Group 1 - Microsoft plans to leverage its access to OpenAI's custom AI semiconductor development to enhance its own chip development efforts [1] - Microsoft CEO Satya Nadella stated the intention to replicate and expand upon OpenAI's developed technologies [1] - A revised agreement allows Microsoft to use OpenAI's models until 2032 and conduct related research until 2030 or until general AI is deemed achieved [1] Group 2 - OpenAI is collaborating with Broadcom to design and produce a custom AI acceleration chip (AI ASIC) [2] - The first batch of AI chips is expected to be delivered as early as next year, posing significant competitive pressure on Nvidia in the AI infrastructure market [2] - AI ASICs and Nvidia's AI GPUs represent two distinct technological paths in AI chips, with AI ASICs offering advantages in cost-effectiveness and energy efficiency for large-scale cloud computing giants [1]
微软(MSFT.US)拟借力OpenAI攻坚定制AI芯片,CEO透露“先复制后拓展”技术路径