List of Flash News about NVIDIA GB200
| Time | Details |
|---|---|
|
2025-10-28 17:07 |
NVDA: Jensen Huang Says H200 Is No. 2 Behind GB200 AI Chip — What It Means for Traders and AI-Crypto Plays
According to @StockMKTNewz, Nvidia CEO Jensen Huang stated that the H200 remains the second-best AI chip globally, ranking behind Nvidia’s newer GB200 AI chip. Source: @StockMKTNewz on X, Oct 28, 2025. Nvidia introduced the GB200 Grace Blackwell platform at GTC 2024 as its next-generation AI system, following the earlier Hopper-based H200 announced in November 2023, establishing GB200 as the newer flagship over H200. Source: Nvidia GTC 2024 keynote and Nvidia H200 product announcement. For traders, this clarifies Nvidia’s data-center chip hierarchy with GB200 at the top and H200 next, a configuration that guides procurement priorities and deployment roadmaps across high-performance AI workloads. Source: @StockMKTNewz and Nvidia product announcements. This hierarchy also matters for AI-linked crypto infrastructure, as decentralized GPU networks and marketplaces reference Nvidia GPUs for AI rendering and compute, including Render Network token RNDR and Akash Network token AKT. Source: Render Network documentation and Akash Network documentation. |
|
2025-09-18 14:03 |
Microsoft Unveils Fairwater, the World’s Most Powerful AI Datacenter: Hundreds of Thousands of NVIDIA GB200 GPUs, 10x Supercomputer Performance, 2 GW Capacity Added
According to Satya Nadella, Microsoft added over 2 gigawatts of new GPU capacity last year and announced Fairwater in southeastern Wisconsin, described as the world’s most powerful AI datacenter. Source: Satya Nadella on X, Sep 18, 2025. Nadella said Fairwater is a seamless cluster of hundreds of thousands of NVIDIA GB200s connected by fiber long enough to circle the Earth 4.5 times, delivering 10x the performance of today’s fastest supercomputer for AI training and inference. Source: Satya Nadella on X, Sep 18, 2025. He stated the system integrates compute, network, and storage so a single job can run at exponential scale across thousands of GPUs from day one, with liquid-cooled closed-loop design requiring zero water for operations after construction and all energy consumption matched with renewable sources. Source: Satya Nadella on X, Sep 18, 2025. Nadella added that multiple identical Fairwater datacenters are under construction across the US, with AI infrastructure already deployed in over 100 datacenters across more than 70 regions worldwide. Source: Satya Nadella on X, Sep 18, 2025. |