πŸ“Œ Topic Cluster: Infrastructure & Compute (INFRA). Subscribe to daily Ai Brief β†’

Infrastructure & Compute (INFRA) Intelligence Hub

Real-time accumulated intelligence, primary research, and executive digests for Infrastructure & Compute (INFRA).

πŸ“… 2026-08-04

SanDisk and SK hynix publish HBF spec

SanDisk and SK hynix have jointly published the HBF (Host-to-Buffer) specification, describing a new memory expansion architecture that can provide GPUs with several terabytes of additional NAND-based capacity using stacked flash modules. The spec details support for up to 16-high NAND stacks and around 3 TB/s bandwidth per device, relying on UCIe interconnect technology. It is designed to offload infrequently accessed data from costly HBM or GDDR memory in high-performance accelerators. The initiative could reduce total system costs for GPU clusters running AI and data analytics workloads, while pushing broader industry adoption of UCIe-based chiplet ecosystems. Vendors are expected to prototype compatible controllers, firmware, and accelerator designs next.

Read Original Source β†—
πŸ“… 2026-08-04

Nvidia open sources cuFile API

Nvidia has open-sourced its cuFile API, a key part of its GPUDirect Storage stack, to accelerate direct data movement between high-speed storage such as NVMe drives and GPU memory with millisecond-level latency. Originally launched in 2021 alongside CUDA Toolkit 11.4, cuFile bypasses CPUs and main memory using direct memory access, reducing bottlenecks and GPU starvation for large-scale AI workloads, including retrieval-augmented generation and agentic AI systems. Nvidia also unveiled the Storage-Next initiative with 40 storage and flash vendors and introduced its SCADA architecture, which seeks to balance ultra-fast GPU data access with Linux-based protections against clobber and potential security gaps.

Read Original Source β†—
πŸ“… 2026-08-04

Anthropic signs $10bn Volta compute deal

Anthropic has signed a $10 billion, six-year compute deal with Volta, a week-old cloud startup. The agreement makes Volta a major supplier of AI infrastructure for running Anthropic’s Claude models. The contract joins a sprawling network of Anthropic compute partnerships with Amazon, Google, Broadcom, AMD, CoreWeave, Akamai, SpaceX, and potential talks with Meta. Volta, newly valued at $2.4 billion, aims to broaden access to Nvidia chips. Volta gains crucial validation, financing, and a long-term anchor customer, while Anthropic deepens its strategy of diversifying compute sources for leverage and resilience. The massive commitments highlight ongoing concentration risks and bets on sustained enterprise AI demand.

Read Original Source β†—
πŸ“… 2026-08-03

TSMC outlines Co Packaged Optics foundry roadmap

TSMC has detailed its co-packaged optics foundry roadmap, outlining how it plans to integrate optical interconnects directly with advanced chips. The article also compares approaches from Intel, Samsung, and GlobalFoundries. Co-packaged optics promise higher bandwidth and lower power consumption by placing optical engines close to switching silicon. Leading foundries are racing to define manufacturable platforms suitable for hyperscale data centers and AI infrastructure. These roadmaps could reshape future networking architectures, influencing switch designs, data center layouts, and vendor partnerships. Customers will watch maturity, cost, standardization efforts, and ecosystem support before committing to large deployments.

Read Original Source β†—