Gimlet Labs Raises $80M to Run AI on Any Chip

Gimlet Labs just landed $80 million in Series A funding for software that does something chip makers have been fighting over: let AI models run across NVIDIA, AMD, Intel, ARM, Cerebras, and d-Matrix processors at the same time.

The startup’s tech tackles a real problem. Right now, companies building AI systems are locked into whatever chips they choose. Want to switch to a cheaper alternative? Too bad. Need to mix different hardware for different tasks? Doesn’t work. Gimlet’s platform sits between the software and the silicon, making it transparent to developers which chips are doing the work underneath.

The funding comes as data centers everywhere are scrambling to source enough GPUs to keep up with AI demand. By letting operators use whatever chips they have on hand—old inventory, cheaper alternatives, whatever—Gimlet removes a massive constraint on deployment.

The company hasn’t disclosed which investors led the round, but the size signals serious confidence in the approach. This isn’t a theoretical problem either. Companies are already losing money on stranded hardware they can’t use for their current AI workloads.