卖什么
AI推理云。
商业模式拆解
买家：谁掌握预算？
未核实Engineering/infrastructure leaders at organizations deploying AI inference at scale (hypothesis from platform positioning).
触发点：什么让他们现在购买？
Inference is becoming the bottleneck as AI agent and product usage grows; need fast, affordable, reliable token throughput.
工作流：什么输入变成什么输出？
Customers run model inference on Groq capacity through layered offerings: GroqMetal dedicated bare-metal infrastructure, GroqCore production inference stack, and GroqAssured enterprise governance; LPU racks deliver about 1,000 tokens/sec/user.
经济账：省了什么，怎么收费？
Sells inference capacity (cloud and dedicated infrastructure) built on its own LPU chips; building hundreds of megawatts across 13 data centers on four continents; unit pricing not on groq.com.
渠道：怎么接触到客户？
Developer self-serve "Start building" plus enterprise/dedicated capacity via contact.
融资历史与报道
| 融资日期 | 轮次 | 金额 | 投资方 | 来源 |
|---|---|---|---|---|
| 2026-08-17 | — | US$ 350 mi | — | news.google.com |
档案汇总自公开报道和公司官网。商业模式部分是 MaxBC 对这些材料的解读，不是已确认的事实；未知之处留空，不作猜测。 更新于 2026-10-12.