The pitch sounds almost too convenient: simulate an entire clinical trial before a single patient swallows a pill, catching costly mistakes in a digital sandbox instead of a hospital ward.
Deep Intelligent Pharma, a Singapore-based startup that automates chunks of the clinical trial process, just closed $60 million in equity financing—its second substantial raise in fewer than ninety days. The February 2 disclosure follows a nearly $50 million round in mid-December, a rapid-fire fundraising sprint that's netted the company roughly $110 million since the final weeks of 2025.
For a sector where drug development timelines stretch across years and budgets routinely exceed $2 billion per approved asset, that's a notable vote of confidence. Whether the technology lives up to its billing is another question entirely.
Money Flows From Familiar Corners
Trustar Capital, the private equity arm of CITIC Capital, led the $60 million infusion alongside Jinyi Capital and Kaitai Capital. Two backers from December's Series D—CDH Baifu and Xinding Capital—doubled down. Index Capital advised on both transactions. The company declined to disclose either a lead investor or its current valuation, a common move for startups navigating competitive fundraising environments but one that leaves outside observers guessing at the company's worth.
The December round itself came seven years after Deep Intelligent Pharma's roughly $15 million Series B from Sequoia China (now rebranded as HongShan Capital). That's an unusually long gap between institutional checks, though the company's founder, Li Xing, appears to have used the intervening period to build out what he describes as infrastructure rather than chasing flashy customer logos alone.
The Product: Synthetic Data Meets Regulatory Gauntlet
Deep Intelligent Pharma doesn't position itself as a traditional contract research organization—the service providers that manage trials on behalf of drugmakers. Instead, it's selling software that automates protocol design, biostatistics, pharmacovigilance monitoring, regulatory document authoring, and translation. The company calls its underlying architecture a "Synaptic Agent Ecosystem," a system it claims comprises more than 10,000 individual AI agents working in concert. The stated accuracy rate: 99.9 percent.
The headline feature is something called AI Digital Rehearsal. Sponsors input study parameters, and the platform generates synthetic data to model how a trial might unfold—patient dropout rates, adverse event patterns, statistical power. The idea is to surface design flaws or operational bottlenecks before committing millions to enrollment and site management.
Li, who logged over a decade at Pfizer, Sanofi, and Johnson & Johnson before founding Deep Intelligent Pharma, argues the approach could sidestep the protocol amendments that routinely add hundreds of thousands of dollars and months of delay to studies. (Research from Tufts Center for the Study of Drug Development has documented those costs, though the center's figures vary by therapeutic area and trial phase.)
It's a compelling narrative. Clinical trials remain one of the most expensive, failure-prone stages of drug development—Deloitte pegged the average cost per approved asset for large pharmaceutical companies at $2.23 billion in 2024. Any tool that promises to de-risk that process finds an audience.
Blue-Chip Names, Unverified Milestones
Deep Intelligent Pharma lists Bayer, Roche, Bristol-Myers Squibb, Merck, and J&J MedTech among its clients—a roster that lends credibility, though the company hasn't detailed the scope or scale of those engagements. In late January, it announced a strategic partnership with Imbioray Bio valued at more than 300 million yuan (roughly $42 million) to develop an AI platform for ACC-NK cell therapy, a niche corner of oncology.
Perhaps the most eye-catching claim: that Japan's Pharmaceuticals and Medical Devices Agency approved an AI-authored Phase I/IIa protocol with zero revisions. That assertion has circulated in trade press coverage, often attributed directly to the company. Independent confirmation in public regulatory records, however, remains elusive. PMDA filings aren't always transparent to outside researchers, which makes the claim difficult to verify—or refute.
The company also points to operational throughput metrics: 200 million words translated for drug licensing deals, IND eCTD packaging completed in approximately two weeks. Those figures underscore the regulatory and operational complexity Deep Intelligent Pharma is targeting, even if they don't directly address the trickier question of how much decision-making authority sponsors are willing to cede to algorithmic systems.
A Bet on Automation in a Risk-Averse Industry
Deep Intelligent Pharma now employs more than 200 people across offices in China, Japan, and Singapore. The company says it serves over 1,000 clients, a figure that likely includes a mix of pharmaceutical giants, biotechs, and research institutions. The business model appears to hinge on subscription or project-based fees, though the company hasn't disclosed pricing.
The back-to-back raises suggest investors see room for automation in a sector historically slow to adopt new technology. Pharmaceutical companies move cautiously—regulatory scrutiny is intense, and the consequences of error are measured in patient safety, not just quarterly earnings. That makes Deep Intelligent Pharma's adoption curve all the more interesting. If sponsors are signing on in meaningful numbers, it signals a shift in how the industry thinks about risk and validation.
Or it signals that the marketing is working, and the operational proof will come later.
What Comes Next
Proceeds from the February round will fund what Deep Intelligent Pharma describes as a "generational upgrade" of its core technology—industry speak that could mean anything from architectural overhauls to incremental feature releases. The company also plans to accelerate its push to become what Li calls "pharmaceutical infrastructure," a term that implies ambitions beyond point solutions.
Whether the platform delivers on its promise depends on factors that venture capital can't solve: regulatory acceptance, sponsor trust, and the willingness of an industry built on decades of manual processes to let software make high-stakes decisions. The $110 million buys runway. It doesn't guarantee arrival.
