Mirendil, a frontier AI lab focused on accelerating AI development, has selected Google Cloud‘s AI Hypercomputer as its infrastructure platform.

The partnership includes a mix of Google’s TPU accelerators and full-stack NVIDIA AI infrastructure on Google Cloud, supporting model pre-training and post-training.

Mirendil aims to accelerate and democratize AI research by building AI systems that can run the research loop faster.

Google Cloud collaborated on end-to-end design and deployment of combined TPU and NVIDIA infrastructure across compute, storage, networking, and control planes.

Mirendil is already live with a cluster of TPU v5P chips, with NVIDIA systems coming online soon.

The ability to choose between TPU and NVIDIA architectures allows Mirendil to match workloads to the best-suited architecture over time.