· 供应链分析
关于 $ORCL、$NBIS 及其他新云厂商 GPU 折旧论点的简要总结： Burry 的一般主张：处于 2-3 年产品周期的芯片资本支出（Capex）不应导致有用寿命的延长。 这一观点错误但具有细微差别： -> 以 $GOOGL 为例，其 7 年前的张量处理单元(TPU) 仍以 100% 利用率运行。 “谷歌称 TPU 需求超过供应，声称 8 年前的硬件迭代‘利用率达 100%’”——数据中心动态 -> 以 $NVDA 为例，A100（PCIe 和 SXM4 变体）仍在运行。它们于 2020 年发布，现在是 2025 年。旧型号保持价值并仍能带来收益。 它们具有残值，5 年后仍能高价转售。他在货架寿命上错了，但在某些超大规模云服务商如何拉伸利用率概念以证明激进资本支出合理性方面，他可能是对的。
Just a TLDR on the GPU depreciation argument for $ORCL, $NBIS and other neoclouds, Burry's general claim: capex of chips on a 2-3 yr product cycle should not result in the extension of useful lives. This is incorrect but nuanced: -> in $GOOGL's case, their TPUs from 7 years ago are still run at 100% utilization. " Google says TPU demand is outstripping supply, claims 8yr old hardware iterations have “100% utilization” -Data Center Dynamics -> in $NVDA's case A100 (PCIe and SXM4 variants), are still in operation. They were launched in 2020. It's 2025 now. Older models keep their value and still deliver equity. They have residual value and still re-sale for a lot 5 years later. He's wrong on shelf-life, but might be correct in how some hyperscalers stretch the idea of utilization to justify aggressive capex.