Artificial intelligence (AI) is already reshaping how scientists discover and develop novel drugs. From predicting protein structures to optimizing clinical trials, AI tools can analyze biomedical data at speeds and scales unimaginable a decade ago. Yet, as the White House’s “America’s AI Action Plan” recognizes, realizing AI’s full potential in drug development will require significant upgrades to the way science is conducted—and to the infrastructure that supports it.

Building Lab Infrastructure

The plan’s call to “invest in AI-enabled science” reflects a fundamental reality: even the most advanced AI models are only as useful as the scientific workflows they serve. In biopharma, AI can generate promising hypotheses—such as identifying a novel drug candidate for a rare cancer. But the bottleneck often occurs in the lab, where traditional experimentation is slow, manual, and resource-intensive.

Automated, cloud-enabled laboratories, like those the plan proposes, could change that. AI-driven high-throughput screening, for example, could enhance the efficiency, accuracy, and cost-effectiveness of drug screenings to significantly accelerate drug development.

Supporting Novel Scientific Organizations

The plan also urges support for Focused Research Organizations (FROs)—non-profit entities designed to tackle large, complex research challenges and produce public goods such as tools and datasets. Past examples include the Large Hadron Collider and the Human Genome Project. These were efforts far beyond the capacity of any single academic lab, company, or informal consortium, and not immediately lucrative enough for industry to undertake.

In biopharma, FROs could tackle pre-competitive challenges too big for one company. One current example, supported by Convergent Research, an incubator for FROs, aims to map the pharmome—identifying all unintended targets of approved small-molecule drugs. The public benefits include better safety pharmacology profiling, new drug repurposing opportunities, and better training data for AI drug development models.

Strengthening the Data Infrastructure

Another core priority in the plan is to “build world-class scientific datasets” while preserving privacy, recognizing high-quality data as a national strategic asset, with steps such as incentivizing researcher data sharing and establishing secure computing environments for restricted federal data.

Privacy-enhancing technologies (PETs)—such as differential privacy, federated learning, secure multi-party computation, and fully homomorphic encryption—can further enable large-scale data sharing and thus merit support. These tools allow sensitive data to be accessed, shared, and analyzed without exposing personal or proprietary information, using advanced mathematical and statistical principles to minimize the risk of re-identification or data leakage.

In AI-enabled drug discovery, PETs can help overcome legal and institutional barriers that often prevent pooling genomic, clinical, and pharmaceutical data. Public-private partnerships (PPPs) provide a framework for applying PETs effectively, particularly in the pre-competitive space. The Innovative Health Initiative (IHI)—the world’s largest biomedical PPP, launched in 2008 by the European Union and the European Federation of Pharmaceutical Industries and Associations (EFPIA)—offers a good example. Its MELLODDY project used federated learning to allow 10 pharmaceutical firms to jointly train an AI model for drug candidate screening while keeping their proprietary data confidential.

In biopharma, the quality, diversity, and accessibility of data determine AI’s predictive ability—whether modeling protein structures, simulating drug–target interactions, or identifying patient subgroups most likely to benefit from a therapy. The Protein Data Bank, for example, was a critical resource for AI-driven protein structure prediction through AlphaFold. Expanding such resources to cover more diverse data types, including imaging and electronic health records (EHRs), could open novel therapeutic opportunities.

Takeaways

AI can accelerate drug discovery and development—but without infrastructure, datasets, and organizational models, much of that potential will remain unrealized. The measures in the AI Action Plan’s “Invest in AI-Enabled Science” and “Build World-Class Scientific Datasets” priorities—funding automated cloud-based labs, supporting FROs, and building high-quality datasets—could shorten the time from bench to bedside and help preserve U.S. leadership in biopharmaceutical innovation.

Sustained public funding for basic science remains the critical gap. Many of America’s most transformative biomedical breakthroughs—from mRNA vaccines to CRISPR gene editing—emerged from decades of federally funded, high-risk research with no immediate commercial application. Continued investment in agencies such as the National Science Foundation and National Institutes of Health is essential to drive foundational AI applications in drug discovery, spur private sector investment, de-risk early-stage innovation, and ensure AI tools address public health needs. Without it, the United States risks ceding ground to rivals pairing AI ambitions with robust investment in fundamental research.