AI and CRO disruption: why IQVIA, Medpace, and Charles River may become more valuable

TL;DR

CRO stocks have sold off hard since early February. Reuters says IQVIA, Medpace, and Charles River Laboratories all fell sharply after Anthropic’s advanced AI agents fueled fears that drugmakers could internalize more clinical-trial work. The market may be right that AI changes the economics of trials. It is wrong about where the disruption lands. TD Cowen estimates AI could compress a late-stage trial from 58 to 47 months. But the hardest part of running a trial is the busywork — patient recruitment across 200 sites in 40 countries, the 2 AM call when a Brazilian site loses its PI mid-enrollment, GCP compliance across 15 different EDC systems. CROs are infrastructure companies that Wall Street keeps pricing like software vendors. AI should make CROs more valuable, not less. The cost-plus model is what has to die. The question worth asking: which CRO will be first to price its contracts based on AI-enabled speed?

CRO stocks have sold off hard since early February. Reuters says IQVIA, Medpace, and Charles River Laboratories all fell sharply after Anthropic’s advanced AI agents fueled fears that drugmakers could internalize more clinical-trial work.

The market may be right that AI changes the economics of trials. It is wrong about where the disruption lands.

TD Cowen estimates AI could compress a late-stage trial from 58 months to 47. That’s real. Protocol design, site selection, data monitoring, signal detection. AI will reshape all of these (even if unfortunately much slower than many of us would like).

In any case, potentially the hardest part of running a clinical trial is the busywork of patient recruitment across 200 sites in 40 countries. Navigating regulatory submissions in jurisdictions with different safety reporting timelines. The 2 AM call when a site in Brazil loses its principal investigator mid-enrollment. Maintaining GCP compliance across a decentralized trial with 15 different electronic data capture systems.

CROs are infrastructure companies that Wall Street keeps pricing like software vendors.

And the real irony: AI should make CROs more valuable, not less. The firms that integrate AI into trial execution will compress timelines and capture the upside. TD Cowen already sees new contracting models emerging, including gain-share arrangements where CROs keep a portion of the time savings they deliver.

But unfortunately, the cost-plus model has served CROs well for decades, and it will be very painful to give it up (you know, old dogs, new tricks?). What is currently saving traditional CROs is mostly the old adage of “nobody ever got fired on hiring IBM”. If the future of your large phase III trial depends on you as a head of clin ops delivering, would you rather go with the approach that has served you well for decades, or try the new thing (and be the one to blame if the new technology is still having some hiccups?).

The question worth asking: which CRO will be first to price its contracts based on AI-enabled speed? Any bets?

Key takeaways

• CRO stocks (IQVIA, Medpace, Charles River) sold off on fears that AI lets pharma companies internalize trial work. The market is misreading the disruption.
• TD Cowen estimates AI could cut late-stage trial timelines from 58 to 47 months. Real productivity gains, but slower in practice than the hype suggests.
• The hardest parts of trial execution — multi-country patient recruitment, regulatory submissions across jurisdictions, GCP compliance across decentralized systems — are not easily internalized.
• CROs are infrastructure companies. Wall Street keeps pricing them like software vendors. AI should increase their value, not decrease it.
• New contracting models are already on TD Cowen’s radar: gain-share arrangements where CROs keep a portion of the time savings they deliver.
• The cost-plus model is the real obstacle. Decades of comfortable margins make the shift to performance-based pricing structurally painful.
• “Nobody ever got fired for hiring IBM” is what currently protects incumbents. The defining question: which CRO moves first to price on AI-enabled speed rather than time and materials?