April 24, 2025
The U.S. biomedical innovation system is world-leading, built by a century of federal investment that peaked last year at $50 billion. That foundation has been thrown into uncertainty by shifts in policy under the second Trump Administration aimed to reduce spend, shift research priorities, and divest from universities. While these changes have appeared to many to come on suddenly, they reflect long-growing concerns with the U.S. scientific system—initiated by issues with trust and reproducibility, barriers to accessing knowledge, and structural limitations on creative risk-taking.
The scientific community is actively seeking additional funders to step in and fill the anticipated funding gap, which may be as big as $20B annually. New pledges are starting to be made. At the state level, California is advancing a bill to fund general scientific research and Texas is advancing a bill to fund Alzheimer’s research. Both of these states have a history of scientific funding, including through the California Institute for Regenerative Medicine and the Texas Cancer Prevention and Research Institute of Texas (CPRIT). Corporate pledges are also happening—Recursion Pharmaceuticals has funded a pre-seed accelerator designed to close gaps in SBIR funding—and groups outside of the US are launching opportunities for funding or training that are aimed at US-based scientists. Other potential sources of funding include philanthropies, family offices, regional governments, private investors, disease foundations, and crowd-funding. If contributions from these groups rise to meet the anticipated $20B funding gap, the system will shift fundamentally—from one organized around a central funder to one sustained by a kaleidoscope of funders. This transition will open-up new and interesting opportunities.
A shift in how science is funded will be accompanied by a shift in how science is practiced. The university, which stands as the current institutional model for science, has been honed over centuries to support the discovery of new knowledge. Over the past 75 years, the US universities have evolved in lockstep with federal science funding—both shaped around the structural frame of an individual lab run by a single investigator and centered on a specific line of inquiry. This mutual reinforcement has created a feedback loop: universities structure themselves to win grants and grants are structured to support the kinds of work that universities are best equipped to host. In many cases, this alignment runs so deep that universities place limits on accepting non-federal funding, particularly when that funding requires different administrative handling. While this structure continues to excel at discovery research, it has struggled to support other modes of science—particularly those that seek to integrate data science practices like machine learning and the increasingly ubiquitous artificial intelligence. These approaches are certainly in use within universities, but they rely on infrastructure that has been bolted onto an organizational model originally built for analog-era research.
At the same time that federal funding has started to pull back from universities, there has been a rise in the creation of independent research institutes. Non-federal funders who are interested in projects that vary from NIH-funded research in scale, duration, and outputs are increasingly choosing to pull scientists out of the university and instead place them into smaller purpose-built institutes. This allows them to be laser-focused on their research mission without the constraints of existing institutions.
Many of these institutions share a key organizing principle: they are built as digital natives with data science practices integrated as a core element of design. An essential requirement is the ability to position biological scientists, data scientists, and software engineers as essential—and equal partners in scientific discovery. This remains hard to achieve in the academic setting where salary caps and promotion structures often discourage technical talent from engaging. These multi-disciplinary teams can be incentivized around shared goals and coordinated activities. In addition, they can be supported by infrastructure explicitly developed with data science in mind. Every piece of data collected can be made machine-readable. Every workflow can be built for automation. All analyses can be structured to be performed and interpreted with machine assistance. This kind of integration of computation, laboratory automation, and AI enables real and meaningful integration of biological and data science in service to their research goals.
Digitally native institutions are increasingly charting the way in the building of public goods—from data to technology—designed to unlock new frontiers in science. An early example of this is the Allen Institutes, which have spent the past 20 years building a series of Institutes with independent domain focus and a shared commitment to release open datasets and tools in leading areas of research. The Chan Zuckerberg Initiative (CZI) has followed a similar path by designing a network of biohubs that each tackle a singular problem in biomedicine—like mapping and engineering human immune cells—with the goal of harnessing technology to advance scientific knowledge. This strategy is also embedded in the Arc Institute—with a focus on genomic research—and in several of the Focused Research Organizations (FROs) designed by Convergent Research—these are purpose-built teams funded for a fixed period to tackle a specific, bounded scientific challenge.
Other digitally native institutions are being built to bridge discovery and application—enabling high-impact research that translates directly into real-world use. Several of these are designed to act as incubators—exploring basic research with the goal of developing products to out-license to biotech. Examples here include Sanford Labs, and Arena BioWorks. By staying highly specialized, these institutes can streamline partnerships and focus on delivering outcomes designed around real-world use cases. With appropriate governance to mitigate any potential conflicts, this provides an important opportunity to integrate discovery and development between nonprofit and commercial entities. Flexible structuring also helps to navigate challenges around technology transfer and licensing from academia.
This approach is even showing up in private institutions. Altos Labs—a $3B for-profit biotech focused on cellular rejuvenation, is the most notable example. Altos has organized itself as three inter-related research institutes that are staffed with scientists drawn largely from academia. This initiative is reminiscent of the corporate research labs that were once a mainstay of the pharmaceutical industry—where in house discovery research was considered a key competitive advantage. The rise of Altos and similar efforts raises the question of whether decentralization of research may lead to a broader shift back towards corporate-led discovery research, this time with team-based data science at their core.
US federal science funding has played more than just a financial role—it has also acted as a unifying force by helping to coordinate scientific judgment across disciplines and institutions to set national research priorities. What happens if that center no longer holds? This is also reflected in how philanthropy supports science today. Philanthropic sources already contribute an estimated $10B-$20B annually to biomedical research. While a few major funders operate with broad strategies and budgets in the hundreds of millions or more, most contribute far smaller amounts—typically between $1M to $20M annually—and are directed toward highly specific research interests. When these contributions complement a strong federal backbone, they play a powerful role in addressing gaps and catalyzing high-risk exploratory ideas. As the system becomes more reliant on distributed funding, the absence of a central coordinating force becomes a bigger issue. It becomes harder to align funding with shared priorities, sustain large-scale infrastructure, or ensure comprehensive coverage across scientific fields. And while the existing model is often criticized for its reliance on consensus and collective reasoning, the alternative risks tilting the research agenda toward a narrower set of interests—with uneven reach and limited accountability.
This is a moment to build with intention. Scientific goals at today’s digitally native institutes are independently directed but tend to cluster around cutting-edge areas with therapeutic potential including in aging and brain science, leaving fields such as women’s health and disease prevention significantly underfunded. In a decentralized funding landscape, a coordinated voice will be more important than ever to help guide this kaleidoscope of funding streams. Organizations like NASEM, FASEB, and disease advocacy foundations may need to step into this role—identifying gaps, surfacing emerging opportunities, and mapping the ecosystem as funding sources shift over time. Further, changes in our system will only succeed if they bring meaningful career pathways for scientists that reward expertise and foster opportunity for the next generation of discovery leaders.
The question isn’t whether the system will change—it already has. The real opportunity now is to shape that change with foresight: to reimagine how we fund, organize, and measure science in a way that reflects the complexity, urgency, and possibility of the moment.
– Lara Mangravite, PhD
First Five
First Five is a curated list of articles, studies, and publications that have caught our eye on the topics of public goods and emerging applied technologies.
1/ Engineering life
Evo2, a biological foundation model can “read and write” in the language of DNA to create RNA and protein with desired characteristics. Released as open source tool by Arc Institute.
2/ How the brain communicates
The MICrONs project mapped anatomical and functional neuronal connectivity across an entire cubic millimeter of primary visual cortex. Released as open data by Allen Institute for Brain Science.
3/ Open evolution of the scientific publication
PLoS is working on effective ways to properly represent attribution as the meaning of scientific contribution changes.
4/ Intermittent Fasting in Plankton
More types of plankton exist than should given the scarcity of shared nutrients. Turns out they take turns consuming them. Article released by the Santa Fe Institute.
5/ Priorities Priorities Priorities
A map of technical bottlenecks that hold back scientific progress just released by Convergent Research. Pick one and unlock the future!
