AI-led selloff in contract research firms may be misjudging disruption risk

By Kamal Choudhury and Siddhi Mahatole

March 31 (Reuters) – Shares of contract research organizations have tumbled on fears that advances in artificial intelligence could allow drugmakers to take clinical trial work in-house, but industry experts say the selloff overestimates how far the technology can replace the sector’s core capabilities.

IQVIA, Medpace and Charles River Laboratories have fallen sharply since Anthropic’s launch of advanced AI agents in February fuelled expectations that drugmakers could rely less on CROs.

A recent wave of partnerships between pharmaceutical companies and AI firms has further added to those worries.

“Could AI eat CROs? Yeah, I think that could be a possibility,” said Thomas Laur, CEO at data analysis firm DNAnexus.

But the nature of services that CROs provide, from patient recruitment to global trial execution, will be difficult to automate or replace, several analysts, industry experts and policy makers told Reuters.

CROs maintain global networks of trial sites and hold proprietary data that pharma companies, especially smaller biotechs, cannot easily replicate, Jailendra Singh, analyst at Truist Securities, said.

That view is echoed across Wall Street. Analysts at TD Cowen estimate that even a fully AI-enabled clinical trial setup would deliver only 10%–15% cost savings for drugmakers.

At the core of the argument is the pharmaceutical industry’s reliance on execution at scale.

Finding even a small pool of eligible patients for an early trial, across diverse demographics and geographies, requires enormous data and site networks that CROs have built over decades. “Pharma companies do not have that same level of data and expertise,” Singh added.

HUMAN ELEMENT

Executives of CROs say AI may streamline parts of the process, but cannot replace the human and operational backbone of trials.

“AI itself can’t reach out to the doctor, enroll the patient, make sure they show up to the appointment on time, record all the data,” said Brigham Hyde, CEO of Atropos Health.

Although AI could automate high-volume tasks such as patient pre-screening, critical decisions still require human oversight, said Ami Bhatt, chairperson of U.S. FDA’s Digital Health Advisory Committee.

Site execution, informed consent and safety monitoring remain firmly in human hands, she said, with accountability ultimately resting on people.

Others point to more fundamental constraints with the technology. AI cannot replace laboratory testing required for drug safety, and its use in direct patient care remains limited by regulatory scrutiny and liability risks, said William Pierce, a former deputy assistant secretary of public affairs at the Department of Health and Human Services.

OPPORTUNITY AHEAD

Rather than replacing CROs, analysts say AI could enhance their value by speeding up trials and improving efficiency.

TD Cowen’s analysts estimate a fully AI-enabled late-stage trial could be completed in 47 months versus a baseline of 58 months, an 11-month reduction.

That kind of timeline compression could become a powerful competitive advantage for CROs that invest heavily in AI.

For instance, reaching the market nearly a year earlier for a drug with estimated peak annual revenues of $1.5 billion could drive roughly $44 million in additional revenue, TD Cowen added.

“We expect new types of contracting to emerge, including gain-share arrangements on AI efficiencies,” the brokerage said.

But the recent stock slide suggests lingering investor unease.

Jim Lee, head of Inflammation and Autoimmunity at drugmaker Incyte, said investors are likely worried about the services that CROs may have to scale back, which could affect their revenue.

For now, analysts maintain there is no evidence of pharma companies cutting spending with CROs because of AI and termed the sell-off as “panic more than (a) real threat”.

For Singh, the bottom line is straightforward: “we do not see AI as a headwind for the industry; if anything, it is more of a tailwind.”

(Reporting by Kamal Choudhury and Siddhi Mahatole in Bengaluru; Writing by Mrinalika Roy; Editing by Shinjini Ganguli)