My first two years of college were in a special medical school program focused on the courses required to get into a pharmacy program, to become a pharmacist. During that time, I worked as a pharmacy technician, gaining valuable firsthand experience with medications and how they are used to treat patients.

But two years into the program, I switched majors and put my medical schooling on hold.

In the mid-1970's, I ended up working at a tech company that shifted my educational focus toward technology and led me to my career in tech. However, my interest in medicine and especially pharmacology has never wavered.

From Pharmacy to Technology—and Back to Drug Discovery

But for most of my career, I've been following emerging technologies that promise to reinvent entire industries. Often, the hype outpaces the reality by several years. Occasionally, however, a technology emerges that really does deserve all the hype it's receiving. One of those is AI-driven biopharma R&D.

The research firm TD Cowen’s recent proprietary survey of 80 industry leaders and experts provides concrete figures to back up what many of us suspect: AI can reduce preclinical drug development costs and timelines by 70%.

The demand signal is already evident in growing investments in software, sequencing equipment, and computational methods to develop more experimental drugs over the next five years.

The emphasis on using more powerful computers for drug discovery came into sharper focus for me when I was allowed to do a deep dive on IBM's vision of using its early quantum computing for drug discovery.

Given my early educational focus on drugs and medicine, this particular research project excited me.

AI is not replacing scientific intuition—and it is unlikely to do so anytime soon. What it can do is make the costly, time-consuming R&D work that precedes clinical trials more efficient. In drug discovery, that means the computer screen is becoming nearly as essential as the lab bench: a core tool for exploring possibilities, testing hypotheses and guiding scientists toward the most promising compounds. That shift is familiar across industries, where digital tools increasingly become part of the craft rather than merely an add-on.

Brendan Smith , director of life sciences equity research at TD Cowen. "With more data, you're able to train AI algorithms better and get a higher likelihood of success. And while some tasks in the earlier phases of the drug discovery process will be transferred to computations, wet labs will remain, and scientists in them will continue verifying the efficacy and safety of medications using the traditional approach."

What makes this a different type of AI disruption from what we've been observing in the rest of white-collar work is that AI is not eliminating jobs but rather shifting workflow locations. That being said, I expect some areas of the pharmaceutical industry workforce to be impacted by the transition towards computation.

There’s also a policy angle to consider. The Trump administration's push for less animal testing in biomedical research will accelerate the shift towards computational tools, 3D human tissues, and other approaches to predict compound toxicity.

Based on TD Cowen survey results, the greatest increase in demand for advanced software capable of simulating biological processes—predicting drug interactions or adjusting dosage based on a patient's age or condition (e.g., newborns or pregnant patients)—is expected between now and 2028.

The money flows into " in silico ” platforms that allow scientists to run thousands of experiments within seconds, simulating toxicity and stability of the compound without using any physical samples. The kind of tooling that was once a science fiction dream has now become a reality.

According to the proprietary survey data, new drug development programs could grow by more than 10% over the next three to five years. With continued investments in lab equipment, this buying spree could mean an increase in spending of about $1 billion.

I always advise people to distinguish between genuine revolution and marketing hype, and this technology deserves the same treatment. At the moment, AI hasn't found a new drug that the FDA would approve. Some investors have voiced concerns about whether the technology will make any meaningful difference for patients soon.

The Promise Is Real, but So Is the Proof Burden

This is a trend I’ve witnessed several times before: a breakthrough technology excels in the early, more controllable stages of a process, then encounters the unpredictability of the real world. AI-driven drug development has genuine potential to make research faster, less expensive and more targeted. But its ultimate value will not be measured by what it can simulate on a computer screen. It will be measured by whether it helps deliver safe, effective medicines to patients.