算算openai 下了多少产能订单

Core Viewpoint - OpenAI is making significant investments in AI server capacity, with a total projected expenditure of $285 billion on chips by 2029, raising concerns about future revenue generation and financing needs [5]. Group 1: Investment Details - OpenAI has partnered with Broadcom to deploy 10GW AI servers, starting in 2026 and completing by 2029 [1]. - The total investment in chip capacity includes $100 billion with NVIDIA, $750 billion with AMD for 6GW, and $1.1 trillion with Broadcom for 10GW, summing up to $2.85 trillion [2][3][4]. - The estimated cost per chip is approximately $30,000 for NVIDIA, $25,000 for AMD's MI450, and $20,000 for Broadcom's TPU [2][3]. Group 2: Revenue Projections - OpenAI's current annual revenue is around $13 billion, with a projected growth rate of 50% per year, leading to a total revenue of $105.6 billion by 2029 [5]. - The projected expenditure on chips will exceed revenue by nearly $180 billion, indicating a potential need for external financing or partnerships with chip manufacturers [5]. Group 3: Market Sentiment - Broadcom's stock performance has been weaker compared to AMD's, suggesting market fatigue regarding U.S. tech stocks [6]. - There is a favorable outlook on Google's new products and TSMC's market dominance, alongside positive trends in storage price increases [7].