We explore the latest data center hardware and infrastructure developments from the last month, spanning networking, memory, CPUs, orchestration software, and chip design.
Data Center Knowledge’s most-read hardware coverage from June 2026 points to a new phase in AI infrastructure. Beyond GPUs, vendors are investing across the stack – networking, memory, CPUs, orchestration software, and chip design – to improve utilization and remove bottlenecks. The month’s biggest announcements also show intensifying competition as established players and newcomers vie to power the next generation of AI data centers.
Here are the most-read data center hardware stories from last month:
HPE at Discover 2026: Networking, Utilization, and Hybrid Quantum
Senior News Writer Shane Snider reported on Hewlett Packard Enterprise’s (HPE) comprehensive AI infrastructure strategy at Discover 2026, including:
Collectively, HPE’s moves target one of AI’s toughest challenges: cutting network-induced latency so that ever-larger GPU clusters stay busy and efficient.
AI Data Centers Squeeze Memory Supply, Coalition Warns A cross-sector coalition of trade associations advised policymakers that surging AI buildouts are straining global memory supply. Memory is quickly becoming a potential chokepoint alongside GPUs and power as deployments scale.
[Nvidia Overtakes Rivals in Data Center Ethernet Switching, IDC Says](/infrastructure/nvidia-overtakes-rivals-in-data-center-ethernet-switching-idc-says)
New IDC data shows Nvidia surpassing competitors in the data center Ethernet switching market, extending its influence beyond AI accelerators. The milestone aligns with Nvidia’s end-to-end AI infrastructure strategy spanning compute, networking, and software.
[Nvidia Says Vera Rubin, Vera CPU on Track, Launches DSX OS to Run AI Factories](/data-center-chips/nvidia-says-vera-rubin-vera-cpu-on-track-launches-dsx-os-to-run-ai-factories)
Chips and hardware writer Wylie Wong examined Nvidia’s roadmap update, highlighting progress on the Vera Rubin platform, the Vera CPU, and the debut of the DSX OS for managing AI factories. Together, these moves reinforce Nvidia’s push to control not only AI chips but also the orchestration layer for entire deployments.
IBM Pushes AI Chip Design Forward with Sub-1-nm NanoStack IBM offered an early look at NanoStack, a research initiative exploring sub-1-nm chip architectures for AI. While experimental, the effort reflects the search for novel device- and system-level approaches as conventional transistor scaling confronts physical limits.
AWS Launches Graviton5-Powered EC2 Instances for AI and HPC AWS rolled out new Graviton5-powered EC2 instances, underscoring that CPUs remain indispensable in modern AI stacks. The launch reflects growing recognition that top performance depends on balancing GPUs with complementary compute.
AWS Graviton5 chip. (Image: AWS)
[QumulusAI’s $124M Deal Spotlights AI Infrastructure’s Utilization Challenge](/business/qumulusai-s-124m-deal-highlights-ai-infrastructure-s-next-challenge-utilization)
As AI infrastructure proliferates, simply adding hardware no longer guarantees returns. QumulusAI’s $124 million deal highlights the priority of keeping expensive clusters highly utilized and economically efficient once deployed.
Qualcomm Lands Meta CPU Deal, Unveils AI Data Center Platform Qualcomm’s partnership with Meta and its new AI data center platform mark a significant push into hyperscale infrastructure. The announcement signals broadening competition as more chipmakers seek roles beyond traditional server CPUs.
The Big Picture #
June’s most-read hardware stories depict an AI ecosystem that’s increasingly systems-focused. The race is no longer only about GPUs. Networking, memory, orchestration software, and end-to-end efficiency are emerging as the decisive technologies shaping the next generation of AI data centers.