The Next Medical Revolution Has Already Entered the Waiting Room

Somewhere today, an artificial intelligence system is helping design a new drug, warning a hospital that a patient may be developing sepsis, or showing a surgeon where cancerous tissue may still be hiding. It works quickly, processes more medical data than any human could reasonably absorb, and has never once complained about being on call during a holiday weekend.

Artificial intelligence has already moved into the practical work of medicine. Dozens of AI-designed or AI-influenced drug candidates have entered human trials, several have produced encouraging early results, and FDA-authorized tools are reaching hospitals, operating rooms, pathology laboratories, and clinical research programs. The technology remains young, and most experimental drugs will still have to survive the long and unforgiving journey through human biology. Biology, after all, has never shown much interest in meeting a quarterly deadline.

That journey has historically been expensive, slow, and filled with failure. According to the FDA, developing a medicine can cost between $600 million and $2.7 billion, take roughly a decade from the first human trial to approval, and carry a success rate below 10%. Researchers identify a biological target, design possible compounds, test them in progressively larger trials, and sometimes discover years later that the treatment does not work well enough or creates unacceptable side effects.

AI may give scientists a more efficient way to navigate that process. It can analyze enormous biological datasets, identify promising targets, design molecules, interpret medical images, find eligible trial participants, and detect patterns that may escape the human eye. We appear to be in the early industrial phase of this transformation: the first reliable machines are running, useful products are coming off the line, and the factory remains far from fully built.

The First Medicines and Medical Tools Are Arriving

• Rentosertib—idiopathic pulmonary fibrosis. This progressive disease scars the lungs and makes breathing increasingly difficult. Insilico Medicine used AI to identify the TNIK target and help design Rentosertib; encouraging Phase 2a results led to the initiation of a Phase 3 trial in July 2026. The drug remains investigational, although reaching Phase 3 marks a major milestone for an AI-generated medicine.
• GB-0895—severe asthma. Generate:Biomedicines used generative technology to engineer an antibody for people whose asthma remains uncontrolled despite conventional treatment. A global Phase 3 trial is recruiting approximately 786 participants and is testing injections given only once every six months. Anyone who has tried to remember a daily medication can appreciate the appeal of a twice-a-year schedule.
• REC-4881—familial adenomatous polyposis. People with this inherited condition develop large numbers of gastrointestinal polyps and face an extremely high lifetime risk of colorectal cancer. Recursion used AI-driven cellular imaging to identify the treatment strategy, and preliminary Phase 1b/2 data showed a median 43% reduction in polyp burden after 12 weeks among the evaluable patients. The study remains small and early, though the results provide an encouraging example of AI uncovering a possible treatment path for a disease with no approved drug therapy.
• ISM6331—advanced mesothelioma. This AI-designed cancer drug targets proteins involved in tumor growth and treatment resistance. In July 2026, the FDA granted the program Fast Track designation for certain previously treated patients with advanced pleural mesothelioma, giving the developer more frequent access to regulators as clinical testing proceeds. Fast Track does not guarantee approval, because even the FDA has learned to avoid guaranteeing anything involving a committee.
• ArteraAI Prostate—prostate-cancer risk assessment. This FDA-authorized software analyzes digital biopsy images to estimate a patient’s long-term risks of cancer spreading or causing death. Physicians can combine those estimates with traditional clinical information when deciding among surgery, radiation, and active surveillance, potentially helping patients avoid care that is either too aggressive or insufficient for their actual risk.
• TREWS—earlier detection of sepsis. Developed at Johns Hopkins, TREWS reviews electronic health records for warning signs of sepsis, a dangerous response to infection that can worsen rapidly. A prospective study across five hospitals found that timely clinician interaction with the system was associated with faster treatment and lower in-hospital mortality. The computer still needs doctors and nurses to act, which is reassuring for anyone who has watched a printer refuse to cooperate for no apparent reason.
• Claire OCT System—breast-cancer surgery. During a lumpectomy, surgeons need to determine whether cancer may remain near the edge of the removed tissue. The FDA approved Claire in March 2026 as a real-time imaging system with an AI feature that highlights suspicious areas while surgery is underway, giving the physician additional information before the patient leaves the operating room. Its intended role is to supplement the surgeon’s judgment and other established methods of assessing the tissue.
• AIM-NASH—fatty-liver disease trials. Evaluating liver biopsies in MASH trials has traditionally required multiple expert pathologists, and their assessments can vary. In December 2025, the FDA qualified AIM-NASH as its first AI drug-development tool, allowing pathologists to use it when scoring biopsies for clinical-trial enrollment and treatment outcomes. The pathologist retains responsibility for the final interpretation, while the software helps make the process faster and more consistent.

The Medical Factory Is Still Being Built

These examples cover several parts of the medical system: target discovery, molecule design, antibody engineering, cancer prognosis, hospital monitoring, surgical imaging, and clinical-trial measurement. Each application removes friction from a different stage. Over time, those individual improvements could compound into shorter development cycles, earlier diagnoses, better-selected treatments, more efficient trials, and higher probabilities of success.

The larger transformation will come as AI converges with robotics and more powerful forms of computing. Surgical robots are already giving physicians greater control and precision during minimally invasive procedures. Medtronic’s Hugo robotic-assisted surgery system received FDA clearance for adult urologic procedures in December 2025, adding another major competitor to a field that is expanding rapidly.

AI may eventually serve as a kind of operating-room co-pilot, helping identify anatomy, guide instruments, anticipate complications, and alert surgeons when a movement approaches a danger zone. I will explore that subject in an upcoming blog, “Scalpels, Robots and the Operating Room of Tomorrow.” Despite the title, I am told the robots will not demand reserved parking spaces—at least during the introductory period.

Quantum computing could push the frontier even further. IBM and Moderna have already demonstrated the use of quantum hardware to model the structure of an mRNA sequence, an early step toward understanding how these machines might help researchers study complex molecular interactions. Significant technical hurdles remain, although hybrid quantum and AI systems could eventually help scientists design vaccines and treatments that are extremely difficult to model on classical computers.

That will be the subject of another forthcoming blog, “Quantum Medicine: Designing the Drugs Today’s Computers Cannot See.” I promise to explain quantum computing without requiring readers to understand how something can be in two places at once. Most parents already encounter that phenomenon whenever they ask who left the lights on.

The trajectory is increasingly visible: faster drug discovery, more accurate diagnosis, treatments selected for individual patients, more precise surgery, and clinical trials that require less time and fewer wasted resources. Progress will arrive unevenly, and plenty of promising ideas will fail along the way. Successful innovation rarely travels in a straight line, particularly when the road runs through government regulators, insurance companies, hospital systems, and the human genome.

For investors and families making long-term financial decisions, these developments matter beyond the excitement surrounding any single company or product. The opportunity spans biotechnology, medical devices, semiconductors, cloud computing, data infrastructure, laboratory automation, and the companies supplying the tools that make modern research possible. Identifying the eventual winners will require patience, diversification, and a healthy respect for valuation.

As always, please call or email us if you would like to discuss how these medical advances may affect your financial plan or how we are positioning portfolios around the companies enabling this transformation. AI can design molecules, spot tumors, and warn doctors about sepsis; it still cannot explain why an old 401(k) contains six funds that appear to own exactly the same stocks. Fortunately, we remain available for that.

Your capital, our expertise, a bespoke creation.