AstraZeneca × Tempus AI
AstraZeneca, Tempus, and Pathos AI entered a multi-year agreement to develop a multimodal foundation model for oncology. [37, 48] Tempus will provide its large library of de-identified oncology data and will receive $200 million in data licensing and model development fees. [43] All three parties will be able to use the completed model for their own R&D to identify novel drug targets and accelerate therapeutic development. [43, 55]
The asset
A multi-year data licensing agreement for access to Tempus's vast repository of de-identified, multimodal oncology data to co-develop a large-scale deep learning model for cancer drug discovery. [43, 48]
- 300 petabytes of data
The record
- Seller status
- Going concern
- Sector
- AI and precision medicine
- Modalities
- clinical_data, genomic_data, transcriptomic_data, radiology_images, pathology_images
- Personal data
- The data provided by Tempus is described as a de-identified or anonymized library of clinical and molecular data. [43, 55]
What it means if you hold data
This transaction demonstrates the high value placed on large, multimodal real-world datasets for training foundational AI models in drug discovery. It establishes a benchmark for data licensing fees in the era of generative AI and highlights a collaborative model where data, pharma, and AI-biotech firms partner to build shared resources.