Sandisk and SK hynix join forces to standardize High Bandwidth Flash memory, a NAND-based alternative to HBM for AI GPUs — Move could enable 8-16x higher capacity compared to DRAM

Phison CEO claims NAND shortage could last a staggering 10 years — says memory 'supercycle' imminent and 'severe' 2026 shortages are at hand

Phison Electronics Corporation's CEO, Pua Khein-Seng, predicts a NAND flash shortage starting in 2026 that could last for a decade, surpassing other industry estimates. Pua suggests a "supercycle" in memory demand is imminent, driven by the AI data center boom. As flash demand surges due to increased storage needs for AI models, companies like SanDisk, Micron, and Western Digital are freezing or raising prices. Pua anticipates SSDs will dominate over HDDs in data centers within the next decade, with flash memory remaining strong for years to come. Morgan Stanley also foresees a "memory supercycle" ahead, influenced by HBM demand and potential market shifts.

Nvidia tipped to be TSMC's first 16A customer, ahead of Apple — Feynman GPUs could make full use of GAA transistors and backside power

Nvidia is set to be TSMC's first customer for the A16 node, utilizing GAA transistors and backside power delivery in its upcoming Feynman GPUs, designed for AI applications. This move breaks Apple's tradition of being the first to adopt new TSMC process nodes. A16's backside power delivery allows for improved performance and architectural freedom, potentially giving Nvidia a competitive edge in AI and HPC markets. By skipping the N2 process and going straight to A16, Nvidia aims to lead in advanced silicon, though at a higher cost. This strategic shift positions Nvidia as a frontrunner in AI chip innovation and could influence the industry's future roadmap decisions.

GamersNexus YouTube channel faces deletion following NVIDIA AI GPU black market documentary

Kioxia’s new 5TB, 64 GB/s flash module puts NAND toward the memory bus for AI GPUs — HBF prototype adopts familiar SSD form factor