Gimlet raises $80M for multi‑silicon inference
Gimlet Labs closed an $80M Series A to run inference across NVIDIA, AMD, Intel, ARM, Cerebras and d‑Matrix chips — investors are backing multi‑silicon inference as a real route to lower cost/latency. That funding increases pressure on single‑vendor stacks and pushes startups to evaluate heterogenous inference strategies. (techcrunch.com (menlovc.com)
The round was led by Menlo Ventures and lists Eclipse, Factory, Prosperity7 and Triatomic as participating investors, according to Gimlet’s announcement. (gimletlabs.ai) Founder and CEO Zain Asgar is described as a Stanford adjunct and a founder with a prior successful exit in TechCrunch’s profile of the company. (techcrunch.com) Gimlet says it emerged from stealth five months ago and reported eight‑figure revenues at launch, signaling early commercial traction. (finance.yahoo.com) The company reports it has tripled its customer base and recently added one of the top‑three frontier model labs and a top‑three hyperscaler to its customer roster. (finance.yahoo.com) Gimlet delivers its technology as both software and an API through a hosted Gimlet Cloud product and positions that offering for large model labs and data‑center deployments rather than ordinary app developers. (gimletlabs.ai) A joint announcement with d‑Matrix says Gimlet Cloud will deploy on d‑Matrix’s Corsair low‑latency hardware and claims up to 10x speedups plus large power‑efficiency gains on frontier AI workloads. (marketwatch.com) TechCrunch reports Gimlet’s stack can slice a model so different portions run on different architectures, a capability the company frames as a route to higher utilization across generations and vendors. (techcrunch.com)
