In a stunning breakthrough, Google’s AI model ‘Cell2Sentence-Scale 27B (C2S-Scale)’ has uncovered a new pathway in cancer therapy, surprising both scientists and investors.

Built on the company’s Gemma open models, this 27-billion-parameter system learned to “read the language of cells.” During testing, it identified a rare compound, silmitasertib (CX-4945), as a potential key to turning “cold tumors”-those invisible to the immune system-into “hot” ones that can be targeted by treatment.

Lab experiments confirmed its accuracy - when silmitasertib was paired with low-dose interferon, it boosted antigen visibility by nearly 50% in human cell models.

Experts say this discovery could revolutionize cancer immunotherapy, proving that AI can go beyond data analysis to generate real biological discoveries.

Google and Yale scientists are now testing more predictions, hinting that the next big cure might come not from a lab, but from an algorithm.