Science in 2025 is deeply under threat. While a fortunate few labs at prestigious institutions continue to pull in grants, the collapse of funding from the United States government threatens to crush an entire generation of scientific talent. The NSF faces potential cuts of up to 50–66% of its current funding³, while the NIH faces a potential 40% slash to its budget² ¹⁷. Spending cuts of this size would eliminate entire institutes. Meanwhile, venture capital in biotech has entered what analysts delicately term a "period of sustainable investing," meaning that money has dried up for anyone not named ‘Flagship’ or sporting a résumé from Genentech's C-suite¹⁰.
But hope is not lost. DeSci at its best offers a fundamental reimagining of the current system. Forget the governance theater of DeSci's first act. DeSci V2 represents something more ambitious: the construction of a parallel scientific infrastructure that operates by different rules¹. Think of it as building Raphael's School of Athens, but onchain. The goal is a new paradigm in the sciences where capital formation moves faster than grant cycles, where postdocs can afford both rent and reagents.
The Valley of Death Gets Deeper
In February 2025, 168 NSF employees received pink slips in a single morning². It became clear that the NSF and the NIH—responsible for providing grants to universities—were not immune to broader federal budget cuts². The limited grants that would continue to go to universities would also not be like the grants of the past. Grants from the government typically allow the recipients to allocate a certain portion of the funding to indirect costs (i.e., overhead costs). Along with the proposed budget cuts, the federal government also indicated that it would cap indirect costs at 15% of any grant. Institutions, which used to be able to allocate up to 40% of any grant to indirect costs, are now no longer able to subsidize overhead costs with grants¹ ². With all of these funding sources shrinking, universities are now limiting, or even halting, admissions for many of their graduate programs.
The human cost of these cuts reads like a tragedy. One postdoc, promised an NIH diversity supplement, learned via email that the funding had evaporated overnight: "I regret to have to inform you that NIH has instructed us not to issue any diversity supplements that are pending³."
The impact on early-career scientists cannot be overstated. Principal investigators report advertising postdoc positions for months without receiving a single qualified application⁴.
The private sector offers no refuge. Despite headlines about AI revolutionizing drug discovery, biotech venture funding remains incredibly selective and risk-off. Of 416 funding rounds in 2024, the vast majority went to companies with proven leadership teams and de-risked assets⁷. The days of writing checks to smart graduate students with interesting ideas are gone, replaced by a flight to quality that feels more like a stampede to safety⁸.
Why Traditional Models Cannot Scale
The crisis runs deeper than budget cuts. Traditional biotech funding operates on a tournament model: many compete, few win, and the losers get nothing. An assistant professor might spend six months writing an R01 grant with a 22% chance of success⁹. Those six months writing the grant are six months that the professor is not doing research, not mentoring students, and not advancing human knowledge. If rejected (and 78% are), they start over.
This system made sense when success rates were higher and competition less fierce. In the 1990s, NIH R01 success rates hovered around 32%⁹. Today's 22% rate transforms persistence from virtue to delusion. Over 50% of NIH funding flows to 10% of institutions, creating scientific dynasties where pedigree matters more than potential.
Emerging markets brim with underutilized talent. Brazil hosts over 1,400 VC-backed biotech startups¹⁰. India's generic drug manufacturers have demonstrated world-class chemistry capabilities. These regions remain excluded from the global research conversation because they lack access to the old boys' network that controls traditional funding.
The Onchain School of Athens
DeSci V2 proposes a different model. Instead of winner-take-all, imagine continuous funding markets. Instead of geographic gatekeeping, imagine borderless collaboration. Instead of year-long grant cycles, imagine capital deployment in weeks.
We imagine a system where milestone-based funding releases capital as researchers hit predetermined targets. Overhead drops from 70% to 5%, meaning more money actually funds science rather than administration. Continuous funding models enabled by novel capital raising eliminate the feast-or-famine cycles that plague traditional research.
But the real innovation lies in the human coordination layer. Traditional science already operates through invisible networks—researchers who share ideas, critique work, and collaborate on problems.
Imagine a prediction market for scientific reproducibility. Before investing millions in clinical trials, stakeholders can bet on whether preclinical results will replicate. Markets aggregate distributed knowledge, identifying potential failures before resources are wasted. Studies show prediction markets achieve 71% accuracy in forecasting replication, far better than peer review's dismal track record.
The implications cascade. A researcher in São Paulo can collaborate with a team in Seoul without either institution's bureaucracy interfering. A brilliant undergraduate in Mumbai can contribute to frontier research without waiting for acceptance to a PhD program. Knowledge flows permissionlessly, and at the speed of the internet.
AI x DeSci
The convergence of AI and DeSci creates possibilities that neither could achieve alone. AI-discovered drugs show significantly higher success rates in Phase I trials compared to 40–65% for traditional approaches¹¹. Centralized institutions struggle to provide this at scale.
DeSci infrastructure can solve these bottlenecks. Federated learning protocols allow pharmaceutical companies to collaborate without sharing proprietary data. The MELLODDY project demonstrated this, with ten pharma companies achieving 12–20% improvements in model efficiency by training on 2.6 billion shared data points¹¹. Each company kept their data private while benefiting from collective intelligence.
Decentralized compute networks make AI accessible to smaller research groups. Instead of paying AWS prices, labs can access GPU resources from networks like Akash or Prime Intellect at a fraction of the cost.
AI can help solve the DeSci quality control challenges. Machine learning models trained on historical data can identify promising research directions, flag potential reproducibility issues, and match researchers with complementary expertise.
Three Futures for DeSci x Biotech
The path forward in DeSci is plural, each addressing different aspects of the broken system:
At the core is a digital Schelling point for science—an onchain School of Athens—where the world's brightest minds gather virtually to tackle grand challenges. Funding flows to ideas through DeFi mechanisms, preventing plutocracy and encouraging participation.
Engagement, curiosity, and joy are sustained in this community via collaborative gamification, perhaps a kind of “prediction republic”—a massive multiplayer game where scientists bet on which research will replicate, which drugs will succeed in trials, which theories will stand the test of time. Their predictions are progressively refined by training models based on the results, improving the fidelity of science.
Value creation emerges from a venture studio model—onchain biotech companies with wet lab work outsourced to the best candidate. A group identifies a promising drug target through community discussion. When the drug succeeds, value flows back to those who participated.
The Choice Before Us
The Sapien Open Science Fund signals commitment to making DeSci V2¹³ a reality, creating structures where everyone—researchers, patients, and funders—shares in scientific success. This isn't just a commitment of capital; it's building a new research ecosystem where aligned incentives replace competing interests.
Science deserves better than its current state. DeSci V2 might be the catalyst that sparks a renaissance. The School of Athens wasn’t a building; it was an idea. The idea that human knowledge advances through open discourse, truth emerges from debate, and wisdom belongs to all rather than a few. DeSci V2 resurrects this idea for the digital age.
Join us in building the future of science. The future of humanity depends on it.
Eleanor Davies, Head of DeSci at Sei Development Foundation
Eleanor is recognized as a leading contributor to the DeSci movement. She leads DeSci at the Sei Development Foundation. In prior roles, she led investments at key projects such as VitaDAO and Convexity Labs.