Google 用 AlphaChip 加速 TPU 芯片设计，Gemini 正进一步放大效果
Google 用 AlphaChip 自己设计 TPU，Gemini 又在加速这条链路，形成模型造芯片、芯片训模型的闭环。文章点出论文数据两年没更新，真实进度外界看不到。关心算力竞争和自研芯片路线的人都该看看这条推文。
Andrew Curran 引用 Google 2024 年 AlphaChip 论文的数据指出，Google 早在同行之前就把 AI 引入芯片设计流程，Gemini 正在加速这一循环。
AlphaChip 是 Google 将强化学习用于芯片平面布局的方法，曾在 Nature 论文中公开，并应用于 TPU 系列的宏单元布局。Curran 特别强调，论文中引用的数字已超过两年没有更新，实际情况可能已远超论文所列水平。
对芯片行业而言，Google 把模型训练、芯片设计、下一代 TPU 串成闭环的做法，形成了硬件层面的正反馈：模型设计出更强的芯片，更强的芯片再训练更强的模型。
Gemini is accelerating all of this. These numbers are from the AlphaChip paper in 2024. God knows what these numbers are now, they have not been updated in over two years. Google started leaning into the AI hardware design loop long before anyone. https://t.co/RPjJpVR5AJ
This report is categorized under AITOP's "Industry" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include AlphaChip, Google, TPU, Gemini.
Source attribution: curated from Andrew Curran (@AndrewCurran_).
