Gimlet Labs, founded by Stanford professor Zain Asgar, raised $80 million Series A led by Menlo Ventures, bringing total funding to $92 million. The startup tackles AI inference bottlenecks with software that simultaneously runs workloads across diverse hardware—CPUs, GPUs, and high-memory systems—boosting efficiency by 3x to 10x.

Traditional AI deployment wastes resources, with existing hardware utilized just 15-30% of the time. Gimlet's orchestration software slices workloads to optimize each component across different architectures. Partnerships with chip giants NVIDIA, AMD, Intel, ARM, Cerebras, and d-Matrix validate their approach to solving hardware fragmentation in AI.

Gimlet launched in October with eight-figure revenues and has since doubled its customer base to include major model labs and cloud providers. With McKinsey predicting $7 trillion in data center spending by 2030, Gimlet's efficiency gains position it to capture a significant portion of the market optimizing existing infrastructure rather than requiring new hardware builds.