Artificial intelligence & machine learning

The promise of AI and machine learning-driven drug discovery has fueled excitement and investment in pharma for years.

Now Bristol Myers Squibb is leveraging those technologies in an approach that not only conceptualizes and creates potential molecules, but validates and forecasts certain endpoints well before clinical trials.

“We don't go into the lab and make things unless we have a good idea of what the forecast looks like,” said Mike Ellis, senior vice president and head of Bristol Myers Squibb’s Discovery & Development Sciences organization. “What are we excited about? What do we think we can learn from this molecule? What may be the concerns that we need to clarify in the laboratory experimentally?”

Ellis is spearheading BMS’ AI efforts, which include developing and implementing proprietary AI/ML tools that can predict both success and potential stumbling blocks before molecules are synthesized, such as “safety liability that could pop up later,” Ellis said.

All of BMS’ “predict first”-enabled small molecule programs now use AI to evaluate properties like efficacy prior to synthesis, a significant leap from just 5% in 2021. BMS is applying a similar approach to nearly half of its large molecule experiments as well.

Optimizing R&D has always been a priority but has become more pressing in the industry. Looming patent cliff losses are set to trigger billions in revenue losses for a number of pharma giants including BMS, which is losing exclusivity for moneymakers like Revlimid, Pomalyst and Sprycel in the coming years.

Those patent losses, along with increased R&D belt-tightening, are prompting companies to use AI to speed and de-risk the clinical development process. Pfizer just expanded its deal with AI firm XtalPi to develop accurate predictive tools for small molecules. Meanwhile, Amgen CEO Bob Bradway recently noted that the company is leveraging AI and machine learning “to choose the right molecules more quickly and bypass the empirical trial and error process.” 

Chemistry meets computation

Ellis served as executive director of chemistry for Celgene until BMS acquired the company in 2019. Before that, he was part of the medicinal chemistry team at Merck & Co. A chemist by training, he’s always “dabbled” with computational tools, including in his graduate work and post-doctoral fellowship, he said.

“From the first time I joined a medicinal chemistry team within the pharmaceutical industry, we were sitting around a table talking about programs with computational scientists in the room, bringing their perspective,” he said. “It spoke to the shift in mindset for the lab experimentalist, and was reflective of some of the ways we changed even our team composition … [to] increase our number of computational scientists within the team [and] the ratio of computational scientists to lab experimentalists.”

Now, he believes the industry — and BMS — is at “an inflection point” for the power of those tools.

‘Choose your own adventure’

There are always going to be multiple solutions to the same problem, and likewise, “multiple molecules that one could get to that could be safe and effective,” Ellis said. The predict-first model allows BMS to increase the probability of advancing the most promising molecules by using AI to guide early decisions before committing resources to lab testing.

While this approach is still like searching for a “needle in a haystack,” it makes the pile of hay a bit smaller by moving away from a funnel-based screening approach and moving toward a “choose your own adventure,” he said, allowing scientists to more precisely prioritize what molecules to make and how to test them based on AI-guided predictions.

“We are enriching what we actually go and make and how we study it based on prediction,” Ellis said. “So rather than potentially redundant or incremental differences between molecules, we're trying to strategically approach what we go in the lab and make [and] what data we generate on those molecules so that it can inform the next cycle as effectively as possible.”

Among the predict-first programs in the company’s pipeline is an effort to find CELMoD agents for sickle cell disease.

BMS is already studying CELMoD agents that target protein degradation to treat multiple myeloma and non-Hodgkin lymphoma, and the company presented positive data for those programs in June.

Although BMS knew it wanted to target protein degradation for sickle cell disease, Ellis said the company hit a “plateau” as it worked toward identifying a clinical candidate.

“We have many parameters that we’re going after in parallel. And often … you pull on one thread of those parameters and the other one breaks,” he said. “You can kind of get to all the pieces that you want in a molecule, except it’s spread across five different molecules … How do we marry things all together?”

That’s when the company “started to bring more computational scientists, as well as models and tools” into the program. Within weeks, they’d crossed the months-long plateau.

“As we started to apply prediction across what we saw as the most important parameters in a composite, computationally we started to move in directions that were different than where we had been before,” he said. “We were able to quickly get us into a better space where ultimately we were able to build each of those parameters into a single molecule as the ultimate goal.”

Ellis called BMS’ commitment to computational tools a “cultural evolution” in everything from determining which molecules move into the lab and how the company structures its teams, to embracing “hybrid intelligence” that marries human and computational skills for “enriching the quality of ideas and accelerating the decision making on programs.”

“What we've appreciated and been proud of is the cultural progression and adoption that we've seen,” he said. “I think it's going to be increasingly embedded in what we do.”

Artificial intelligence is taking shape as a major revolution in biopharma, from drug discovery to clinical trials. The wheels of technology spin quickly, and as the seeds of AI germinate, leaders are wrapping their heads around how far it’s come and where it’s going now.

Dr. Lynda Stuart, who serves as the inaugural president and CEO of the Fund for Science and Technology, a new venture capital group, has watched AI go from a fringe idea to a real-world effort.

“It’s an interesting time for the AI field in biotech because there’s a massive transition from a nascent technology to being a real tool that you can use,” Stuart said. “That transition from the theoretical to the real demonstrates an enormous opportunity for good — to define responsible use and mitigate risk.”

Stuart hearkens back to early days in the field. She worked with biochemist David Baker, who won half of the 2024 Nobel Prize in Chemistry alongside the founders of AlphaFold. His ideas, considered revolutionary now as AI takes center stage, weren’t always seen that way, Stuart said.

“[Baker] had this crazy idea that he was going to design proteins from scratch — everyone thought he was part of the lunatic fringe,” Stuart said.

There is always a process of hitting the gas and pumping the brakes on new technology, Stuart said, and the FFST is trying to find that line, calling for a mindset that puts AI on the right track for innovation but keeps potential hype in check.

“To understand and address this tension in innovation is to set up a framework that is agile enough to allow innovation to move at the pace it should move and then know when to regulate,” Stuart said. “We’ve proposed something called an ‘if-then’ framework, where if certain capabilities are reached, then you deal with them — but you shouldn’t get all panicked about something that’s 10 years away.”

Tradition vs. progress

A paradox of the pharmaceutical industry is that, while cutting-edge science drives every significant advancement, leadership is sometimes less enthusiastic about embracing the new ways. Financial and regulatory risk play a role in that tension.

“This was my obsession and focus for many years,” said Dr. Mathai Mammen, who served as head of R&D at pharma juggernaut Johnson & Johnson. When the AI revolution was still operating in much smaller computer science circles, he promoted using datasets and machine learning to enroll clinical trials, for instance.

“Ten years ago, I started seeing the evolution in technology and the convolutional neural network technology take off with the availability of electronic data,” Mammen said. “I knew this would yield a revolution, whether on the drug discovery or drug development side, and when I came to J&J, this became a mantra and something important to try and drive.”

Now, as CEO of the biotech Parabilis Medicines, Mammen recalls the resistance to AI, even though company leaders broadly supported the concept.

Still, full buy-in took time and was hard won, Mammen said.

“For some people, there’s a tried-and-true approach, and they feel there’s too much company and personal risk in rationalizing trying something new,” Mammen said. “That collective experience of being successful is in some ways a source of resistance, and the more experienced and successful a place has been, the more resistance there might be.”

Mammen said he was able to “drive through” that resistance with like-minded executives at J&J by focusing on near-term problems. That meant drilling down on the use of data rather than implementing rudimentary tech before the company was ready to jump in.

“We changed the organizational structure to give the data science team real authority — that comes with some reluctance because you’re taking authority away from someone, but those individuals quickly saw they were apt to be more successful with this new partnership,” Mammen said. “These are the step changes that led to organization-wide change from nothing to being probably the leading data science organization in pharmaceuticals.”

Mammen attributes the success of J&J’s portfolio at the time to how he and his team were able to deploy those tools and that mindset as early adopters.

State-of-the-art

Sanofi debuted a new campus in 2022 in East Cambridge, Massachusetts, as a chance to build not only for the scientific needs of today, but for technology coming down the pike, particularly in the field of immunoscience.

“There are going to be things you can’t foresee,” Matt Truppo, global head of research platforms and computational R&D at Sanofi, said during a press tour of the Cambridge Crossing facility in June. He pointed out that procedures that once once took months soon only took weeks to perform, and “now we’re doing multiple structures a day enabled in part by AI leveraging extremely large data sets.”

The company had to build with an eye to the future.

“What you try to do is build with the vision in mind of what you’re going to need, and when you find that those needs have accelerated beyond what you planned, then you go back and address them,” Truppo said.

The two buildings on the 900,000-square-foot campus were designed to foster collaboration among working groups of scientists and company leaders, and the R&D team had to also think about data in that way.

“We’re at a stage where new models are moving so quickly in academia and industry that you need to be able to test them and internalize them rapidly, evaluate them, see who you want to partner with and where your internal models are superior,” Truppo said. “We have the advantage that this building is quite new, but even with that, we have to keep up, and human intuition is an important part of it.”

Keeping the human aspect and the traditional scientific process intact as AI becomes a more widely used tool will keep the industry from spinning out of control — even as leaders explore new realms, the FFST’s Stuart said.

“I’m hoping the hype won’t get away from itself,” she said. “You always have to ground truth in real science — if you set out to make something and you fail, then that’s still teaching us something.”

Sponsored content

By IQVIA

Almost overnight, artificial intelligence moved from point solutions to transforming workflows throughout research and development. As technical capabilities advance, we will see the industry move beyond point solutions with predictive and descriptive AI to integrated processes and workflows embedded with Large Language Models and agentic AI. Agentic AI will reshape clinical research processes as we know them today, with AI-powered workflows being applied from design, to startup and throughout execution to expedite trial activities, reduce white space and improve quality.

One of the key areas of impact we will see with agentic AI is its ability to overcome the limitations of siloed data. Historically, gaining insights from varied, disparate data sources was a slow, manual endeavor that was often static and lacked comprehensiveness. The effort to synthesize insights garnered from multiple data sources into a holistic view to facilitate decision making has brought another set of challenges. Moving forward, specialized agents will be deployed to derive insights and drive actions from the data. For example, if a sponsor wanted to understand asset potential within an indication, an agentic framework could be built, including a:

• Retrieval agent to access literature, conference abstracts and other documents
• Trial search agent to capture information from trial registries
• SQL agent to engage with proprietary clinical trial data
• Specialized agent to create predictions based on related existing approvals, mechanisms of action and patient populations to name a few

The true power of an Agentic system lies in its orchestrator agent — an LLM-powered ‘command center’ that doesn’t just coordinate, but actively mobilizes a network of specialized agents. These agents operate autonomously, but within defined guardrails, leveraging real-time data and integrated tools to generate cohesive and context-rich outcomes that drive decisions and action. Agentic AI's action-oriented capabilities enable it to autonomously initiate and manage tasks, driving efficiency and ensuring timely completion of critical activities. By embedding these capabilities into clinical development workflows, sponsors can make more confident decisions on asset and indication selection and trial design to increase the probability of technical, regulatory and commercial success.

In addition to querying data for strategic business decisions, we will see changes throughout study planning and conduct that not only create efficiencies but also lead to improved data quality. It will become commonplace for clinical researchers to leverage AI assistants to enable them to quickly and easily query data, create documentation and execute tasks — with human oversight. This will lead to sponsor benefits across clinical trials, such as:

• Optimized and de-risked protocol designs
• Accelerated site activation
• Faster patient recruitment
• Enhanced oversight and agility
• Accelerated data and submission readiness

As clinical development evolves, success will hinge not just on connecting data, but on activating it. An Agentic AI platform provides intelligent collaborators that can navigate the broader clinical technology ecosystem, interfacing with platforms and executing tasks across domains. These agents don’t just surface insights; they act on them — drafting protocols, triggering workflows and accelerating trial milestones. This shift from passive data aggregation to proactive orchestration marks a fundamental leap in how R&D will operate.

The future of clinical development is being redefined by agentic AI — not through abstract and incremental innovation, but through measurable acceleration. By embedding intelligent agents into workflows, sponsors can reduce white space, compress timelines and streamline execution across clinical phases. These systems don’t just analyze —they act — initiating tasks, resolving blockers and adapting in real time. The result is a more agile, responsive R&D engine that delivers faster outcomes without compromising on quality.

The FDA plans to roll out its new AI tool, called Elsa, “ahead of schedule and under budget,” according to Commissioner Dr. Marty Makary, who offered up the most information to date about the agency’s new tool and AI goals this week. But questions remain.

After completing a pilot program of AI-assisted scientific reviews, the FDA announced it was taking an “aggressive” approach to scale up the use of Elsa agencywide by June 30, but progress has moved up the timeline. In a video announcement, Makary described how Elsa will modernize the agency to help employees, including scientific reviewers and investigators, reduce non-productive busywork. 

“The agency is using Elsa to expedite clinical protocol reviews and reduce the overall time to complete scientific reviews,” Makary said. “One scientific reviewer told me what took him two to three days now takes six minutes.”

Other tasks performed by Elsa included “summarizing adverse events to support safety profile assessments, conducting expedited label comparisons and generating code to facilitate the development of databases for nonclinical applications.”

The launch has been criticized by some FDA employees as “rushed,” possibly to account for significant personnel cuts at the agency enacted by the Department of Government Efficiency over the last several months, STAT News reported.

Makary also attempted to address security concerns in the announcement, noting the language model-powered AI tool wasn’t trained on data submitted by the industry and data is held within a secure “GovCloud environment.” The agency provided few other details regarding how Elsa was trained and how much it’s been tested.

Questions remain

The FDA will plan to focus Elsa’s scope on administrative tasks such as summarizing documents and data extraction. This is the right approach, according to Panna Sharma, CEO and president of Lantern Pharma, an AI biopharma.

“These are exactly the applications where current generative AI … completely excels and where we see similar gains in drug development and scientific review internally at Lantern Pharma,” he said. “The key will be maintaining this performance as the system scales across different review types, medicines, diseases, and with differing or incomplete background information or literature.”

The agency noted that Elsa’s introduction will be just “the initial step in the FDA’s overall AI journey,” with the potential to add more tasks like data processing down the line. In May, the FDA also said it would refine its features over time and gather feedback.

“The big question I have focuses on the continuous learning and improvement side of things. How regularly will the FDA audit some of those AI outputs and help identify certain error patterns or maybe hallucinations or problematic prompts that might require retraining? ” said Dr. Chase Feiger, CEO and co-founder of Ostro, an AI company that works with many of the largest pharmas.

One area of concern is transparency. Will pharmas know if AI was used in their reviews, for example? If an AI-generated analysis leads to a rejection or delay, sponsors may face the challenge of “limited visibility into how much AI influenced the decision, raising concerns about opaque or unvalidated reasoning,” wrote lawyers from Hogan Lovells in a May blog post.

Other legal experts see the potential for AI tools to have too much influence in review decisions.

“If sponsors are aware that AI was used in scientific review of their applications, the agency’s use of AI could become a topic in future appeals, requests for supervisory review, or Formal Dispute Resolution Requests following unfavorable decisions on premarket applications,” King & Spalding lawyers wrote in a recent client alert.

A handful of regulatory and data experts peppered the FDA’s LinkedIn post of the video announcement with questions about the more nuanced details of the tool, alongside many singing praises of the agency’s modernization efforts.

“So many questions,” wrote Dr. Hugh Harvey, founder and managing director of U.K.-base Hardian Health, a software and AI medical device consultancy firm. “[W]hat verification and validation was done. Where is the public info? The FDA expects industry to do these things, so why not them?”

“They should release a summary of how they developed and validated this AI, including how they used all the internal filings and submissions to build their algorithms,” suggested Troy Trboyevich, director of regulatory affairs at Philips.

AI is here

Even without a detailed explanation, Elsa’s rollout shows the FDA is pursuing an internal AI strategy regardless of outside opinion. 

The FDA has been under pressure to create clear regulations for AI in the pharma industry for a few years and released a draft guidance in January to show how the agency will assess the risk of AI models in drug development. While it offered the first look at a possible framework for how pharmas using AI will be evaluated by regulators, many AI experts were left wanting more details. The first steps provided in the guidance also underscored how the industry is still in the early stages of AI adoption.

“I find it very exciting as well as surprising that the FDA is putting their foot on the gas in terms of helping roll out AI internally within the agency, because it actually aligns quite nicely with the adoption speed that we're seeing within the pharmaceutical manufacturing side of the table,” Feiger said.

And for AI life sciences companies, Elsa’s debut may be a source of validation.

“The timeline is aggressive, especially given that the scope and rigor of the initial pilot testing wasn't fully showcased, but this reflects the urgency FDA feels about modernizing their review processes and doing more with less staff,” Sharma said.

When Eli Lilly hired computational pathology expert Thomas Fuchs as the company’s first-ever chief AI officer in October, the announcement showcased pharma’s demand for a very specialized combination of skills. 

While many AI functions in pharma are outsourced, pharma and biotech companies are building up their internal AI workforces, triggering a higher need for leaders with a background straddling tech and health care.

A recent survey of about 500 industry professionals by the CRO Parexel showed it’s a pressing issue for drugmakers.

“A key insight is that 51% of biopharmaceutical leaders consider AI experts as one of the top three roles they need to fill in the next three to five years,” Keri Mattox, chief business officer at Parexel, said via email.

But recruiting experts for such specialized roles isn’t easy. At Lilly, Fuchs is tasked not only with leading the company’s strategic direction for AI initiatives in drug discovery, clinical trials, manufacturing, commercial activities and internal functions, but also with identifying, building and managing AI and machine learning solutions. 

“These are not easy skills to train, and finding candidates can be challenging due to the specialized skills required. Competition is fierce,” Mattox said.

What to look for in an AI exec

Ideal candidates for pharma and biotech AI roles have experience that combines “technological expertise with a deep understanding of drug development,” according to Mattox.

Fuchs’ resume weaves together experience in multiple industries. At Mount Sinai, he worked in digital health, AI and computational pathology, and was dean and inaugural department chair for AI and human health. He also had stints at Memorial Sloan Kettering Cancer Center, NASA's Jet Propulsion Laboratory and the California Institute of Technology.

“They should be proficient in AI and data management, with broad therapeutic and drug development knowledge — not just a siloed area,” Mattox said of candidates. “This includes the ability to critically assess AI outputs and integrate them into decision-making, which requires a strong grasp of both algorithms and clinical contexts.”

But not all candidates will have a perfect mix of skills. Usually, they’re usually stronger in one field than the other, according to Bill Holodnak, CEO and founder of the executive recruiting firm Occam Global.

“We see very few people [who] come with equal competence or equal duration in either [healthcare or tech],” he said.

Many digital experts also often start in one field and move into the other. But part of the successful hiring process is finding leaders who are equally passionate about and committed to both disciplines.

“Are they believable in their commitment to do this? Because we don't want to waste our time and our clients’ time with somebody who's doing this as a hobbyist or as just a way to scratch an intellectual itch,” Holodnak said.

Motivation is also important, especially if a candidate has “uneven” skills on paper. Someone who has more experience in the tech world, but is driven to move into health care or the life sciences because of personal experience with a certain disease would have a strong motivation to succeed in a multidisciplinary role, Holodnak said.

Other “soft skills,” are important, too, like “critical thinking, problem-solving and agility to use AI in a way that creates efficiency,” according to Mattox. 

“These skills enable the workforce to adapt quickly to changing technologies and methodologies, navigate complex challenges and envision disruptive and innovative approaches,” Mattox said. “Candidates also need to understand the importance of establishing measurable outcomes and outlining clear KPIs that demonstrate the value of AI.”

Cultural factors could also come into play including the need for tech professionals to adjust to the slower speed and more complicated regulatory hurdles of the life sciences world.

“They're used to creating a product, selling it and then selling the company,” Holodnak said. In the life sciences, “there's a longer timeline and there's much greater complexity in a product development life cycle.”

When it comes to finding talent, Mattox suggests active outreach, such as “engaging with universities and tech hubs,” as well as providing ongoing AI and other tech training for the current workforce.

“The companies actively working to develop and retain their teams are best prepared in the current environment,” she said.

AI-driven drug discovery has gotten much of the glory in pharma. 

Isomorphic Labs — which is leveraging an AI drug discovery platform based on Nobel Prize-winning AlphaFold software — pulled in the largest biotech investment haul this year to date with a $600 million fundraising round in March. 

And drug discovery is a prime focus for many of the highest valued AI companies working in pharma. Tempus AI, which has a market cap of about $9 billion, got its start leveraging molecular data for precision cancer care and has since expanded its capabilities to include drug discovery. Its services have been used by almost all of pharma’s largest public companies. 

So far, none of this momentum has resulted in a new drug approval.

Insilico Medicine is one of the frontrunners in that race with a candidate it calls the “first end-to-end generative AI-assisted drug,” that could soon move into phase 3 after garnering positive topline results late last year.

But before an AI-discovered drug proves its worth in the clinic, the technology is shining in other areas such as diagnostics and data management, according to Jessica Owens, co-founder of the healthcare-focused venture capital firm Initiate Ventures.

“I think there are going to be winners on the periphery far before we’re high-fiving about AI-discovered drugs,” she said.

These AI-driven advances may fly under the radar despite their impact because the space is currently dominated by small companies with “baby point solutions,” Owens said.

“They are all picking off one piece of the preclinical value chain,” she said.

Startups offering single service-based solutions also don’t attract as much buzz on the investment scene. 

“Historically, VCs don’t like service companies because the margins don’t work as well and they’re too hard to scale,” Owens said. “With those business models, it’s hard to get that hockey stick kind of growth.”

But that isn’t to say these startups aren’t offering strong ROI. Here’s where the technology is delivering breakthroughs throughout pharma. 

Clinical trial gains 

While shaving time off of drug discovery for a single asset can be a boon for one company, Owens sees wider value in AI technologies that offer to streamline development for the entire industry — especially in clinical trials. 

In particular, Owens pointed to Axiom Bio, a startup taking aim at predicting liver toxicities, which it says are linked to 20% to 25% of clinical drug failures. Using what it describes as the “world’s largest primary human liver dataset,” Axiom tested its AI model against assays used by Pfizer and AstraZeneca that demonstrated it was just as effective at a much lower cost. 

Ultimately, Axiom is positioning the technology to help phase out animal modeling and plans to extend its use to other organ systems including the heart and brain. 

AI solutions are also making an impact for ICON, a large research, development and commercialization firm based in Ireland. The company has implemented several AI tools, some of the proprietary, that tackle onerous aspects of clinical trial work including document storage, endpoint strategies and resource forecasting. 

In one case, ICON’s site selection tool, One Search, helped a sponsor bounce back from enrollment challenges by quickly identifying “high-performing sites” and reversing “enrollment shortfalls,” said Gerard Quinn, ICON’s vice president of IT innovation and informatics, in an email. 

Across the clinical trials industry, AI companies are working to develop solutions that automate patient enrollment and recruitment, and “match patients to trials in minutes rather than months,” Owens said. 

Diagnostics and the data deluge 

Initiate’s portfolio is also giving rise to several biotechs. The VC is currently building a startup called Tensor Bio that’s developed an AI-powered blood test for MASH that will create personalized treatment plans. 

Pinpointing effectiveness will be critical as new and high-profile MASH therapies come into play.

“This is big,” Owens said. “The first drug for MASH was recently approved, but there was a graveyard of drugs that failed before that. Some GLP-1s [could get] label expansions but none are silver bullets, so we are moving towards a combo treatment approach.” 

Another Initiate-backed company, Cornerstone AI, has developed a solution for a major behind-the-scenes problem in healthcare — cleaning large amounts of “messy” real-world data. 

“What they’re doing is not glamorous, but it’s super important,” Owens said.

Cornerstone has worked with a wide swath of companies, including biopharmas and large healthcare systems, to leverage an AI assistant that Owens said works like a “spellcheck for data” by identifying and correcting errors. 

“What would take four data scientists four months to do, can now be [done] in a week with one data scientist,” Owens said. 

Areas of potential failure — and opportunity 

AI capabilities aren’t quite ready to steal the spotlight in every area of R&D. 

Like many industry leaders, Owens argued that pharma still has a long way to go before animal modeling can be replaced by alternative testing methods based on AI, partially because the technology can’t yet replicate the intricacies of biology. 

“There’s a lot [that] AI can’t predict with respect to biology. It’s just too complex,” Owens said. “But we’ll get there.” 

And although pharma’s AI space has already been flooded with scores of companies, there are still lingering gaps in the market.

For example, there could be an opportunity for a company pulling multiple solutions under one roof and providing more end-to-end drug development services based on AI. 

“We are interested in companies linking these solutions together,” Owens said. 

And as the AI market continues to develop, the companies that ultimately succeed will also likely have a platform based on proprietary — rather than public — data.

“The way I would predict who the winners are going to be is by … looking at companies that were built on proprietary data,” Owens said. “Those are the companies that will create value over time.”

The fundamentals of leadership haven’t changed much in the more than three decades Dr. Mikael Dolsten has been a prominent figure in the pharma industry, most recently as chief scientific officer and president of R&D at Pfizer.

But the world is changing rapidly, driven in part by technological advances like AI and the proliferation of data-driven approaches. Leaders can react to those changes either with their heads in the sand or their eyes wide open, said Dolsten, who capped his career with a record-breaking vaccine approval during the COVID-19 pandemic before stepping down from the Pfizer role last year.

“On a personal level, leadership is quite similar to what it’s always been,” Dolsten said. “It’s just become technically more complex — but curiosity is the real key.”

Technology has been a major driver not just of industry breakthroughs but of the slower tectonic forces underneath those advances. To understand how the currents are flowing, a leader needs to be inherently curious, Dolsten said.

“I’ve been fascinated with how much technology plays a role as a constant driver of change at an incremental innovation level,” Dolsten said. “The convergence of technology, data-driven biomedicine and curiosity on a personal level has made these three decades so amazing for me, and I hope the next two to three decades will continue to be the same thrill.”

In his post-Pfizer career, Dolsten has embarked on several paths that sate his curiosity within the wider world of biopharma, taking board seats at smaller companies that he believes can shape the future of the industry. 

Earlier this year, Dolsten joined the board of Immunai, an AI drug discovery and development company “mapping” the immune system to find new treatments for a range of diseases. The move underscores the direction he sees the industry heading in terms of technological advances, and how he believes leaders should relish the changes.

Ostriches and chameleons

When adopting new technologies or discoveries, R&D leaders can be “the bird that puts his head in the sand and doesn’t see what’s happening around it,” or they can take it all in with eyes on a swivel, “reacting constantly with a 360-degree view,” Dolsten said.

The savviest leaders will understand that a tech-friendly approach requires co-existence between humans and AI, Dolsten said.

“You have human intelligence and you have artificial intelligence, and you need to be aware of how to manage risk and biases in both,” Dolsten said. “Putting human and artificial intelligence next to one another can play such a tremendous role for pharmaceutical companies in overcoming biases or limitations in how we look at the world.”

In exploring that dynamic, Dolsten looks to his mentor and colleague Daniel Kahneman, a Nobel Prize-winning behavioral economist, who told him that “a life where you wouldn’t be able to see the color of reality in its sharpest nuances isn’t a life worth living.”

Kahneman, who died last year at the age of 90, taught Dolsten that biases in human intelligence can in fact be mastered through informed decision making. AI, when used the right way, can augment that process, Dolsten said.

The takeaway? Knowledge is the lifeblood of drug development, and advances in computing power offer it in droves.

“Whether it’s large-language models or sequencing DNA … as [long as] you’re well aware of your model, I would say artificial intelligence that analyzes large data sets is less biased than human intelligence,” Dolsten said. “That’s why I’m so fascinated when companies collaborate with Immunai and bring to the table different types of insight.”

The Immunai team, led by CEO Noam Solomon, leverages a tech and AI perspective to start speaking “the language of the future, how biology and medicine should be spoken and looked upon,” Dolsten said.

Immunai’s platform, which includes an atlas of immune cells and a development-focused immunodynamics engine, can help solve some of the challenges faced by drugmakers like AstraZeneca and Teva Pharmaceuticals, both of which have signed multi-year collaboration agreements with the company. While the pipeline is still under wraps, the companies are investigating opportunities in oncology and immunology.

The goal, Dolsten said, is to “roll up our sleeves and define the questions that therapeutic-oriented companies need to have answered.”

Personal ecosystem

With any new, wide-ranging technology like AI comes the inevitable hype, and Dolsten said it’s critical that leaders do their homework. Putting those two letters ahead of a business proposal doesn’t necessarily give it value, he said.

“My homework always starts, whether I’m filling a board seat or overseeing a partnership, with asking for references of what they’ve done recently in the area, some real case reports,” Dolsten said. 

More than anything, Dolsten said it’s important for biopharma leaders to ask questions rather than assume they have the answers, especially at a time of technological upheaval.

“I encourage people to take a step back and look at the questions that will be dominant over the next five years — you don’t need to continue with the tools or teams that make you feel comfortable,” Dolsten said. “Getting out of your comfort zone is helpful, so I welcome change, and I think it’s necessary that we have the openness and willingness to find new ways to work together.”

Dolsten’s series of board appointments since stepping down from Pfizer point to what he sees as the next wave of biopharma innovation. The former exec currently has a seat on the boards of gene therapy specialist Rocket Pharmaceuticals and oral biologics maker Orbis Medicines, and holds advisory roles that include the protein structure AI company Chai Discovery. He said more announcements will be made this year.

These steps are a way for Dolsten to explore his scientific curiosity and build a network that bridges the industry’s past and its future.

“My ambition here is to create a personal ecosystem, and I think my contribution is much larger in each of the companies because of the connectivity I bring,” Dolsten said. “I hope I can connect companies that are making breakthroughs in one area, but allow them to seamlessly communicate with tools that can be adjusted and used across different areas.”

Part of what drives this push into future technologies was the frustration Dolsten encountered in the drug development process at companies like Pfizer.

“The awareness of what it takes to make a product that brings value to patients and is also commercially successful was a continuous challenge — with some great successes and some setbacks — in my work with Pfizer,” Dolsten said. “It’s a complex playing field from launching a drug and allowing the innovation to be recognized and rewarded.”

Although Dolsten is diversifying his post-Pfizer career, he looks at AI as the great connector.

“I hope that Immunai becomes a central node for me as they try to decode the language and the letter of diseases and help this overall ecosystem be better at using that language in communication that leads to products,” Dolsten said. “The crown jewel of my career was my time at Pfizer, but I’m very curious about this next chapter and playing a role in a different way.”

After a few years of booming interest in AI platforms, the rubber has hit the road. Now, the market is watching closely to see if drugs discovered with the new tech will pan out in the clinic — and are still worth backing. 

Signs have been good so far for Insilico Medicine’s lead candidate, which it claims is the “first end-to-end generative AI-assisted drug.” The company announced positive topline results for a phase 2 study of the idiopathic pulmonary fibrosis treatment in November, which Insilico said demonstrated “encouraging clinical efficacy” by improving lung function for all 12 weeks of the trial. Insilico now plans to charge ahead into pivotal trials with the drug that was recently rechristened from ISM001-055 to rentosertib.

But results have been murkier for other AI front runners.

Recursion Pharmaceuticals revealed mid-stage data for its candidate in September that demonstrated safety but only showed a “trend” toward efficacy endpoints. 

Some industry watchdogs suggested the drug may be doomed to fail for the tough-to-treat neurovascular condition called cerebral cavernous malformation. And Recursion, which has built a diversified oncology and rare disease pipeline using an AI platform leveraging “one of the world’s largest proprietary biological and chemical datasets,” has been bolstering its clinical prospects in other directions.

Last year, Recursion picked up Exscientia, an AI drug discovery company facing its own challenges after losing a partnership with Bayer, in an acquisition worth nearly $700 million.

Have these industry shifts and uneven clinical results dulled the market’s appetite for AI drug discovery companies? 

While the COVID-19 market boom has cooled broadly across biotech, a major funding round for a Google-backed AI startup last month indicates there’s still momentum.

In fact, 71% of venture capital investments across U.S. industries in the first quarter of 2025 were tied to AI, according to PitchBook.

And in pharma, which has long been hampered by high rates of clinical setbacks, any new approach that brings even a modicum of efficacy or speed to the R&D process could trigger huge returns — a pursuit investors are still rewarding. 

Here’s a look at the biotech that raked in that major haul and at other startups that have recently won funding to chase the AI dream. 

Isomorphic Labs’ potential AI goldmine

Months after snagging the Nobel Prize in Chemistry, the CEO of Isomorphic Labs helped the company bag $600 million in its first round of external fundraising. 

The company’s AI drug discovery platform is based on AlphaFold software, which can predict protein structure, DNA, RNA and ligands that bind to targeted proteins, while creating new proteins as well. The invention was rolled out by Google DeepMind and led to a Nobel Prize for three scientists, including Demis Hassabis, CEO of both Isomorphic and Google DeepMind. 

Hassabis predicted last year that AlphaFold could help create an AI drug discovery business worth more than $100 billion — a mouth-watering proposition that investors are clearly eager to get behind.

The company’s ultimate aim is to shrink the average discovery time from five years to two. Isomorphic is also exploring emerging modalities and developing its own pipeline of candidates targeting cancer and autoimmune diseases.

"AI and machine learning have long held the promise of transforming drug discovery, yet few companies over the past decade have unlocked their full potential,” Dr. Krishna Yeshwant, managing partner at GV, one of the firms that participated in the funding round, said in a release. “After witnessing the extraordinary pace of innovation at Isomorphic Labs, we believe their pioneering approach will redefine AI-powered drug discovery.”

Thrive Capital led the $600 million funding round in March with additional boosts from GV and Alphabet.

Isomorphic also landed research collaborations with Eli Lilly and Novartis aimed at discovering small molecules for undisclosed targets last year. Combined, the research partnerships are worth up to $2.9 billion. 

Other recent AI wins

Isomorphic’s $600 million haul was the largest biotech funding round so far this year. But other AI companies have also scored significant backing.

Insilico nabbed $110 million last month in a series E round led by Value Partners Group, which has been around since the 1990s and claims to be one of the largest independent asset management firms in Asia.

The fresh capital will help Insilico refine its AI models and algorithms while it advances its pipeline of clinical candidates, the company said.

A duo of smaller upstarts also pulled in funding this year.

Simon Kohl, a scientist who worked on the second iteration of AlphaFold and co-led DeepMind’s protein design team, arrived on the biotech scene with a startup of his own in February. Called Latent Labs, the company’s AI platform is focused on “applying the latest generative techniques to design proteins from scratch,” it said in a release. Latent emerged from stealth with $50 million in funding.

Two notable names from their respective fields also unveiled an AI drug discovery startup. LinkedIn co-founder Reid Hoffman and Columbia professor Dr. Siddhartha Mukherjee, author of a landmark cancer history book “The Emperor of all Maladies,” have joined forces to launch Manas AI. The startup will leverage a Microsoft cloud computing platform and AI to design and hunt for treatments in cancer, autoimmune and rare disease, including “some previously considered unreachable,” it said in January.  

Manas AI raised $24.6 million in seed funding and will initially set its sights on breast cancer, prostate cancer and lymphoma.