Foundation models, virtual cells, AI-designed molecules, self-driving labs, and the companies putting them to work.

Biology is becoming computable, and the consequences run the length of the field.

At the foundation, virtual cells and biological foundation models are replacing incomplete mental models with systems that predict rather than guess. Downstream, that predictive power is compressing the therapeutic pipeline: targets validated faster, molecules and proteins designed generatively, toxicity anticipated before the clinic, delivery engineered to reach the right tissue. Around it, labs are learning to run themselves, with autonomous hardware and AI agents closing the loop between hypothesis and evidence without waiting on human pace. And at the far end, where good candidates still fail, AI is reshaping trial design, patient stratification, and the real-world evidence that determines whether a therapy actually helps anyone.

This theme follows that arc from first principles to patient outcomes, and asks the question the field keeps circling: where is it real, and where is it still a promise?