As memory continues to dominate the AI hardware conversation, a research note from Korea Investment & Securities (KIS) outlines that prices might not drop. Since memory is directly tied to the productivity of an AI GPU, hyperscalers have placed long-term orders for capacity. KIS believes that even if this time period elapses, the impact of the shortage on memory prices might persist due to the chips' ability to drive key metrics of AI GPU performance. Memory Chips' Ability To Drive Up System Level Performance Could Keep Prices High, Says Research Firm The research note explains that since expanding memory capacity […]

Read full article at https://wccftech.com/memory-prices-wont-drop-even-as-shortage-eases-korean-research-firm-warns-hyperscalers-locked-in-long-term/

Korean semiconductor giant Samsung has achieved 80% production yield with its 8-nanometer chip manufacturing process, according to a media report from the Seoul Economic Daily. Samsung's yields are a frequent feature of industry discussion, with multiple media reports highlighting the firm's struggles with production efficiency. A process technology's yield is a crucial metric of its commercialization, and lower yields often lead to foundries having to bear the cost of defective products which their customers are unable to use or sell. NVIDIA Backed Groq Relying On Samsung For Its Chip Production Requirements According to the details, NVIDIA backed Groq has ordered […]

Read full article at https://wccftech.com/report-samsung-hits-80-yield-on-4nm-process-as-nvidia-backed-groq-ibm-and-baidu-pile-onto-its-foundry/

Korean memory manufacturer SK hynix has announced a yield improvement for its hybrid bonding packaging technology for high bandwidth memory (HBM) modules. Hybrid bonding enables memory chip manufacturers to bond memory layers with each other without relying on bumps. The direct contact enables higher speeds and improved efficiency through lower heat generation. SK Hynix's technical leader, Kim Jong-hoon, revealed the development at the Beyond HBM — Core Technologies of Advanced Packaging: From Next-Generation Substrates to Modules conference in South Korea, reports The Elec. Latest Packaging Technology To Come Into Play With Next-Generation HBM 4 Memory Chips High bandwidth memory is […]

Read full article at https://wccftech.com/sk-hynix-verifies-12-die-hybrid-bonded-hbm-stack-but-wont-disclose-yield-figures-as-next-gen-hbm4-ai-memory-race-heats-up/

Google and NVIDIA have teamed up to provide users with access to as much as one million NVIDIA GPUs to power up the freshly launched A5X instances. The announcement is part of the pair's latest collaboration to reduce inference costs and improve token throughput. Their A5X system relies on NVIDIA's network accelerators that enable the development of single and mutli-cluster computing infrastructure for AI workloads. The A5X Instance: Purpose-Built for Agentic AI Workloads The A5X instances are Google's latest products that are designed specifically to run agentic artificial intelligence workloads. They are part of Google's AI Hypercomputer portfolio which also […]

Read full article at https://wccftech.com/nvidias-rubin-lands-inside-googles-virtual-machine-stretching-multi-site-clusters-to-nearly-1-million-gpus/