Anthropic and ByteDance enter AI drug discovery, data challenges persist

Anthropic and ByteDance have entered the AI drug development sector, signaling a growing interest in leveraging artificial intelligence for pharmaceutical innovation. Anthropic is establishing a wet lab to test its models in biological experiments and has partnered with Novo Nordisk and Bristol-Myers Squibb, while also acquiring Coefficient Bio. ByteDance's newly formed lab has completed its first round of financing. AI is currently being used to accelerate early-stage drug development, reducing the search space for candidate molecules and experimental iteration time. However, the costly and time-consuming clinical trial phase remains challenging to shorten. Three distinct approaches are emerging among players in the field. AI-focused companies target specific problems within drug development, while large model companies aim for general capabilities, using a single base model with a vertical toolchain to create a scientific workbench. Traditional pharmaceutical firms possess vast data but lack a unified data foundation. Large model companies invest heavily in model training, prioritizing data from failed experiments, which is as valuable as successful samples. Commercialization paths are divided, with AI drug development companies typically combining AI-CRO, software tools, and their own pipelines. Self-developed pipelines are seen as the core value driver for these companies. The market awaits results from Insilico Medicine's Phase III trial, as real clinical data will ultimately determine the credibility of AI-driven drug discovery approaches.