HPE Discover 2026, the company's main annual conference, ran from 16 to 18 June in Las Vegas. The theme of Antonio Neri's keynote: HPE positions itself first of all as a networking company for the AI era, building compute and storage around the network. With the Juniper integration complete, that direction is consolidated into HPE AI Data Center Solutions. Here is what actually matters for data centre projects.
Networking at the centre of the strategy
The bulk of the announcements covered networking for AI factories, both for connecting accelerators inside a rack (scale-up) and between nodes (scale-out):
- QFS5252 — a scale-up switch on two Broadcom Tomahawk 6 chips with UALink over Ethernet and liquid cooling, for binding GPUs into a single domain.
- QFX5250 — a scale-out switch with 102.5 Tbps of throughput and liquid cooling for horizontal AI cluster scaling.
- QFX5140 — a 16 Tbps inference switch in 1U on the Broadcom Trident 5 for model serving infrastructure.
- PTX12000 — a router with 800G support and coherent ZR/ZR+ optics for data centre interconnect.
- MX301 — a 1.6 Tbps edge router in 1U with 400G ports for inference at the edge.
- SRX4700 — a 1.4 Tbps firewall in 1U with quantum-resistant cryptography.
Servers and the AI platform
- HPE ProLiant Gen12 DL394 with NVIDIA accelerators — a platform for training and inference.
- HPE ProLiant EL2000 — a compact edge server on Intel Xeon 6.
- HPE Private Cloud AI — an updated stack with NVIDIA integration and Alletra Storage MP X10000, offering multi-node inference, a single gateway and a shared KV cache. It comes in three tiers: Private Cloud AI for enterprises, AI Factory at Scale for hyperscale and AI Factory Sovereign for government and defence. Systems built on NVIDIA Vera get confidential computing by default.
Storage
HPE Alletra Storage MP X10000 is storage for AI workloads with file and object access and real-time data enrichment. The logic is simple: next to accelerators you need not only the network but fast access to large volumes of unstructured data.
Management and ecosystem
- HPE Mist and the Marvis virtual assistant now manage HPE Networking CX switches through natural-language actions.
- HPE CloudOps extends Morpheus to legacy VMware environments.
- HPE Unleash AI is a partner programme around the AI ecosystem.
What it means for customers in Uzbekistan
The practical conclusion from Discover 2026: in AI infrastructure the network stops being plumbing and becomes the defining element — it drives both the cost and the achievable performance of the cluster. Even outside large AI factories the trends apply to ordinary data centres: the growing role of 400 and 800G, liquid cooling arriving in dense configurations, demand for fast storage for unstructured data and the move to quantum-resistant perimeter security.
Most of the announced platforms are not immediately orderable, but the current HPE ProLiant Gen11, Aruba CX and Alletra lines already cover these tasks and remain current. We will size a configuration for a specific workload and calculate the power and cooling infrastructure: info@itsmart.uz, Telegram or the contacts page.
Source and photographs: ServeTheHome, Patrick Kennedy, 16 June 2026 — servethehome.com. Published with attribution.
Related guides and tools: HPE ProLiant Gen11.
