Google unveiled an advanced foundation model in the Cell2Sentence‑Scale family that proposed and (per company reports) experimentally validated a novel drug combination for cancer. Built on the Gemma model lineage and a 27‑billion‑parameter architecture, the system generated candidate combinations and supported preclinical validation workflows. The announcement marks a high‑profile example of large models moving from hypothesis generation to tangible experimental follow‑up. Google described the work as a milestone for AI‑driven biomedical discovery but did not provide full peer‑reviewed datasets in the initial release. While promising, industry scientists caution that AI‑generated leads require extensive preclinical and translational vetting; nonetheless, the result underscores accelerating industry investment in foundation models for drug discovery.