For the first time, an artificial intelligence-discovered drug has produced clinical evidence that it works in human patients, marking a watershed moment for a field that has consumed over $11 billion in investment without a single approved medicine to show for it. The question now is whether this breakthrough signals a genuine transformation in how we discover drugs, or whether it represents an isolated win in a sector that may be overheating with capital.

The proof point is rentosertib, an AI-discovered and AI-designed molecule developed by Insilico Medicine for idiopathic pulmonary fibrosis. In a Phase IIa trial, the drug improved lung function by an average of 98.4 milliliters of forced vital capacity at a 60-milligram daily dose, compared to a decline of 20.3 milliliters in the placebo group. The results were published in Nature Medicine, a top-tier peer-reviewed journal, giving the field its first genuine end-to-end validation that artificial intelligence can identify and design molecules that actually help patients.

Yet this single success sits against a sobering backdrop. Across approximately 348 funding rounds in 2025, the AI drug discovery sector attracted $11 billion in capital. In any other industry, that level of investment with zero approved products would trigger serious questions about whether the market is pricing in a realistic timeline. In drug development, where clinical trials routinely span a decade or more, the calculus is different. The field is essentially betting that the technology will prove itself over the next 24 months, when a wave of clinical readouts will either validate the approach or expose it as overhyped.

What's Actually Changed in AI Drug Discovery?

The shift happening right now is less about revolutionary breakthroughs and more about a fundamental change in how the industry operates. Three major trends are reshaping the landscape:

• From Single Molecules to Platforms: Pharmaceutical companies have stopped buying individual drug candidates and started acquiring entire AI platforms, signing multi-target collaborations and gaining access to proprietary models and infrastructure rather than betting on one molecule at a time.
• From Narrative to Evidence: The conversation has shifted from theoretical promises about AI transforming drug discovery to actual peer-reviewed clinical data, making each new readout disproportionately important to the field's credibility.
• From Western Dominance to Global Competition: The first major proof point came from a China-rooted company, and Hong Kong has emerged as a primary listing venue for AI-native biotech, signaling a geographic shift in where innovation and capital are concentrating.

The pipeline itself is real and growing. AI-originated drug programs entering clinical trials have expanded from roughly 3 in 2016 to 67 by 2023 to more than 200 by early 2026. That compounding growth suggests the field is producing candidates at an accelerating pace. The open question is whether those candidates will convert into approved drugs at rates that justify the investment.

How Are Major Pharmaceutical Companies Responding?

Big pharma's behavior offers a clearer signal than any press release. The largest pharmaceutical companies are not waiting passively for AI startups to deliver finished drugs. Instead, they are aggressively acquiring capability and platform access, signaling confidence that the technology is real enough to bet on.

• Eli Lilly's Multi-Front Strategy: Lilly has become the most aggressive buyer in the space, partnering with Isomorphic Labs (which has generated over $1.7 billion in milestone payments from Lilly alone), acquiring generative AI capability through Chai Discovery, and backing insitro's machine learning and functional genomics platform alongside other top-ten pharmaceutical companies.
• Repeat Partnerships as Validation: Isomorphic Labs has signed partnerships with three major pharmaceutical companies: Lilly, Novartis (approximately $1.2 billion in milestones), and Johnson & Johnson, demonstrating that renewals and repeat customers are the strongest validation signal in the field.
• New Deal Structures Emerging: Noetik signed a five-year licensing partnership with GlaxoSmithKline featuring a $50 million upfront payment and a subscription-based framework for non-small cell lung cancer and colorectal cancer models, introducing a software-like deal structure to AI drug access.
• Platform Access Over Single Assets: Recursion partnered with Roche and Genentech on a $150 million-plus antibody-discovery collaboration, while maintaining ongoing partnerships with Bayer, showing that established pharmaceutical companies are paying for repeated shots on goal from validated engines rather than betting on individual molecules.

The pattern is unmistakable: when multiple independent top-ten pharmaceutical companies back the same AI platform, it signals that the science is real and the business case is compelling. This kind of third-party validation carries more weight than any single funding round.

Why Is the Clinical Evidence Still So Sparse?

One readout does not a revolution make. The honest assessment is that the evidence base supporting AI drug discovery remains thin. Rentosertib's success is genuinely important, but it represents a handful of programs, not a body of literature. To properly evaluate clinical claims in this space, experts recommend distinguishing between three different levels of AI involvement.

An AI-discovered drug means the artificial intelligence identified the target or the initial hit compound. An AI-designed drug means the AI generated the actual molecule that went into patients. An AI-assisted program means AI accelerated an otherwise conventional drug development process. Only the strongest claims, like rentosertib, represent genuine end-to-end AI involvement from target identification through molecular design. When evaluating any "AI drug breakthrough" headline, the endpoint matters too: a hard clinical endpoint like lung function beats a biomarker, and a large Phase III trial carries more weight than a small Phase IIa result.

Recursion Pharmaceuticals offers a model for disciplined pipeline management. After acquiring Exscientia, Recursion guided investors to expect approximately 10 readouts across an 18-month window, including a Phase II ALDER readout around the first quarter of 2026, while simultaneously cutting weaker programs in 2025. This combination of scale plus discipline is healthier than claiming everything works, and it sets a standard for how the field should communicate about its pipeline.

How Are AI-Native Biotech Companies Accessing Public Markets?

The capital markets have reopened for AI-native biotech companies, a sign that investors believe the field has moved beyond pure speculation. Generate Biomedicines and Eikon Therapeutics both completed initial public offerings in early 2026, raising approximately $400 million and $381 million respectively. Insilico Medicine, the company behind rentosertib, listed on the Hong Kong Exchanges and Clearing in December 2025 at approximately $293 million, while XtalPi raised approximately $785 million on the same exchange.

The emergence of Hong Kong as a primary venue for AI biotech capital is significant. It reflects both the geographic shift in innovation and the role that China-rooted companies are playing in validating the technology. Mega-rounds persist across the sector, with Xaira launching with over $1 billion, Isomorphic raising $600 million in 2025, and Iambic, Genesis, and Chai each adding nine-figure rounds. Capital is not the constraint in this field; validation is.

What Should You Watch Over the Next 24 Months?

The next two years will determine whether AI drug discovery is a genuine technology shift or an investment bubble. The field has produced its first real signal: a peer-reviewed clinical readout, a cluster of billion-dollar pharma partnerships, and a reopened public-market window. What it has not yet produced is a pattern of approvals.

The defining tension of the moment is the gap between capital and clinical output. More than 200 AI-originated drug programs are now in the clinic, and the pipeline is compounding. The conversion of those programs into approved drugs will determine which interpretation of the current moment proves correct: whether $11 billion of annual investment represents a sector pricing in a genuine option on transformative technology, or whether it represents capital chasing a narrative that has not yet delivered at scale.

" }