AI 热点

Perplexity 年化收入达 10 亿美元,现金消耗约为 Anthropic 的五分之一

Rohan Paul更新于 11月18日 05:33

Perplexity has reached a $1B annual revenue run rate on 5x less cash burn than than this week’s giant AI IPO.

AI 导读

Perplexity 年化收入运行速率已达 10 亿美元,现金消耗约为本周大型 AI IPO 主体的五分之一。创始人 Aravind Srinivas 在 Semafor 采访中称,该运行速率在 2026 年初低于 2.5 亿美元,8 月突破 7.5 亿美元;Anthropic 的 S-1 显示其在 2023 和 2024 两年烧掉 73 亿美元运营现金才达到约 10 亿美元。他将低消耗归因于结构差异:Perplexity 不训练旗舰前沿模型,而是把任务路由给 Claude、GPT 和 Gemini,并认为用户关心任务是否完成,而非由哪个模型执行或消耗多少 token。

推荐理由

AI 评分 62

Perplexity 以低于 Anthropic 的现金消耗达到 10 亿美元年化收入,其不训练前沿模型、转调 Claude、GPT、Gemini 的路线提供了一个成本结构对照。

主题

标签

正文

Perplexity has reached a $1B annual revenue run rate on 5x less cash burn than than this week’s giant AI IPO.

@AravSrinivas (Perplexity founder) with his interview with Semafor

Anthropic’s S-1 shows $7.3B of operating cash burned in 2023 and 2024, the two years it took to reach about $1B.

Aravind said the run rate sat below $250M at the start of 2026 and passed $750M in August.

His explanation for the thin burn is structural: Perplexity doesn’t train a flagship frontier model and instead routes work to Claude, GPT and Gemini.

Customers, he argued, care whether the task got done rather than which model ran it or how many tokens it consumed. https://t.co/sCXOYzhXAb

来源

Rohan Paul