Big Pharma Pools Data to Accelerate AI-Driven Drug Discovery

Leading pharmaceutical companies are increasingly collaborating to harness the power of artificial intelligence (AI) in the search for new medicines. This collaborative push, involving data pooling and the development of advanced AI models, aims to speed up drug discovery and improve treatment options for patients worldwide.

The Rise of Federated AI in Drug Development

Traditionally, drug discovery has been a lengthy and expensive process. AI offers the potential to significantly accelerate this process by identifying promising drug candidates and predicting their efficacy. Yet, the success of AI models relies heavily on the availability of large, diverse datasets. Recognizing this, several major pharmaceutical companies are now pooling their resources to create more robust and effective AI tools.

Key Collaborations and Initiatives

A significant initiative is the Federated OpenFold3 Initiative, launched in March 2025 with the support of AbbVie and Johnson & Johnson. This program has recently expanded to include Astex Pharmaceuticals, Bristol Myers Squibb (BMS), and Takeda Pharmaceutical. The Columbia University lab of Mohammed AlQuraishi, Ph.D., is developing OpenFold3.

The core of this collaboration involves contributing data from thousands of experimentally determined protein-slight molecule structures. This data is shared using a federated computing platform developed by Apheris, which allows companies to collaborate without physically transferring or exposing sensitive data. This approach addresses critical data privacy and security concerns.

Benefits of Data Pooling

Pooling data from multiple sources offers several advantages:

• Larger Datasets: A larger dataset leads to more accurate and reliable AI models.
• Increased Diversity: Diverse datasets improve the generalizability of AI models, making them applicable to a wider range of patients and diseases.
• Enhanced Predictive Power: More comprehensive data allows AI models to better predict the binding affinities of small molecule-protein and antibody-antigen interactions, crucial for drug development.
• Protection of Proprietary Information: Federated learning ensures that each company retains control over its own data while still contributing to the collective effort.

Company Perspectives

Hans Bitter, Ph.D., head of computational science and data strategy at Takeda, highlighted the potential of this initiative, stating that the prediction tool could be “transformative in how we discover and develop small molecule therapeutics.” Paul Mortenson, Ph.D., senior director of computational chemistry and informatics at Astex, emphasized that the system will improve medicinal chemistry models “while keeping proprietary science protected.”

Recent Regulatory Scrutiny of Pharmaceutical Marketing

While the industry focuses on innovation, regulatory oversight remains a key factor. In September 2025, the FDA initiated a crackdown on potentially misleading or deceptive direct-to-consumer pharmaceutical advertising, issuing numerous letters to major drugmakers, including AstraZeneca, Bristol Myers Squibb, Novartis, and Takeda. FDA Targets Big Pharma in Deluge of Marketing Letters

Looking Ahead

The increasing collaboration between pharmaceutical companies and the adoption of federated AI approaches represent a significant shift in the drug discovery landscape. As these initiatives mature and generate results, we can expect to see a faster and more efficient development of new and innovative therapies. The success of OpenFold3 and similar projects will likely pave the way for further data-sharing collaborations and the widespread adoption of AI in the pharmaceutical industry.