New physics‑grounded AI models and collaborations are accelerating early drug discovery and biological simulation. Researchers showcased Boltz‑2 and related large quantitative models that predict binding affinities rapidly, complementing structure‑prediction advances such as AlphaFold3. These models aim to cut reliance on resource‑intensive experimental screens. On infrastructure, NVIDIA expanded collaboration with the Chan Zuckerberg Initiative to scale the Virtual Cells platform and the Billion Cells Project, increasing compute and datasets available for community model development. The partnership ties GPU power to open biological models and large single‑cell datasets. Taken together, model accuracy and infrastructure scale are lowering barriers to computational lead identification and mechanistic simulation — a shift likely to accelerate preclinical target selection and compound triage across biotech.