人工智能如何改写药物开发经济学_45页_2mb

Summary of "From Code to Clinic: How AI Is (and Isn’t) Rewriting the Life of a Drug"

Core Content

This report explores the evolving role of artificial intelligence (AI) in the drug discovery process, analyzing how AI is reshaping but not replacing the physical (wet lab) work that remains central to pharmaceutical research. The focus is on the antibody drug discovery workflow, the impact of AI on different stages, and the implications for the life science tools and services industry.

Main Points

• AI is not replacing wet lab work, but rather reshaping it. AI serves as a "tool in the tool kit" for hypothesis generation and idea creation, but the validation, characterization, and testing phases still require physical experimentation.
• AI's role is primarily in the design and learning phases of the drug discovery cycle (D-L), which in turn increases the demand for build and test phases (B-T). This creates a net tailwind for the life science tools ecosystem.
• The most significant economic opportunity for AI lies in reducing clinical failure rates, particularly in Phase II trials, rather than simply lowering the cost of generating hypotheses.
• High-quality experimental data is critical for AI models to perform effectively. The report emphasizes that AI's impact depends heavily on the availability of robust, standardized data from wet lab experiments.
• The use of AI in drug discovery is at an inflection point, with increasing integration into the process, leading to more data points and a greater need for tools that support the validation and testing phases.
• Investor sentiment has been mixed, with some fearing a decline in wet lab utilization. However, the report argues that AI adoption is not likely to significantly reduce overall R&D spend, as historical innovations (e.g., NGS) have also had limited impact on total R&D costs.
• The life science tools industry is not in decline, and most companies face low to modest residual disruption risk from AI, with a notable exception in the hypothesis creation phase.

Key Information

AI's Role in Drug Discovery

• Hypothesis generation is where AI is making the most impact, but this is not the most cost-intensive part of the drug discovery process.
• AI is not a substitute for wet lab validation; rather, it enhances the efficiency of the entire process, especially when combined with high-quality data.
• The "dry lab" (computational) and "wet lab" (experimental) are both essential, and AI is not causing the obsolescence of wet lab work.

Impact on the Drug Discovery Workflow

Investment Outlook

• The life science tools sector is a net beneficiary of AI adoption due to the increased demand for validation and testing tools.
• Alpha is available both if AI becomes a dominant tool in discovery and if it fails to do so, as the physical validation steps remain essential.
• Investors should consider adding exposure to the sector as end-markets recover and growth accelerates in 2026.

Ecosystem Overview

• The drug discovery ecosystem consists of two core layers:
  1. Companies generating biological insights and turning them into therapeutics.
  2. Companies providing tools and services that enable these discoveries.
• Collaborations are increasing, with major partnerships between pharma, biotech, and AI firms. These include:
  • Eli Lilly with InSilico Medicine, NVIDIA, Nimbus Therapeutics
  • Roche with NVIDIA
  • Takeda with Iambic Therapeutics, Nabla Bio
  • Sanofi with CytoReason, Formation Bio, BioMap
  • AstraZeneca with CSPC Pharmaceutical Group, Absci
  • Bristol Myers Squibb with Insitro
  • Pfizer with PostEra
  • Novartis with Generate Biomedicines, Isomorphic Labs
  • GSK with Noetik
  • Bayer AG with Cradle
  • Amgen with PostEra
  • Abbvie with BigHat Biosciences

Public Company Coverage

• High exposure to preclinical drug discovery is found in:
  • Bio-Techne (~60% of revenue, 25% early discovery, 35% translational)
  • Revvity (~50%)
  • Twist Bioscience (~45%)
  • 10x Genomics (~100%)
• Low to no AI disruption risk is associated with:
  • Pharma packaging companies (e.g., West, Stevanato, Aptar)
  • Bioprocessing companies (e.g., Repligen, Mettler-Toledo)
  • Downstream QC-focused companies (e.g., Waters, Agilent)
• Key winners include:
  • Twist Bioscience: Benefiting from increased demand for synthetic DNA in AI-driven antibody discovery.
  • 10x Genomics: Playing a critical role in enabling AI to create and adapt biological models.

Conclusion

AI is transforming the drug discovery process by enhancing the efficiency of hypothesis creation and accelerating the design-build-test-learn (DBTL) cycle. However, the validation and characterization steps—which are the most expensive and time-consuming—remain physically intensive and not easily automated. As a result, the life science tools industry is not at risk of obsolescence but rather of increased demand for its services. The report suggests that AI is a tailwind, not a disruption, for most companies in the sector, and that the economic value of AI lies in improving the quality of drug candidates entering clinical trials.