Wafer, an AI infrastructure optimization startup that uses autonomous agents to squeeze more performance out of GPUs, has raised $40 million in new funding at a valuation north of $200 million. The company also turned down acquisition offers along the way.
Wafer closed a $4 million seed round in April 2026. Five months later, it’s sitting on a valuation roughly 50 times larger.
The GPU utilization problem nobody talks about
The average GPU in production environments runs at about 20% utilization. Wafer’s technology deploys autonomous AI agents to profile and tune inference workloads across different hardware and model architectures. The system automates what would otherwise require teams of specialized engineers spending weeks manually optimizing configurations. What makes the approach particularly interesting is its focus on non-Nvidia chipsets. While Nvidia dominates the AI training market with its CUDA ecosystem, the inference side of the equation is more fragmented. Wafer positions itself as the optimization layer that makes those alternatives viable.
From seed to series at breakneck speed
