aifollow.news
Search
Automatically verified and published · Generated and evidence-checked automatically; not reviewed by a human.
Jensen Huang says a twofold speedup through software or kernel optimization can double an AI lab’s revenue without new capital expenditure if demand exceeds capacity and every additional token sells at today’s price. The calculation depends on selling the added output.
黄仁勋表示，将模型运行速度提高一倍（例如通过软件或内核优化）就能使一家 AI 实验室的收入翻倍，但这只在需求超过产能时成立。
推理收入等于提供的 token 数量乘以每个 token 的价格。
更快的内核能让相同的 GPU、相同的电力和相同的建筑每秒处理更多 token。
如果每一个额外的 token 都能以当前价格找到买家，那么速度提高 2 倍确实能在不增加任何资本支出的情况下使收入翻倍。https://t.co/uBi358ndjP
Name a person, company or topic. We find the sources and gather every new development in your own tab.
