A drug discovered by an algorithm. An organ-on-a-chip used to test pharmacology. An injection taken twice a year that cuts cholesterol in half. These are either on the market or in late-stage trials right now, and they didn’t get there by chance.
Behind them is a set of deeper shifts rewriting how drugs are discovered, developed and brought to patients. For smaller biotechs, these shifts matter in concrete ways. They affect how you design your program, what investors expect to see and how persuasively you can make your case to regulators. These are five worth understanding now.
Trend one: AI is redefining how drug programs are built
Not long ago, the question was which biotechs were using AI. Today, many are. The more relevant question now is whether it’s changing outcomes, and where in the development process it’s earning its keep.
AI started with target identification and has moved well beyond it. It now influences everything from how clinical programs are designed and patients identified to which endpoints are chosen and how regulatory submissions come together. The tools are more capable, the precedents are accumulating and the programs that treat AI as a box-checking exercise are starting to fall behind those using it to make better decisions earlier.
Consider Insilico Medicine’s drug candidate rentosertib. The company identified a novel anti-fibrotic drug target using a predictive AI approach. It then used generative chemistry optimized by machine learning to derive the first compounds for testing. As a result, it took 18 months from target discovery to preclinical candidate nomination, with positive results published in Nature Medicine in June 2025.
The regulatory picture reflects this maturation. The FDA has reviewed more than 500 submissions incorporating AI components since 2016, and issued its first draft guidance on AI in drug development in 2025. Later in the year, the agency formally qualified PathAI’s AIM-MASH through the Drug Development Tool Biomarker Qualification Program, making it the first AI-based tool cleared for clinical trial use. The tool helps pathologists score liver biopsies in trials for MASH, a serious progressive liver disease. It’s a narrow application, but the qualification pathway it establishes is significant for any program considering how to incorporate AI-generated evidence into a regulatory submission.
On the industry side, NVIDIA and Eli Lilly announced an AI co-innovation lab backed by up to $1 billion over five years at the J.P. Morgan Healthcare Conference in January 2026. The lab is designed to co-locate pharmaceutical scientists and AI engineers in a shared facility focused on rethinking how drugs are discovered and developed. Commitments at that scale signal that AI is shifting from experimental capability to infrastructure.
This doesn’t mean the technology has fully arrived, though. No fully AI-discovered drug has yet received FDA approval, and that distinction matters because it clarifies where the value currently lies. The programs benefiting most from AI are using it to sharpen early decisions and reduce the cost of being wrong, not to sidestep the clinical work that ultimately determines whether a drug works in people.
A McKinsey report reinforced this point, finding that many of the largest gains from AI may come less from molecule design itself and more from improving decision-making at the study, program and portfolio levels. That includes helping teams decide which programs to advance and which to stop earlier.
Trend two: Precision medicine is becoming the default
Targeted drug development was once synonymous with oncology. That’s no longer the case. The expectation that a drug should work for a clearly defined patient population, and that you can show who that is and why, now extends across immunology, neurology, cardiovascular disease and rare conditions.
The reason is straightforward. A trial that enrolls a broad mix of patients, many of whom may not have the biology a drug is targeting, produces murky results. Real effects get buried, regulators see a weaker case and investors see more risk. A well-defined patient population produces cleaner data, a stronger regulatory story and a more convincing pitch.
Two examples show how far this has spread beyond oncology. In cardiovascular disease, companies are developing treatments specifically for patients with elevated lipoprotein(a), a genetic risk factor present in roughly 20 percent of the global population that standard cholesterol-lowering therapies don’t address. Several of these drugs are now in large late-stage trials.
In neurology, new blood-based tests are showing early promise in enabling researchers to identify patients at risk for Alzheimer’s before symptoms appear. The result in both cases is a development pathway defined by the underlying biology, not just the disease label.
Trend three: RNA therapies are expanding beyond infectious diseases
Most people’s introduction to RNA-based medicine came through the COVID-19 vaccines from Pfizer-BioNTech and Moderna. Those programs demonstrated, at a scale never before attempted, that RNA medicines could be designed and deployed faster than any previous therapeutic modality. Less visibly, that moment also compressed the learning curve for an entire field and made infrastructure that had taken decades to build available to a much broader set of programs.
Since then, RNA-based therapies have advanced into oncology, rare genetic diseases and metabolic conditions. One approved drug in this class, Leqvio—a type of RNA medicine called a small interfering RNA, or siRNA—lowers harmful cholesterol levels by around 50 percent with an initial loading dose followed by two injections per year thereafter. In 2025, the FDA expanded its label to allow the drug to be prescribed as a first-line treatment for certain patients, placing it alongside therapies that have been standard of care for decades.
Trend four: Cell and gene therapy is delivering on its potential
For a long time, cell and gene therapies were defined more by their potential than their track record. Early gene therapy trials in the 1990s ended in setbacks that stalled the field for years. It has come a long way since.
The clearest example is CAR-T cell therapy. These treatments, which reprogram a patient’s own immune cells to recognize and attack cancer, went from experimental to standard of care faster than many in the field expected. The FDA approved both Kymriah and Yescarta in 2017, and the approach has since expanded across multiple blood cancers to become a routine part of oncology treatment guidelines.
Gene therapy has taken a longer road. But the evidence that it works, and lasts, is now catching up. Long-term follow-up data for hemophilia gene therapies have shown that single-infusion treatments can provide multi-year benefit, directly addressing the durability question that had shadowed the field for years.
Some of the most notable data from 2025 pushed the field further still. In a phase 1 study, CRISPR Therapeutics reported that a single infusion of a gene-editing CRISPR-Cas9 therapy targeting ANGPTL3 resulted in a large, sustained reduction in harmful cholesterol and triglycerides in patients with difficult-to-treat lipid disorders whose levels remained dangerously elevated despite maximally tolerated therapy. The results, described by the lead study investigator as unprecedented, were published in the New England Journal of Medicine.
Whether that holds up over longer follow-up remains to be seen. The trial enrolled 15 patients with lipid reductions sustained through at least 60 days of follow-up, and longer-term durability data are still pending. Still, the broader picture is encouraging: gene therapy is delivering durable results for rare diseases and showing early promise for common conditions that patients once managed for life.
Trend five: Non-animal methods are on the rise
Drug development has long relied on animal studies to evaluate safety and effectiveness before human trials. While these models have contributed greatly to biomedical progress, they often fail to predict with confidence how humans will respond to a drug.
A growing set of alternatives is beginning to close that gap. Collectively known as non-animal methods, or NAMs, these include lab-grown human tissues, organoids, organs-on-chips, advanced imaging and computational modeling, all designed to replicate human biology more directly. By working with human-relevant systems from the start, researchers can gain earlier and often more reliable insights into how a drug is likely to behave in the body.
In practice, these methods are already reshaping early discovery and preclinical work. Scientists can now grow human tissues from stem cells and use them to model diseases in the lab, allowing them to screen thousands of drug candidates quickly and identify those most likely to succeed. Microphysiological systems, such as liver-on-a-chip or lung-on-a-chip platforms, simulate interactions between tissues and help researchers study toxicity and metabolism in ways that traditional cell cultures can’t. Meanwhile, advances in artificial intelligence and computational toxicology allow researchers to predict drug effects using large biological datasets, reducing the need for animal testing while accelerating decision-making.
Looking ahead, NAMs are expected to play a central role in creating a more efficient and human-relevant drug development ecosystem. Regulatory agencies in many regions are beginning to incorporate validated NAM approaches into safety assessments, and collaborations between academia, industry and regulators are expanding the evidence base for these technologies. Although animal studies will still be required in some contexts during the transition period, the long-term vision is a drug development process that relies primarily on predictive human-based systems. As these technologies mature, they have the potential to reduce costs, improve safety predictions, and bring more effective medicines to patients faster.
The takeaway
Drug development is becoming faster, more targeted and more dependent on getting foundational decisions right early—before the cost of changing course becomes prohibitive.
For smaller biotechs, the possibilities are significant. Programs built on well-defined patient populations, strong biological rationale and smart use of available technology are better positioned than ever to compete with larger players. The companies that will define the next decade are the ones that make the right calls at the beginning, when it’s still possible to make them. That’s what KreaMedica is here to help you do.