← Back to Home

Competition in AI Drug Discovery Extends to Partnership Networks: Strategy Survey of 25 Companies Maps Industry Roles

Pharmaceutical companies integrate AI into R&D workflows, AI startups build platforms and drug candidates, and technology giants supply models and computing power. A Japanese strategy survey describes how these three groups connect their resources, while leaving a key question: can these partnerships translate into verifiable drug development outcomes?

By SURL BioNews

A promising molecule proposed by a model still faces many hurdles before becoming a drug. Who can provide disease data, arrange experiments, and then advance candidate molecules into human studies is shaping the industry landscape for AI drug discovery. A strategy survey released in August by Japan’s TPC Marketing Research indicates that partnerships among pharmaceutical companies, AI startups, and large technology companies are gradually becoming an important way to integrate technology, data, and drug pipelines.

The “2026 AI Drug Discovery Business Strategy Survey” was published on August 24, and TPC issued an announcement on August 26. International Business Times also reported its key findings. According to the publicly available report page, the survey ran from May to August 2026. The 121-page report covers 11 pharmaceutical companies, 10 AI startups, and four technology companies, including AstraZeneca, Recursion, Insilico Medicine, NVIDIA, Amazon, Google, and Microsoft. These figures represent the range of companies covered by the research and cannot be treated as an industry-wide survey of adoption rates.

The applications described by TPC have expanded from identifying disease targets and screening compounds to molecular design, ADMET prediction, and preclinical and clinical development. ADMET refers to a drug’s absorption, distribution, metabolism, excretion, and toxicity, concerning how a candidate molecule moves through the body, how it is processed, and what harm it might cause. If models can help predict these properties, they may help researchers prioritize validation work; however, the predictions themselves still require experimental testing. The survey also explicitly excludes general administrative uses, such as reducing data entry, focusing on work directly related to drug discovery and development.

The three types of companies focus on different parts of the process. According to TPC’s announcement, pharmaceutical companies emphasize integrating AI into existing R&D workflows; AI startups both build platforms and create and develop drug candidates; and technology companies provide models, cloud services, GPUs, and high-performance computing infrastructure. TPC considers that pharmaceutical companies obtain external technology, data, and computing power through partnerships, making competition increasingly a matter of how entire sets of resources are combined. From this perspective, whether partnership networks can connect computational results with actual R&D work offers a clue to understanding the industry landscape.

However, the public materials present an outline of strategies. TPC notes that drug candidates created or designed with AI involvement have entered clinical development, but its summary does not provide trial results that can be checked individually, model validation datasets, or comparative data on R&D costs and success rates. The survey can therefore support the description that AI applications are expanding, but it is not yet sufficient to demonstrate that AI has broadly improved new drug R&D outcomes. Entry into human studies also does not mean that efficacy and safety have been established.

Background Context

The discussion of AI drug discovery is changing as drug candidates move toward clinical development: researchers must not only identify molecules but also establish traceable, reproducible evidence. The survey also includes AI and machine learning guidance from Japan’s PMDA, the United States’ FDA, and Europe’s EMA within its scope of analysis, indicating that regulation has entered companies’ strategic considerations. However, the public table of contents does not provide enough detail to compare the agencies’ requirements, nor can it be interpreted as indicating that any platform has received regulatory recognition.

The survey offers a view of how the division of work behind AI drug discovery is taking shape. Connecting data, models, computing power, and drug pipelines through partnerships may help research ideas move more smoothly into validation. Whether these partnerships actually shorten development timelines and reduce failures still requires experimental and clinical data from specific programs. Industry networks are taking shape, while outcomes must be measured for each individual drug candidate.

References

  1. International Business Times
  2. TPC Marketing Research / TPC Bibliotheque
  3. Mapion News / DreamNews — TPC Marketing Research company release