Gimlet lands $80M
Gimlet Labs raised an $80M Series A to build a multi‑silicon inference cloud spanning NVIDIA, AMD, Intel, ARM, Cerebras and d‑Matrix — investors and customers are backing heterogeneous stacks. That funding strengthens the multi‑vendor inference narrative and raises the risk startups design to avoid single‑vendor lock‑in. (techcrunch.com) (menlovc.com)
Series A was led by Menlo Ventures with participation from Eclipse, Factory, Prosperity7 and Triatomic. (gimletlabs.ai) Gimlet publicly launched in October 2025 after a seed round led by Factory that raised roughly $12 million. (siliconangle.com) The founding team lists Zain Asgar (CEO), Michelle Nguyen, Omid Azizi and Natalie Serrino, and Menlo notes Asgar previously founded Pixie and served as a Stanford adjunct. (techcrunch.com) Gimlet says its customer base has tripled since launch and now includes “a top frontier lab” and a hyperscaler. (gimletlabs.ai) The company reported eight‑figure revenue at launch and the CEO has estimated many AI apps use existing hardware only about 15–30% of the time. (techcrunch.com) Gimlet claims 3–10× speedups on models larger than 1 trillion parameters within the same power envelope and says delivering those gains requires substantial investment in physical datacenter infrastructure. (gimletlabs.ai) Menlo frames the opportunity as one where GPUs remain the workhorse for batch inference while SRAM‑focused accelerators and CPUs provide dramatic latency or orchestration advantages, and where older GPU generations are being redeployed rather than retired. (menlovc.com) The stack is positioned to let teams import existing PyTorch or HuggingFace pipelines and to be delivered as software, an API, or via Gimlet’s own cloud. (menlovc.com)
