Scripps Gets 14M GPU Hours From Huang Foundation

Scripps Research said on 19 October 2026 that it has received an in-kind grant of more than 14 million hours of compute time from the Jen-Hsun and Lori Huang Foundation to push AI deeper into biomedicine. The hours come through the foundation’s partnership with CoreWeave using NVIDIA GPUs. There is no dollar value attached in the announcement — this is compute, not a cash gift.

For lab scientists and drug hunters, the pitch is scale: enough GPU time to train and run models that would otherwise sit in a queue or never leave a laptop. The projects described are research plans and tool-building efforts, not published clinical results.

In brief

Scripps Research receives more than 14 million GPU hours from the Jen-Hsun and Lori Huang Foundation via CoreWeave/NVIDIA (Scripps release, 19 October 2026).
Focus: Bryan Briney’s “Virtual Germinal Center” antibody model and Stefano Forli’s “dark” proteome simulations for transient drug targets.
In-kind only — no dollar figure; tools are meant to be openly available; these are plans, not finished medicines.

What the compute is meant to power

Professor Bryan Briney’s Antibodies for All project aims to build a “Virtual Germinal Center”: an AI system that models how human antibodies mutate, bind their targets and are selected during an immune response. Trained on millions of paired antibody sequences and structural data, it is intended to let researchers test vaccine candidates and search for therapeutic antibodies in silico before wet-lab work.

Professor Stefano Forli’s team is targeting the “dark” proteome — thousands of human proteins whose dynamic behaviour has resisted structural study. By simulating them at scale under realistic conditions, the group hopes to catch how those molecules flex and shift, revealing transient druggable pockets that static structures miss.

Both efforts will also draw experimental inputs from the labs of Andrew Ward, Christopher Parker and Benjamin Cravatt, Scripps says. Ward, helping lead the collaboration with the foundation, said pairing Huang Foundation computing with Scripps science aims to shorten the path from breakthrough to something that can help patients.

Open tools, no price tag

Jensen Huang, co-president of the Huang Foundation, said scientists at Scripps are using AI at scale to simulate antibody evolution and a dynamic map of the human proteome, and that “the resulting tools will be openly available for scientists everywhere searching for new medicines.” That open-availability pledge is part of the philanthropic framing; the release does not describe a commercial licence or a fee schedule.

The award’s size is stated only in GPU hours. Compressing timelines and training next-generation models on Scripps datasets are the benefits the institute emphasises — not a cash equivalent. Readers comparing this to dollar-denominated AI grants should treat the 14 million hours as an in-kind capacity gift.

Our read

This is compute philanthropy aimed at two of biomedicine’s hardest modelling problems: immune repertoire dynamics and undruggable proteins. The compliance-and-trust angle for open scientific AI is whether the promised open tools actually ship with enough documentation and access for labs outside La Jolla. Watch for first model or dataset releases; until then, the story is capacity granted, not cures delivered.

How we sourced this

Primary: Scripps Research news release (19 October 2026). GPU-hour total, CoreWeave/NVIDIA route, project descriptions and Huang quote are from that page. No dollar value is given there.