought we were going to talk about something like HPE helping customers source power, this is more about how Siemens Energy is using HPE for predictive failure. ## Final Words That was an absolutely crazy keynote, with how much was covered in around an hour. Perhaps the bigger takeaway is the order and messaging. HPE absolutely hammered networking announcements. At [Dell Tech World 2026 It’s All About Sovereign and On-Premises AI](https://www.servethehome.com/dell-tech-world-2026-its-all-about-sovereign-and-on-premises-ai/) we got a completely different viewpoint that was focused more around business transformation being powered by compute. Also of note, Jensen was on stage at Dell Tech World, but was not on stage at HPE Discover this year. Clearly, NVIDIA is a significant HPE partner, but this felt like a networking company that is doing some AI compute, whereas Dell’s keynote felt more like a compute company that sells networking gear alongside it. I am getting kicked out of the keynote room now, so off to the next talk.