AI Infrastructure

Astera Labs plans custom connectivity solutions supporting NVLink for heterogeneous AI infrastructure with tailored performance and efficiency.

AMD and HPE announced expanded collaboration on the open AMD “Helios” rack-scale AI platform incorporating HPE Juniper Networking switches and Broadcom technology for AI clusters. HPE will offer this architecture globally in 2026. The Herder supercomputer at HLRS Germany will use AMD and HPE systems for HPC and AI workloads.

Analysis reveals overlapping AI data center announcements, power constraints, and timeline gaps distort actual capacity growth projections.

Dell'Oro analysts assess optical circuit switches as a power-efficient solution for connecting large-scale AI GPU clusters.

SONiC’s open-source platform supports scalable AI networking with multi-vendor hardware and high-speed data transfer protocols.

Aviz ONES integrates with NVIDIA Spectrum-X Ethernet to provide AI GPU network orchestration, observability, and lifecycle automation for enterprises.

Analysis shows overlapping AI data center announcements and power limits distort actual growth, emphasizing need for verified capacity data.

Weekly roundup covering a DARPA materials contract, Copilot data security controls, network-as-a-service, and industrial control advisories.

Aviz Networks and NVIDIA discussed AI-driven infrastructure integration at NVIDIA's GTC event, highlighting their collaboration.

AI and HPC workloads demand fast, lossless, scalable GPU networking. NVIDIA Spectrum-X Ethernet delivers this with Spectrum-4 switches, NVIDIA SuperNICs, and accelerated computing platforms. Aviz ONES (Open Networking Enterprise Suite) complements the infrastructure to enhance enterprise deployments.

The post Secure Mediation & Translation: How Itential MCP Enables Controlled AI-Infrastructure Integration appeared first on Itential.

GNE 2025 discusses how Mplify standards are powering the network fabric of the AI era. NaaS is positioned as the core of AI infrastructure.