Key Highlights
- VivaTech 2025 panel revealed that biopharma’s biggest AI hurdle isn’t compute power but access to AI-native, interoperable data from 20 exabytes of fragmented sources.
- Executives from NVIDIA, Sanofi, and TetraScience called for global data-sharing frameworks, sovereign AI strategies, and collaborative “science factories.”
- TetraScience reports 80% reduction in lead clone selection time, showcasing early but impactful gains in AI-powered drug development workflows.
The 20-Exabyte Problem: Data, Not Models, Is the Limiting Factor
At The Rise of Scientific AI panel during VivaTech 2025, top leaders from TetraScience, NVIDIA, and Sanofi agreed that the greatest obstacle in biopharma isn’t the AI models—it’s the fragmented, vendor-locked data. TetraScience CEO Patrick Grady emphasized that “raw data trapped in proprietary formats has zero utility for AI.” Sanofi’s Emmanuel Frenehard added that anonymized, longitudinal patient data is essential for building predictive “digital twins,” while NVIDIA’s Rory Kelleher contrasted the open data landscape in AI language models with the current biological data bottleneck.
Forging Alliances: AI Progress Demands Data Collaboration, Not Competition
The panel stressed that AI progress in biopharma cannot happen in silos. NVIDIA’s Kelleher pointed to partnerships with TetraScience and Sanofi as key to building scalable, AI-ready datasets. He called it a “moral imperative” to address the declining ROI in drug development. Frenehard advocated for cross-border data-sharing protocols, and Grady positioned TetraScience’s Scientific Data and AI Cloud as a critical bridge between lab complexity and AI readiness—augmenting rather than replacing scientific expertise.
From Vision to Validation: AI Is Delivering Results Today
While the panel explored ambitious visions such as NVIDIA’s AI-powered “science factories,” the discussion remained grounded in present-day achievements. Sanofi emphasized the importance of cautious optimism given biology’s inherent complexity. Meanwhile, Grady shared real-world examples where TetraScience helped customers cut lead clone selection time by 80%, proving that even in today’s workflows, AI can deliver meaningful efficiencies.
Balancing Sovereignty with Science: A Unified Approach to AI in Healthcare
The session concluded with a powerful exchange on AI and data sovereignty. Kelleher advocated for national strategies to safeguard patient data, while Frenehard warned that excessive regulation could stifle innovation. Grady struck a middle path: “Sovereignty isn’t about isolation—it’s about enabling nations to securely contribute to a federated, global intelligence for the benefit of all.”
For more information, visit TetraScience, NVIDIA Healthcare, and Sanofi.
By