Posted inAI News Nvidia Pushes Its New AI Rack to 2028, Over a Year Late Nvidia pushes its Kyber AI rack a full year to 2028. A 78-layer PCB midplane blocks fabrication, and the fallback design was cancelled.
Posted inAI News Meta Meta Compute: Zuckerberg Rents Out Spare AI GPUs Meta Compute wants to rent Zuckerberg's spare AI GPUs to AWS, Azure and GCP customers. A 182.9 billion dollar infra bet ready to monetize.
Posted inAI News Deepseek DSpark: DeepSeek Boosts AI Speed 85% on GPUs DeepSeek releases DSpark, a speculative decoding framework boosting model speed 60 to 85% on GPU. An open-source statement.