Boston, Massachusetts, March 16, 2026
Manifold Bio, a platform therapeutics company focused on AI-driven drug discovery, announced breakthrough results from a joint study with NVIDIA demonstrating million-scale experimental validation of AI-designed protein binders. The study evaluated 1 million protein binder designs across 127 biological targets, generating over 100 million protein-protein interaction measurements, establishing a new benchmark for scalable validation of generative AI models in biopharmaceutical research.
AI-Driven Protein Design Achieves High Validation Success
The study validated Proteina-Complexa, NVIDIA’s generative AI model for protein binder design, using Manifold Bio’s proprietary high-throughput experimental platform. The results demonstrated that functional binders were successfully identified for 68% of tested targets, confirming the model’s capability to generate biologically relevant protein interactions at scale.
This achievement highlights the power of combining de novo AI-driven molecular design with real-world experimental validation, enabling researchers to move beyond theoretical predictions toward practical therapeutic discovery applications. The ability to test such a large number of designs simultaneously represents a major advancement in drug discovery efficiency, where traditional approaches are often limited by low throughput and high costs.
Massively Multiplexed Platform Enables Breakthrough Throughput
Advertisement
Manifold Bio’s platform leverages multiplexed molecular synthesis and measurement technologies, allowing simultaneous evaluation of millions of protein interactions within a single experiment. This approach enables scaling laws in protein design, where increasing the number of generated candidates directly improves the likelihood of identifying effective binders. By matching AI inference speed with experimental throughput, the study demonstrates a new paradigm in biopharmaceutical R&D, where computational predictions and laboratory validation operate at comparable scales.
The platform also integrates in vivo data generation, providing physiologically relevant insights into protein behavior, which is critical for advancing candidates toward clinical development. This capability significantly enhances the reliability of AI-designed therapeutics.
Accelerating Future Therapeutics Through AI and Data Integration
The study also generated thousands of new annotated protein structures, creating valuable datasets for future AI model training and continuous improvement. By combining generative AI models with large-scale experimental feedback loops, Manifold Bio is establishing a self-reinforcing discovery engine that accelerates the identification of novel therapeutic candidates. This approach supports the development of next-generation biologics, including antibodies and protein-based drugs, and has the potential to transform how pharmaceutical companies discover and optimize treatments.
As the industry increasingly adopts AI-driven drug discovery platforms, the ability to validate designs at scale will be critical for translating computational innovation into clinical and commercial success. The collaboration between Manifold Bio and NVIDIA demonstrates how advanced computing, AI, and experimental biology can converge to redefine the future of precision medicine and therapeutic development.
Source: Manifold Bio press release
