Research Note: HPE’s Updated AI Factory

At its recent Discover event, HPE announced an expansion of its NVIDIA-based AI Computing portfolio with three distinct AI factory configurations targeting enterprise, service provider, and sovereign deployment scenarios.The offerings center on the upgraded HPE Private Cloud AI platform, which integrates NVIDIA Blackwell GPUs with HPE ProLiant Gen12 servers, custom storage solutions, and orchestration software.

Research Note: AMD Raises its Game at its Advancing AI 2025 Event

AMD Instinct MI350

AMD announced a comprehensive portfolio of AI infrastructure solutions at its recent Advancing AI 2025 event, positioning itself as a full-stack competitor to NVIDIA.

The announcements include the immediate availability of MI350 Series GPUs with 4x generational performance improvements, the ROCm 7.0 software platform achieving 3.5x gains in inference, and the AMD Developer Cloud for broader ecosystem access.

AMD also previewed its 2026 “Helios” rack solution, which integrates MI400 GPUs, EPYC “Venice” CPUs, and Pensando “Vulcano” NICs.

Research Note: NVIDIA NIM Agent Blueprints

NVIDIA

NVIDIA launched its new NIM Agent Blueprints, a catalog of pre-trained, customizable AI workflows to help enterprise developers quickly build and deploy generative AI applications for critical use cases, such as customer service, drug discovery, and data extraction from PDFs.

Research Note: IBM Telum II & Spyre AI Accelerators

IBM Telum II & Spyre

At the Hot Chips 2024 conference in Palo Alto, California, IBM unveiled the next generationj of enterprise AI solutions: the IBM Telum II processor and the IBM Spyre Accelerator. These new technologies should meet the demands of the AI era, providing enhanced performance, scalability, and AI capabilities. Both are expected to be available in 2025.

Research Note: Inside the IBM Research NorthPole Accelerator

IBM Research

IBM Research has developed and released details on a groundbreaking AI chip called NorthPole, which could revolutionize AI hardware systems. Unlike traditional computer chips, NorthPole integrates processing units and memory on the same chip, eliminating the von Neumann bottleneck and significantly improving efficiency.