Has the AI Drug Discovery Hype Train Derailed? A Reality Check for Pharma and Biotech
When smarter algorithms hit the wall of outdated biology
In 2021, venture capitalists poured over $2 billion into AI-powered drug discovery startups. The hope was clear: artificial intelligence could compress a 10-year, multi-billion-dollar R&D cycle into a few years and a few million dollars. Companies like Recursion, Exscientia, and Insilico Medicine emerged as poster children for this new era, promising a revolution that would change the trajectory of pharmaceutical research forever.
But fast forward to 2025, and the story looks much different.
Stock prices are down. Investors are wary. M&A activity is heating up—not because of success, but because of pressure to survive. What happened?
⸻
💡 AI Was a Tool, Not a Cure
AI drug discovery promised efficiency: faster molecule generation, better hit prediction, and streamlined clinical trial design. In some areas—like protein structure prediction (e.g., AlphaFold)—it delivered.
But the core promise of AI in pharma wasn’t just better predictions. It was better outcomes. And so far, those outcomes haven’t materialized. Despite billions invested, the number of AI-discovered drugs that have reached approval can be counted on one hand. Many are still years from even entering Phase II.
The problem isn’t AI. It’s the assumptions AI is built on.
⸻
🧬 Bad Biology, Amplified
AI learns from data. But if that data reflects flawed or outdated models of disease, AI simply amplifies the problem. Garbage in, garbage out—at scale.
Today’s drug discovery pipelines are still rooted in a symptom-first, single-target view of biology. This view has led to generations of “blockbuster” drugs that treat downstream effects while ignoring the upstream causes of illness—often hidden in immune dysfunction, systemic pathogen behavior, and tissue isolation that begins long before diagnosis.
Training AI on this framework doesn’t fix the problem. It accelerates the development of drugs that do exactly what previous ones did—often without solving the actual disease.
⸻
📉 Investors Are Waking Up
Insilico, which once boasted one of the first AI-designed drugs to enter clinical trials, has seen delayed timelines and increased scrutiny. Recursion’s lofty valuations have been trimmed by the realities of biology’s complexity. Even Big Pharma partnerships with AI startups are becoming more cautious, with milestone-based deals replacing blank-check enthusiasm.
This isn’t failure—it’s correction. But it reveals something important: the AI drug discovery revolution didn’t falter because AI failed. It faltered because we’re still solving the wrong problems.
⸻
🔄 RewriteBiology: A Different Approach
At RewriteBiology, we believe the real opportunity isn’t in designing smarter drugs for outdated models—it’s in rewriting the biological models themselves.
Our research suggests that many chronic diseases and aging-related illnesses are driven by a stealthy, progressive form of biological isolation caused by pathogens that evolve to escape the immune system over time. These processes begin far earlier in life than we currently recognize. By the time symptoms appear, the body has already lost communication between systems.
Instead of chasing targets that appear late in the game, we focus on identifying the root mechanisms that disconnect the body’s ability to self-regulate and heal. This changes the entire drug development equation.
The future isn’t about “AI versus biology”—it’s about empowering a new biology with the right tools, including AI, to finally cure disease at its root.
Learn more and join us in building that future:
