The Money Keeps Flowing
Recursion Pharmaceuticals is having a good week. The AI-driven drug discovery company pulled in a $12.5 million milestone payment from its partnership with Rallybio, a signal that its pipeline isn't just generating headlines — it's hitting contractual benchmarks that trigger real dollars. New clinical candidates are also advancing within the Recursion fold, suggesting the company's machine-learning-first approach to identifying drug targets is continuing to churn out prospects at a pace traditional pharma would struggle to match.
For Recursion, the Rallybio milestone is more than a cash event. It's validation of a model: partner early, let AI accelerate the discovery and optimization phases, and let milestone structures reward progress rather than just outputs. In a capital-intensive industry where most drug candidates fail before they ever reach a patient, collecting milestone payments along the way is a meaningful way to fund the next wave of experiments.
But the Clinic Is a Different Animal
Timing couldn't be more pointed. Even as Recursion celebrates its pipeline momentum, a wave of expert commentary is pressing the broader AI drug discovery sector on its most uncomfortable question: can any of this actually work in humans?
The concern isn't with the technology's ability to find patterns in biological data — AI has proven remarkably good at that. The worry is the gap between computational promise and clinical reality. Drug discovery has always been littered with candidates that looked brilliant in a dish or a model and collapsed in a trial. Critics and researchers are now asking whether AI is accelerating the discovery of better drugs, or simply accelerating the discovery of more candidates — many of which will fail for the same biological reasons they always have.
The scrutiny is fair. Investment in AI-powered pharma has been extraordinary over the past several years, with billions flowing into companies promising to compress decade-long timelines and slash failure rates. But the clinical track record is still thin. Most AI-originated drug candidates are still in early-stage trials. The field hasn't yet produced a clear, unambiguous success story where AI made the decisive difference in getting a safe, effective drug to market.
Two Stories, One Tension
Taken together, today's developments capture the exact moment the AI pharma sector finds itself in: flush with capital and pipeline activity on one side, facing a credibility test on the other.
Recursion's milestone payment is a data point in the optimists' column. It means a partner believed enough in the science to structure a deal around it, and then saw enough progress to release funds. New clinical entries mean the company's discovery engine is still running. These are real-world signals, not just pitch-deck projections.
But the broader expert skepticism reflects something equally real: the industry has been here before with other transformative technologies that promised to reinvent drug development. Genomics, combinatorial chemistry, high-throughput screening — each reshaped parts of the process without eliminating the brutal attrition rate that defines pharmaceutical development. AI may be different in degree, but the clinical gauntlet doesn't change just because the upstream process got smarter.
What to Watch
The next 18 to 24 months will be telling. As AI-originated drug candidates move deeper into Phase II and Phase III trials, the sector will either begin accumulating the clinical evidence it needs to silence doubters, or face a reckoning about how much of the early promise was real versus hype. Recursion, with its expanding pipeline and partner relationships, is one of the companies best positioned to provide an early answer.
For now, the milestone check clears. The harder milestones — the ones measured in patient outcomes — are still ahead.