Tag: #supercomputer17 results
SoftBank's internal IT department successfully developed Japan's top-performing AI supercomputer in just four months. This 'President's Project' highlights the capability of internal teams to execute high-stakes infrastructure projects.
xAI expanded Colossus supercomputer to 555,000 NVIDIA GPUs across three Mississippi facilities, totaling 2 GW power and $20B investment, making it the world's largest AI training cluster. It runs on on-site natural gas turbines to bypass grid limitations. Environmental lawsuits challenge permit compliance for the turbines.
SoftBank invested 122 billion yen in an AI supercomputer. Developers explain the technical and reliability reasons for choosing NVIDIA GPUs despite initial hardware challenges.
xAI has announced a new compute partnership with Anthropic, providing access to Colossus 1. This supercomputer is the world's largest and fastest-deployed AI cluster, supporting Anthropic's AI R&D and services.
OpenAI is scaling its Stargate project to build compute infrastructure for the Intelligence Age and powering AGI. The initiative includes adding new data center capacity to meet surging AI demand.
SpaceX secures Cursor with 600 billion options before IPO, paying 100 billion collaboration fee—swapping compute for ecosystem. Memphis supercomputer to San Francisco incurs 40ms latency barrier. Cursor hits 1.75 trillion valuation at 90x price-to-sales ratio as Musk bets on future ecosystem.
SpaceX and Cursor have partnered to develop the world's best AI for coding and knowledge work, leveraging Cursor's coding tool and SpaceX's Colossus supercomputer. SpaceX may invest $10B or acquire Cursor for $60B later this year. Cursor will use the infrastructure to overcome compute bottlenecks and scale its models.
Monarch is a new API that simplifies running distributed training jobs on massive clusters. It particularly aids complex setups like distributed reinforcement learning. The tool addresses key challenges in debugging large-scale training.
Elon Musk revealed xAI's Colossus 2 supercomputer cluster is currently training six AI models. This update highlights ongoing massive compute efforts at xAI.
NVIDIA CEO Jensen Huang personally delivered the first DGX GB300 AI supercomputer to Kapasi. This marks the initial rollout of NVIDIA's latest high-performance AI infrastructure system. The event underscores the excitement around the new GB300 platform.