Loading page…
The Huang Foundation’s grant supplies NVIDIA computing through CoreWeave. UW expects initial model releases from late 2026 into early 2027, subject to research progress and experimental validation.
Listen to this story
Announced October 8, the University of Washington’s Institute for Protein Design received nearly 7 million GPU-hours from the Jen-Hsun and Lori Huang Foundation to support protein-design AI research. The capacity, provided through CoreWeave using NVIDIA GPUs, is research infrastructure—not a finished model—and could help the institute pursue new medicines, materials and environmental applications. The practical test will be whether the additional compute leads to experimentally validated designs and usable open releases.
UW Medicine expects the added computing power to reduce model learning and evaluation time from months to days; this is a projected benefit, not a measured result.
Research targets include more precise cancer therapeutics, durable vaccines, work on currently untreatable diseases, and enzymes that degrade toxins and plastics.
UW plans to share models, software and data openly, with initial releases anticipated from late 2026 into early 2027, subject to research progress and experimental validation.
The University of Washington’s Institute for Protein Design has received nearly 7 million hours of computing time to develop open-source AI for designing proteins. Announced October 8, the Jen-Hsun and Lori Huang Foundation’s in-kind grant supplies computing access for research aimed at new medicines, materials and technologies—not a finished set of models.
The foundation is providing the capacity through its partnership with CoreWeave, using NVIDIA graphics processing units, or GPUs.
UW Medicine says the additional computing power will cut algorithm learning and evaluation time from months to days. That is the institution’s expected research benefit, not a published performance result.
Baker, a UW biochemistry professor and 2024 Nobel laureate in chemistry, tied that faster training to more precise design across several kinds of molecular technology.
This computing power will allow us to train models on the immense datasets we are collecting that define how matter interacts, at speeds that were unavailable to us before.
David Baker, director of the Institute for Protein Design
The institute’s work combines predicting protein structures with designing new proteins that have particular functions. UW describes applications extending beyond medicine into technology and sustainability. The computing grant is intended to help scientists tackle complex design problems across those fields, with the announcement identifying several research targets:
The institute plans to openly share AI models, software tools and research data so scientists worldwide can adapt the work to their own areas of study. It also intends to publish findings in open-access publications.
Initial AI model releases are anticipated to begin in late 2026 and continue into early 2027. UW makes that schedule conditional on research progress and experimental validation.
Loading discussion...
Join the conversation
Explain when open sharing should be a funding condition.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.