AI-Discovered Drugs Reach 173 Clinical Programs as Pivotal Trials Approach

# AI-Discovered Drugs Reach 173 Clinical Programs as Pivotal Trials Approach

The pharmaceutical industry has spent decades promising that artificial intelligence would revolutionize how medicines are made. In 2026, the numbers are finally catching up to the narrative -- and the next 12 months will determine whether AI-designed drugs can clear the highest bar in medicine: regulatory approval.

More than 173 AI-discovered drug programs are now in clinical development worldwide, according to recent industry analyses. Roughly 94 sit in Phase I safety trials, 56 have advanced to Phase II efficacy studies, and 15 have reached Phase III -- the large-scale pivotal trials that regulators require before granting approval. Between 15 and 20 additional programs are expected to enter pivotal trials before the end of the year, creating a wave of late-stage data that could reshape the industry's calculus on AI's value in the lab.

No AI-discovered drug has yet received FDA approval. But the pipeline is approaching critical mass, and the first approval is now projected to arrive in 2026 or 2027 with roughly 60 percent probability, according to independent analysts.

Zasocitinib Leads the Pack

The most advanced AI-linked candidate is zasocitinib, also known as TAK-279, a once-daily oral TYK2 inhibitor for plaque psoriasis. The molecule was originally designed through a computational chemistry collaboration between Schrodinger and Nimbus Therapeutics before being acquired by Takeda for $6 billion. In Phase III trials, approximately 70 percent of patients achieved clear or almost clear skin at week 16, with safety data consistent with earlier studies.

Chinwe Ukomadu, MD, PhD, senior vice president and head of Takeda's Gastrointestinal and Inflammation Therapeutic Area Unit, called the results a validation of the approach. "Our Phase 3 results demonstrate that highly selective TYK2 inhibition can offer many people with moderate-to-severe plaque psoriasis the potential for clear or nearly clear skin," Ukomadu said. "We are working as quickly as possible with regulators to advance a potential new therapeutic option for patients seeking a safe, effective and convenient oral treatment."

Takeda is on track to submit a New Drug Application to the FDA starting in fiscal year 2026, positioning zasocitinib as possibly the first AI-discovered drug to reach the market.

Insilico Medicine's End-to-End Test

If zasocitinib represents AI's role in molecular optimization, Insilico Medicine's rentosertib -- formerly ISM001-055 -- represents something more ambitious. It is the first drug in which both the disease target and the molecular compound were identified using generative AI, making it a true end-to-end test case for the technology.

Rentosertib targets TNIK, a kinase implicated in idiopathic pulmonary fibrosis. Phase IIa results, published in Nature Medicine in 2025, showed that patients receiving 60 mg once daily experienced a mean improvement in lung function of +98.4 mL, compared to a mean decline of -20.3 mL in the placebo group -- a clinically meaningful separation. The drug also met its primary safety endpoint.

Alex Zhavoronkov, PhD, founder and CEO of Insilico Medicine, said the data point toward larger trials ahead. "These results not only suggest that rentosertib has a manageable safety and tolerability profile, but also warrants further investigation in larger-scale clinical trials of longer duration, demonstrating the transformative potential of AI in drug discovery and development," Zhavoronkov said.

The Efficiency Argument

Beyond individual molecules, the aggregate data are beginning to support AI's broader value proposition. AI-discovered molecules have demonstrated an 80 to 90 percent success rate in Phase I trials, compared to a historical industry average of roughly 52 percent. A Boston Consulting Group and Wellcome Trust analysis projected that applying AI in early-stage drug R&D could yield time and cost savings of 25 to 50 percent through the preclinical stage.

Those efficiency gains matter in an industry where bringing a single drug to market traditionally costs more than $2 billion and takes over a decade. If AI can meaningfully compress either timeline or cost, the economic implications are enormous.

Regulatory Framework Takes Shape

The FDA is also moving to formalize how it evaluates AI-generated evidence. A draft guidance released in January 2025, titled "Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products," introduced a risk-based credibility assessment framework. It requires drug sponsors to define the regulatory question their AI model addresses, assess model risk, and document outcomes -- creating a structured pathway for AI-driven submissions.

Startup Ecosystem Expands

The momentum is attracting fresh capital and institutional support. Nebius, the AI cloud infrastructure company, launched its second annual AI Discovery Awards on April 24, offering $100,000 in GPU cloud credits to the top drug discovery startup, with $50,000 and $30,000 for runners-up. Applications are open through April 30, and the program is open to startups from pre-seed through Series D. Last year's inaugural competition drew 257 applications from around the world.

What Comes Next

The central question for 2026 is straightforward: can AI-discovered drugs survive the brutal attrition of Phase III trials at rates meaningfully better than the industry's historical roughly 50 percent success rate at that stage? If the answer is yes -- even modestly -- AI will move from an optional efficiency tool to a structural requirement for competitive drug development. If the answer is no, the technology will face a credibility reckoning, regardless of how promising the early-phase data looked.

Either way, the industry will know soon. The data are coming.