Big Pharma is shifting its AI strategy from targeting specific diseases to funding the underlying operating system of human biology.

For years, artificial intelligence in drug discovery was treated like a sniper rifle. Pharma giants hired AI startups to find single molecules for specific diseases. But GSK’s expanded $110 million pact with Relation Therapeutics signals a structural pivot. The industry is moving toward “foundation models of the cell.”

The Data Bottleneck

The collaboration centers on MORGAN, a platform designed to predict how cells react to genetic and drug interventions. But algorithms are useless without clean data. In biology, high-quality training data is scarce and messy. Rather than scraping public databases, Relation is building automated wet labs to manufacture its own biological data at scale.

This is not an isolated experiment. It follows Relation’s massive $1.7 billion drug discovery pact with Novartis in late 2025. Pharma is realizing that buying pre-packaged AI tools is not enough. They must own the data generation pipeline.

Unproven Biological Maps

The financial stakes are soaring, but the biological reality remains highly uncertain. Predicting cellular behavior is infinitely more complex than predicting the next word in a text document. Cells are dynamic, chaotic, and poorly understood. If these foundation models fail to translate into actual clinical candidates, the industry will have spent billions mapping a territory it still cannot navigate.