UW secures $10M to expand AI infrastructure, prioritizing public-interest research and data security
Washington Sen. Patty Murray toured the University of Washington on Friday after securing $10 million in federal funding to expand the university's infrastructure for data-intensive AI work. Her message was clear: AI progress shouldn't be steered only by billionaires and shareholder incentives.
"If just billionaires are creating and using AI for their own projects that make money, then we lose out on most of the benefits of AI," Murray said. She emphasized the role of public institutions in delivering value in healthcare, environmental science, workforce training, and job creation.
The funding, allocated through Congressionally Directed Spending in the Commerce-Justice-Science appropriations bill, will support Tillicum - UW's next-generation computing platform launched in October. University leaders say the investment will speed up research cycles, expand access for students and faculty, and reduce reliance on commercial cloud providers.
"This keeps us at the front of AI research," said Andrew Connolly, director of the eScience Institute. Magdalena Balazinska, director of the Paul G. Allen School of Computer Science & Engineering, added that public universities answer to taxpayers - and that access to compute is often the first question prospective faculty ask.
During the visit to the eScience Institute, students demonstrated projects built on sensitive personal and scientific data, including a voice-driven health system that tracks symptoms and produces summaries for clinicians. Running these workloads on UW-owned infrastructure helps keep sensitive data on campus and out of third-party clouds.
Having in-house compute also shortens iteration loops for students and faculty. During the tour, researchers discussed open-source work such as OLMo from the Allen Institute for AI, and how local capacity supports faster experimentation without long queue times or cloud procurement friction.
Murray, the top Democrat on the Senate Appropriations Committee, framed compute as foundational infrastructure. "If you don't have the computers, if you don't have the basic infrastructure, you're stymied," she said. "This benefits everybody - jobs, better healthcare, and a stronger pipeline of innovators in Washington state."
Additional allocations include $3 million for new fan blades at UW's Kirsten Wind Tunnel and $1.5 million for upgrades to the Radiocarbon Lab. The broader federal spending package also boosts scientific agencies such as the National Institute of Standards and Technology (NIST), countering proposals from President Trump to sharply cut federal research spending.
Why this matters for research teams
- Data governance and privacy: On-campus compute helps keep sensitive datasets local, easing collaboration with clinical and field partners while reducing third-party exposure.
- Faster iteration: Ready access to GPUs and storage shortens experimentation cycles compared to cloud budgeting, procurement, and scheduling overhead.
- Recruiting and retention: Guaranteed compute is a deciding factor for faculty candidates and graduate students evaluating where they can do their best work.
- Open science momentum: Local capacity supports open models and transparent methods, strengthening reproducibility and community contributions.
- Pragmatic hybrid strategies: Keep sensitive pipelines on-prem; burst to cloud for one-off scale needs. Plan for data locality, cost ceilings, and shared queues.
What to watch next
- Tillicum access policies, queue management, and support for multidisciplinary projects (health, climate, materials, and beyond).
- Faculty hiring, grant competitiveness, and student recruitment tied to guaranteed compute resources.
- Partnerships with hospitals and research centers that need secure, high-throughput workflows.
- Efficiency metrics: utilization, energy impact, and cost-per-experiment vs. commercial cloud.
Bottom line: Strategic public funding for compute changes what's feasible on campus - from sensitive health AI to open-source models - and gives researchers time back to run more experiments with fewer barriers.
If your lab or institute is building AI skills for 2026 projects, you can scan curated options by role here: Complete AI Training - Courses by Job.
Your membership also unlocks:
