Asia-Pacific AI Drug Discovery Moves Beyond the Demonstration Phase: What Nine Major Platforms Have Actually Delivered Is More Than Model Scores

The easiest things to showcase in AI drug discovery are speed, model parameters, and enormous headline collaboration values. Yet the hardest question remains: What has it actually delivered? A review of nine drug discovery platforms in Asia-Pacific instead assessed whether candidates had entered human studies, whether collaboration payments had actually been made, whether results had undergone peer review, and whether third parties could reproduce and validate them, in an effort to bring the industry narrative back to verifiable outputs.

The review covered Insilico Medicine, XtalPi, METiS TechBio, BioMap, Nxera Pharma, PeptiDream, Standigm, Galux, and Hummingbird Bioscience. According to the report, five of the nine platforms have advanced platform-derived molecules into human trials, five have secured at least one publicly disclosed collaboration payment that was contracted or paid, and six have established a peer-reviewed record. However, only four have third-party evidence that can be described as independent reproduction.

One of the most closely watched examples of clinical progress is rentosertib, an idiopathic pulmonary fibrosis candidate for which Insilico Medicine used an AI workflow to propose both the target and chemical structure. The source article states that a Phase 2 study conducted in China with 71 participants showed a lung-capacity signal, and that the drug entered a Phase 3 registrational study in July 2026. However, the Phase 2 trial was designed primarily to assess safety, the efficacy estimate had a wide interval, and some participants discontinued treatment because of liver injury or abnormal liver function. Whether Phase 3 can reproduce the effect will be the crucial test of both the platform’s value and the drug’s risks.

The outputs delivered by the other platforms are not all the same kind of thing. METiS’s MTS-004 is a known therapeutic strategy that uses AI to improve formulation and has reportedly completed a Phase 3 trial, but it cannot be directly compared with a drug in which algorithms were involved from target identification through the creation of a new molecule. Nxera’s M4 receptor agonist has been advanced to Phase 3 by a partner, but its core approach is closer to structure-based drug design. PeptiDream, meanwhile, has accumulated substantial revenue through its peptide screening platform and multiple pharmaceutical partnerships, but does not itself claim that its clinical assets were designed by AI.

These differences reveal an issue easily obscured by promotional language: “AI drug discovery” may refer to target prioritization, molecule generation, formulation design, protein engineering, or even simply the addition of machine learning to traditional computational chemistry. Unless the point at which AI is involved is clarified first, ranking a Phase 3 formulation result alongside a novel protein design still undergoing laboratory validation has very limited scientific meaning.

Commercial figures likewise require closer examination. The review found that, in many collaborations, upfront payments actually received were far below the maximum total milestone values cited in headlines. Some platforms have recorded audited revenue, while others have only research collaborations with undisclosed terms. This does not mean the latter lack technical value, but it does show that “potential multibillion-dollar deals” cannot substitute for drug candidates, clinical data, or cash revenue.

The review still relies primarily on company announcements, exchange filings, and existing papers. Some papers, transaction terms, and comparative model results could not be verified, and it does not demonstrate that AI can improve the overall clinical success rate. Its more reliable conclusion is therefore appropriately restrained: AI drug discovery in Asia-Pacific is no longer limited to computer simulations, but the real tests still lie ahead—whether candidates can pass large controlled trials, safety reviews, and regulatory assessments, and ultimately become treatments that patients can use.