Gimlet Labs has developed what it calls the first multi-silicon inference cloud, software and an API that slices and orchestrates agentic AI workloads to run simultaneously across CPUs, GPUs and high-memory systems. The company claims its approach can speed AI inference by 3x to 10x at comparable cost and power, and can split model execution across different chip architectures. Gimlet has partnered with chip makers including NVIDIA, AMD, Intel, ARM, Cerebras and d-Matrix and targets large model makers and data centers rather than individual developers. The startup publicly launched in October with reported eight-figure revenues and says its customer base more than doubled in four months, now including a major model maker and a very large cloud computing company. Gimlet employs around 30 people and was founded by Zain Asgar, Michelle Nguyen, Omid Azizi and Natalie Serrino, who previously worked together at Pixie (acquired by New Relic).

Total raised

$80M

Funding rounds

1

Latest round

Series A

Latest activity

Mar 2026