About Cleanroom

Every research institution should be able to offer its researchers private AI.

Cleanroom lets universities and research institutions run private AI on the GPU clusters they already own, by placing AI models in GPU memory that research jobs leave unused. No new hardware, no per-use cloud bills, and data stays on campus.

Team

The founders

Hugo Chiang

Co-founder, business lead

NYU student

Hugo leads customer discovery, partnerships and strategy.

Tim Chao

Co-founder, technical lead

Northwestern student

Tim leads Cleanroom's technical work on GPU systems.

Where we are

Universities want campus AI. Our next step is showing that shared memory doesn't slow research down.

  1. Demand

    Universities are building it themselves

    Universities such as UC San Diego and the University of Michigan have built their own campus AI services, and research computing teams we've spoken with explained why.

  2. Data

    The unused memory is real

    Our analysis of 62,827 GPU-hours on a U.S. research university cluster found that 56% peaked at a quarter or less of GPU memory. We also have requests for GPU usage data in progress at other universities. See the evidence.

  3. Next

    A controlled experiment

    We're running research software alongside AI models to measure slowdown (our target is under 5%), response time, and how safely memory is handed back.

Work with us

Researchers and engineers: let's talk.

If you work on GPU systems, HPC or research computing and this problem interests you, we'd like to hear from you as a possible collaborator, advisor or future teammate.

Get in touch

See what your cluster could offer your researchers.

We're looking for research computing teams to work with. In a first conversation, we'll compare notes on your cluster and what your researchers need from AI.