Gimlet Labs builds AI inference orchestration software that runs workloads across diverse hardware types.
$80MSeries A
An $80M Series A for inference orchestration signals that the market is past the 'which GPU to buy' phase and into 'how do we run workloads across whatever hardware we already own'—a pragmatic shift as GPU scarcity eases and cost optimization becomes the real lever. Gimlet's likely burning this on sales/go-to-market and R&D to support more silicon types (TPUs, custom accelerators, CPUs), which means they're betting enterprises will standardize on orchestration layers rather than lock into single vendors. If you're building any layer of the inference stack (quantization, routing, caching), watch whether Gimlet becomes the de facto control plane—it changes what you optimize for.
