Gimlet Labs raises $80M
Gimlet Labs closed an $80M Series A to build hardware‑agnostic inference infrastructure that abstracts NVIDIA, AMD, Intel, ARM, Cerebras and d‑Matrix — a bet on portability for production GenAI workloads. (techcrunch.com)
Menlo Ventures led the Series A and the round lists Eclipse Ventures, Factory, Prosperity7 and Triatomic as participating investors. (gimletlabs.ai) Gimlet’s founding team is led by Zain Asgar with co‑founders Michelle Nguyen, Omid Azizi and Natalie Serrino, and the company publicly launched from stealth in October 2025 with reported eight‑figure revenues. (techcrunch.com) The company says its Gimlet Cloud slices agentic compute graphs into fragments and maps those fragments to the optimal hardware at runtime, yielding reported 3x–10x end‑to‑end speedups on >1T‑parameter frontier models within the same power envelope. (gimletlabs.ai) Gimlet has tested SRAM‑centric Corsair accelerators in its stack and published comparisons showing 2–5x interactive request speedups and up to 10x gains on energy‑optimized configurations versus GPU‑only setups. (gimletlabs.ai) Under the hood Gimlet describes an intelligent workload orchestrator, a hardware‑agnostic compiler that lowers fragments to accelerator implementations, and autonomous kernel generation to produce optimized runtimes for differing silicon. (gimletlabs.ai) Since launching, Gimlet says its customer base has tripled and now includes a top frontier model lab and a hyperscaler, and the company’s seed plus Series A brings total capital raised to roughly $92 million. (gimletlabs.ai)
