Gimlet Labs完成3亿美元B轮融资，押注AI推理性能优化

AI基础设施公司Gimlet Labs宣布完成3亿美元B轮融资，由Andreessen Horowitz领投，Sapphire Ventures、Menlo Ventures、Arm、Samsung Ventures、Tiger Global Management等18家机构参投。1

该公司致力于构建首个多芯片云平台，通过异构硬件和软件栈优化来提升AI推理性能。1根据公司披露，这些技术方案可在同等功耗下实现5至10倍的性能提升，或在同延迟约束下提升吞吐量。1Gimlet方面指出，这样的优化对应对日益增长的token需求和推理延迟要求至关重要。1

自5个月前完成A轮融资以来，Gimlet已新增数十亿美元合同收入、吉瓦级数据中心管道，并管理数百兆瓦容量。1在token生成方面，过去12个月增长6倍，该公司预计到2030年将增长20倍。1行业层面，AI投资已接近每年1万亿美元，到2030年累计投资预计达7万亿美元。1AI数据中心功耗在2025年约为18吉瓦，预计到2030年将增至54吉瓦。1

Gimlet Labs, an artificial intelligence infrastructure company, has closed a $300 million Series B funding round led by Andreessen Horowitz 1. The company is developing what it describes as the first multi-chip cloud platform, designed to optimize AI inference performance through heterogeneous hardware and software stack improvements 1.

According to the company, its approach delivers between 5 and 10 times greater performance at equivalent power consumption, or similar throughput at equivalent latency 1. This advancement addresses growing demand for token generation and reduced inference latency in AI systems 1. The funding round included participation from 18 institutional investors, including Sapphire Ventures, Menlo Ventures, Arm, Samsung Ventures, and Tiger Global Management 1.

Since completing its Series A funding round five months prior, Gimlet Labs has secured billions of dollars in new contract revenue, built a gigawatt-scale data center pipeline, and brought hundreds of megawatts of capacity under management 1. Token generation volumes have increased six-fold over the past 12 months, with projections suggesting a 20-fold increase by 2030 1. Industry investment in AI infrastructure has nearly reached $1 trillion annually and is projected to accumulate to $7 trillion cumulatively by 2030 1. AI data center power consumption is expected to grow from approximately 18 gigawatts in 2025 to 54 gigawatts by 2030, a threefold increase 1.