Big Pharma is no longer just experimenting with AI; it is embedding massive neural networks directly into the core of drug discovery.
The era of the small-scale AI pilot is over. AbbVie’s new partnership with Iambic Therapeutics proves that pharmaceutical giants are now betting on massive, specialized models to solve their toughest chemistry problems.
At the center of this deal is Enchant v3, a newly launched 41-billion-parameter multimodal model trained on over 6,000 molecular properties. This is not a generic language model adapted for science. It is a highly specialized system built to predict how molecules behave in the human body.
The Scale of the Bet
For Iambic, this marks its third major pharmaceutical partnership of the year, following deals with Takeda and Bayer. The San Diego-based startup is rapidly consolidating its position as a preferred AI engine for the industry.
Pharma companies are looking for shortcuts in immunology, neuroscience, and oncology. They are turning to multimodal transformers because traditional physical screening is too slow and too expensive. By integrating these models, they hope to shrink the timeline between target identification and clinical trials.
The Reality Check
But massive parameters do not guarantee clinical success. Training a model on thousands of molecular properties is impressive on paper, but biology remains notoriously unpredictable.
The startup’s financial runway, backed by a 100 million dollar venture round and recent debt financing, gives it the resources to refine these models. However, the pressure is now on to deliver actual molecules that work in living organisms, not just on screens.
The real test for Enchant v3 will not be the speed of generating drug candidates. It will be whether those candidates can survive the brutal transition from digital simulation to human clinical trials.
