Co-founder and CEO of Wafer, an AI inference cloud that uses AI agents to optimize open-source LLMs for heterogeneous chip architectures, demonstrated AMD parity with Nvidia at half cost.
no scored calls yet — needs a stated position or a categorical verdict, with a matured window vs SPY
Wafer uses AI agents to automate GPU kernel engineering, enabling open-source LLMs to run on AMD chips at 80-90% of Nvidia performance for half the cost, proving a chip-agnostic future where software optimization unlocks heterogeneous compute.
Wafer's AI-driven kernel optimization enables AMD GPUs to achieve 80-90% of Nvidia's throughput on open-source models like GLM and Kimi at roughly half the cost, demonstrating that software can close the hardware performance gap.