Simulation-First Clinical Trials Could Change What CROs Are Worth

Today, two private biotech companies — Harvest Integrated Research Organization (HiRO) and Differentia Biotech — signed a memorandum of understanding to combine clinical trial operations with AI-powered simulation. Neither is publicly traded. You can't invest in this deal.

But what they are trying to do points to a structural shift in the clinical research industry, and the companies you can invest in are already feeling the pressure.

Clinical research organizations, or CROs, run clinical trials for drug companies. IQVIA, the largest, has a $43 billion market cap. ICON, the second largest, is worth about $13 billion. The entire CRO market is projected to grow from $93 billion in 2026 to $140 billion by 2031. This is a big, growing industry.

The question nobody is asking yet is whether simulation changes what a CRO is worth.

Here's the basic business. A drug company has a molecule. It needs to prove the molecule is safe and effective in humans. That process — from first-in-human trials through late-stage studies — costs billions and takes years. The industry average to bring a drug to market is now over $2.6 billion. CROs exist because running these trials requires an operation: recruiting patients, managing sites across countries, tracking data, satisfying regulators. It's a logistics and compliance business at scale. You pay a CRO to execute.

Differentia's platform, called DHARMA, does something different. It builds "digital twins" — computational models of how a disease works, how a drug interacts with the biology, and how different patients might respond. You simulate a trial before you run one. Their website claims this can reduce the number of patients needed by 20 to 60 percent. That's not a cost optimization. If true, it changes the size and economics of every trial a CRO runs.

HiRO brings the operational data — trial results, site performance, patient recruitment patterns from 1,600 trials across 85 countries. Differentia brings the simulation engine. The idea is that the simulation gets better when it's trained on real operational data from trials that actually happened, and the operations get better when they're guided by simulations that tell you what's likely to work before you spend the money.

Most people look at this and see a CRO adding AI. That's the obvious frame. But the more interesting question is what happens when the simulation is good enough that it becomes the primary decision-making tool, and the trial becomes a narrower confirmation exercise.

If the simulation says "this dose won't work, try a different one," the sponsor changes course before enrolling patients. If it says "you only need 200 patients instead of 500," the CRO runs a smaller trial. If it says "recruit these patients, not those," the logistics get simpler. The CRO doesn't disappear — someone still has to run the trial. But the center of value shifts from execution to simulation. The company that tells you what to do becomes more valuable than the company that does it.

This is the same pattern we've seen in other industries. When software replaced manual processes, the companies that built the software captured more value than the companies that managed the people doing the work. When automation improved manufacturing, margins migrated upstream to design and planning. Simulation-first drug development would shift value upstream from trial execution to trial design.

The public CROs know this is coming. IQVIA rebranded itself last year as an "AI-driven technology platform." The stock dropped 22% in a month after that announcement, even though the company reported 10 percent revenue growth. The market was voting on something: rebranding to AI doesn't change the underlying business model. IQVIA still earns most of its revenue running trials and selling data. Saying you're an AI company doesn't make the execution business a simulation business.

ICON has had a rougher trajectory. Its stock dropped from over $200 to the $66 range before recovering to around $170. Forward PE sits at roughly 19 times, which sounds cheap — but the trailing PE is over 480 times, meaning last year's earnings were a thin sliver relative to the share price. The market is pricing in a recovery that hasn't fully arrived.

Here's the hard part. The CRO business is built on scale and compliance, not technology. You need sites in multiple countries, relationships with hospitals, data infrastructure that satisfies the FDA and European regulators, and teams that know how to navigate every country's rules. That's hard to replicate. But simulation could make less of that hard stuff matter. If the trial is smaller and better-designed, you need fewer sites, fewer patients, less of the compliance machinery.

I suspect the real test isn't whether CROs adopt simulation. It's whether simulation changes CRO economics in a way that the incumbents can't adapt to. The big CROs are bloated organizations with thousands of employees. They can buy a simulation tool or build one internally. But their incentive structure is built around running more trials, not running smaller ones. There's a tension here that's not immediately visible: the technology that makes trials more efficient could also make the incumbent CROs' existing operations less valuable.

The HiRO-Differentia deal is small in absolute terms — a memorandum of understanding between a 450-person CRO and a smaller simulation company, with no disclosed financial terms. But it's a signal. The companies that are building for the next version of this business are already forming alliances. They're not waiting for IQVIA to decide what AI means for its 50,000 employees.

What to watch. If simulation-first trial design becomes real, you'd see it in a few ways: biotech sponsors start asking CROs to simulate before committing budget; trial sizes shrink while success rates rise; and the revenue per trial for CROs drops even as the number of trials stays steady. The publicly traded CROs that adapt fastest will be the ones that can pivot their existing infrastructure — their sites, their data, their relationships — into a simulation-informed model. The ones that can't will find themselves running smaller trials at lower margins, serving a market that's getting smarter about what it buys.

The question for investors isn't whether AI will change clinical trials. It's whether the companies that make money running trials today are structurally capable of leading the shift to the companies that will make money designing them tomorrow. That answer won't come from a press release. It'll come from trial sizes, margins, and the pace at which sponsors stop trusting execution alone and start demanding simulation first.