Top artificial intelligence researchers and venture capital are increasingly migrating toward life sciences and drug discovery, marking a structural shift in the AI-for-Science sector. Recently, AI-native biotech firms have secured substantial funding ahead of clinical milestones. Chai Discovery, co-founded by former OpenAI researcher Josh Meier, closed a $400 million Series C round at a $3.8 billion valuation on July 14, focusing on generative AI for protein and antibody design. Simultaneously, former OpenAI scientist Miles Wang is launching a new AI drug discovery venture with approximately $200 million in talks, while Xaira Therapeutics and EvolutionaryScale have mobilized over $1 billion and $142 million respectively, leveraging breakthroughs from Nobel laureate David Baker and former Meta protein-language model teams. Parallel moves by Chinese startups underscore a global talent reallocation from general AI to biomedical applications. This migration stems from converging economic and scientific factors. Training frontier foundation models now requires hundreds of millions of dollars, creating insurmountable capital barriers outside of major tech firms. Consequently, researchers are pivoting toward applied science to achieve measurable real-world impact. Life sciences emerged as the primary destination due to the direct transferability of sequence modeling, self-supervised learning, and reinforcement learning techniques to biological problems. Furthermore, the pharmaceutical industry high R&D costs and low clinical success rates make it highly receptive to AI-driven risk reduction. The scientific precedent for this approach was established by AlphaFold success, which validated AI utility in biology years before the generative AI boom. Unlike first-generation AI drug discovery firms that treated machine learning as a specialized tool for target screening and molecular optimization, the new wave places foundation models and reinforcement learning at the organizational core. Investors are front-loading capital, betting on platform potential rather than immediate clinical results. However, the sector still confronts historical bottlenecks, particularly the lack of comprehensive, unbiased biological data. Publication bias, experimental batch effects, and proprietary clinical data restrict model training. To overcome this, leading companies are implementing dry-wet laboratory closed loops, where automated experiments continuously generate structured data to iteratively refine algorithms. Commercial strategies are diverging into platform and full-stack models. Platform companies like Chai and Cradle focus on providing molecular design software and partnering with established pharmaceutical firms for clinical development and commercialization. Conversely, full-stack ventures such as Xaira and Isomorphic Labs are integrating in-house wet labs and drug pipelines to capture downstream asset value, bearing higher clinical risk in exchange for potential milestone payments and licensing revenue. The trajectory of this sector will ultimately depend on whether these integrated systems can reliably produce clinically validated therapeutics, transforming early technological promise into sustainable biomedical innovation.