Snowcap Compute secures $23 million for AI superconducting chips
Snowcap Compute, a startup working on super-efficient AI chips, just raised $23 million.
The round was led by Playground Global, with Cambium Capital and Vsquared Ventures joining in.
Former Intel CEO Pat Gelsinger not only led the investment but also joined their board, saying power limits are a huge challenge for today's data centers.
Snowcap's chips could revolutionize data centers
Snowcap's chips use superconductors—meaning electricity flows with zero resistance—so they can deliver up to 25 times better performance per watt than current chips (even factoring in cooling).
While they need to stay super cold, these chips could help data centers handle massive power demands; for context, NVIDIA's upcoming "Rubin Ultra" server is set to use around 600 kilowatts.
The team includes ex-NVIDIA and Google chip leaders
The team includes superconducting chip experts Anna Herr and Quentin Herr, ex-NVIDIA and Google chip leaders, and CEO Michael Lafferty (formerly at Cadence Design Systems).
They're aiming to launch their first chip by late 2026 and hope to make AI hardware way more efficient.