AI-Driven Drug Discovery in China — From Concept to Clinical Reality in 2026

I. Introduction: The Inflection Point

For years, AI-driven drug discovery was described as "promising" — a technology perpetually on the cusp of transforming pharmaceutical R&D but never quite delivering. In 2026, that changed. China now has both an approved AI-assisted original drug and an AI-discovered drug in Phase III clinical trials. The conceptual era has ended; the clinical era has begun.

II. Case Study 1: Xijiao Yistewei (艾普司韦) — China's First AI-Assisted Approved Drug

On September 3, 2026, the National Medical Products Administration (NMPA) granted conditional approval to Xijiao Yistewei tablets (盐酸伊司特韦片, brand name: 艾普司韦), developed jointly by Westlake University, Westlake Laboratory, and Westlake Pharmaceuticals. It is China's first original drug approved with AI-assisted R&D at its core.

The development timeline:

Metric	Detail
Discovery to clinical trial completion	3.5 years
Compound library screened	49 billion compounds
Development approach	AI-driven DEL library + computational prediction + experimental validation closed loop
Indication	Adult mild-to-moderate COVID-19 infection
Approval type	Conditional approval

The underlying platform utilized a DNA-encoded library (DEL) of 49 billion compounds, with AI algorithms predicting binding affinity and drug-like properties. This enabled researchers to identify lead candidates far faster than traditional high-throughput screening. Westlake University Professor Huang Jing, a key researcher on the project, noted that the AI approach fundamentally changed the economics of early-stage discovery.

The 3.5-year timeline from molecular discovery to clinical trial completion is notable. The industry's traditional benchmark — often cited as "10 years and $1 billion" — has been substantially compressed. While it would be premature to attribute the entire acceleration to AI, the AI-driven discovery phase clearly played a significant role in reducing the time spent on hit identification and lead optimization.

III. Case Study 2: Rentosertib — The First AI-Discovered Drug in Phase III

On September 9, 2026, Insilico Medicine announced the first patient dosing in GENESIS-IPF-3, a Phase III clinical trial of Rentosertib (ISM001-055) for idiopathic pulmonary fibrosis (IPF). This is the world's first Phase III trial of a generative AI-discovered drug.

Trial design:

Parameter	Detail
Trial name	GENESIS-IPF-3 (NCT07687459, CTR20262475)
Lead site	Peking Union Medical College Hospital
Design	Randomized, double-blind, placebo-controlled
Enrollment target	320 patients across 47 centers in China
Duration	52 weeks
Primary endpoint	Annual decline rate in forced vital capacity (FVC)
Mechanism	TNIK inhibition (targeting Wnt, TGF-β, and NF-κB signaling pathways)

The mechanistic rationale is notable: Rentosertib inhibits TNIK (TRAF2- and NCK-interacting kinase), a target that was itself identified through AI-driven analysis of multi-omics data. Both the target and the molecule were discovered using Insilico's generative AI platform.

Phase II data informing the Phase III design:

• In the 60 mg once-daily arm, patients demonstrated a mean FVC improvement of +98.4 mL at 12 weeks, compared with −20.3 mL in the placebo group.

In an exploratory proteomic analysis published in Nature Biotechnology in 2026, researchers applied six independently developed proteomic aging clocks to patient samples. Rentosertib-treated patients showed consistent reductions in predicted biological age, with the strongest signal observed at Week 4 in the 30 mg twice-daily group — approximately 3–4 years of biological age reversal on several clocks, and up to 6 years on one clock.

These anti-aging findings are exploratory and hypothesis-generating rather than confirmatory. However, they illustrate a broader trend: AI-discovered drugs may be associated with novel biological signatures that were not anticipated during the discovery phase.

IV. AI Pharmaceutical Financing in 2026: A Surge in Capital

The capital markets have taken notice. In the first half of 2026, global AI pharmaceutical financing reached $6.5 billion, a 42% increase year-over-year and a record for any comparable period. China accounted for approximately 28% of this total, making it the world's second-largest AI pharmaceutical investment market after the United States.

China biotech financing highlights (H1 2026):

• Total raised: $4.64 billion, exceeding the $4.58 billion raised in all of 2025 — a 269% surge from the $1.26 billion recorded in H1 2025.
• Within China's biotech financing landscape, AI drug discovery accounted for 25% of deal value, making it one of the top three most active segments alongside cell & gene therapy (30%) and ADC platforms (20%).
• Domestic biopharma financing in China reached $5.0 billion across 193 deals, with late-stage rounds (Series D and beyond) surging 200%.

The confluence of clinical validation and capital inflow suggests that AI drug discovery in China has entered a self-reinforcing cycle: successful clinical programs attract investment, which funds more ambitious programs, which generate more clinical data.

V. What AI Is Actually Changing — and What It Isn't

What AI has demonstrably accelerated:

• Hit identification and lead optimization: The 49-billion-compound DEL library screening for Xijiao Yistewei would have been computationally infeasible without AI.
• Target identification: Rentosertib's TNIK target emerged from AI-driven multi-omics analysis rather than traditional hypothesis-driven research.
• Timeline compression: The 3.5-year Xijiao Yistewei development cycle, while perhaps the fastest case, illustrates the upper bound of what is achievable.

What remains uncertain:

• Attrition rates: AI may improve early-stage success rates, but Phase II and III failure rates remain the primary driver of R&D costs. Long-term data on AI-discovered drugs' Phase II/III success rates are not yet available.
• Regulatory acceptance: While NMPA has approved an AI-assisted drug and accepted an AI-discovered drug into Phase III, the FDA and EMA have not yet issued comprehensive guidance on AI-generated evidence packages.
• Generalizability: Both case studies involve well-validated targets (viral protease for COVID-19, TNIK in fibrosis). Whether AI can reliably deliver first-in-class drugs against entirely novel targets remains to be seen.

VI. Conclusion

2026 will be remembered as the year AI drug discovery moved from PowerPoint to prescription pad — or at least to Phase III clinical trial. The approval of Xijiao Yistewei and the Phase III advancement of Rentosertib represent genuine milestones, not incremental progress. For international patients, the practical implication is clear: the therapies reaching Chinese clinics in the coming years will increasingly originate from AI-driven discovery platforms, potentially offering novel mechanisms of action and faster development timelines.