United StatesHealthcare system cost inflation is becoming a defining political challenge for rapidly ageing societies in North America, Europe and Asia. Having risen steadily in recent decades, around 10% of global GDP is now allocated to healthcare; in the US, it is almost 17%.1
Could artificial intelligence (AI) be the miracle cure? Signs are encouraging that it could transform the efficiency of the drug development process, accelerate innovation and enable more drugs for more therapeutic areas to successfully reach the market. It is also already being used to enable hospitals to operate more efficiently and effectively.
There are undoubtedly risks, especially given sensitivities concerning healthcare. Longer term, though, we perceive compelling opportunities in the US$10tn global healthcare industry for innovative companies that can safely leverage AI’s potential to reduce the cost of healthcare delivery and improve patient outcomes.
Accelerating drug development
Drug discovery and development is notoriously inefficient: it typically takes a decade from inception for a product to reach the market. Assuming a 20-year patent on new molecular entities, pharmaceutical companies then have about 10 years to recoup their investment and turn a profit.
If AI models can accelerate the preclinical stages of drug development, it could transform the industry’s economics. Blockbuster drugs generate billions of dollars in annual revenues: two or three additional years of sales under patent protection could materially enhance returns on investment.
More effective, AI-led drug discovery could also bring more novel products to market. The number of new molecular entities approved by the US Food and Drug Administration (FDA) has been broadly static over the past three decades, edging up only slightly.2
Source: US Food and Drug Administration, January 2025. Products approved by the Center for Biologics Evaluation and Research, including vaccines and gene therapies, are not included in this drug count.
Subhead: New drug approvals by the US Food and Drink Administration (FDA), by year
Overview: This bar chart shows the number of new drug approvals by the US FDA each year between 1994 and 2024. It shows the breakdown between new molecular entities and biologics (which are medications derived from living organisms and reproduced in living cells).
Overall, this chart highlights how the number of new drug approvals has gradually followed an upward trend over the period, especially since the mid-2000s.
The success of AlphaFold, an AI system developed by DeepMind (which is part of Alphabet), in accurately predicting complex protein structures indicates what AI could make possible.3 A protein’s 3D structure determines its exact functionality in the human body; before AlphaFold, this could only be discovered experimentally in an extremely time-consuming manner.
While there have not yet been any fully AI-designed drugs approved by the FDA, multiple standalone AI drug development companies have succeeded in advancing AI-generated drugs into clinical trials, and Big Pharma has been quick to ink partnerships with them.
The promise of AI plotting a speedier route through drug discovery seems genuine: generative AI drug creation company Absci says it can accelerate the pre-clinical stage from six years to between 18 and 24 months.4 AI seems particularly adept at the target validation and lead optimisation parts of the drug discovery process – essentially interrogating vast datasets to find the right biological target in the body for a specific drug molecule to bind to.
Most large pharmaceutical companies now integrate AI into their drug development process. Among those at the vanguard is Eli Lilly, which has launched an AI platform that offers access to drug discovery models trained on its proprietary molecule datasets.5 The US-listed company is also building the industry’s most powerful AI supercomputer with Nvidia to identify and validate new molecules.6
Focusing clinical trials
The end-to-end costs of successfully developing a new drug have increased from an average of US$1.3bn to US$2.2bn over the past decade.7 Over two-thirds of costs are incurred at the clinical trial stage, when a drug goes into human testing.
If AI can improve the speed and success of trials – currently only around 10% of drugs pass this stage – the industry’s research and development costs per drug could tumble.8 Savings should ultimately enable more drugs to be brought to market, widening patient access and improving health outcomes.
Source: Deloitte, March 2025
Subhead: Average R&D cost from drug discovery to launch (US$bn)
Overview: This line chart shows the average R&D costs, for each new drug successfully brought to market, between 2013 and 2024, based on analysis by Deloitte.
Overall, this chart illustrates how average R&D costs, from drug discovery to product launch, have almost doubled over the period.
The leading clinical research organisations (CROs), which partner with the pharmaceutical industry for drug research, trials and commercial support, are fully embedding AI into their workflows.
In designing a clinical trial, AI is being used by the likes of IQVIA, a US-listed CRO, to identify the sub-group of patients most likely to respond to a drug. It can help define inclusion and exclusion criteria, optimise site selection and dosing, and draft filing submissions.
In one example, IQVIA used predictive analysis and machine learning to analyse vast datasets (including claims, electronic medical records and prescription data) to identify patient populations with Alzheimer’s disease, with 80% precision.9 These findings could help overcome the challenge of identifying and recruiting patients that has historically dogged trials targeting Alzheimer’s.
In parallel, IQVIA is collaborating with Nvidia to automate complex workflows across the therapeutic life cycle using agentic AI.
Regulators are supportive of the industry’s use of AI to improve efficiency. Indeed, the FDA now uses AI within the regulatory review process: its own large language model, Elsa, aims to streamline scientific evaluations, accelerate clinical trial protocol reviews and identify drug manufacturing sites that need inspecting.
Streamlining hospital workflows
Hospitals represent the largest area of healthcare spending in the US, comprising 31% of total expenditure in 2023.10 Financial savings in hospital administrative and operating processes are being targeted through the application of AI to automate revenue cycle management, managing suppliers and patient scheduling.
As well as yielding efficiencies for hospital operators (who often have fine profit margins), AI can be used to help improve patient care and working conditions for healthcare professionals. Burnout risk is a systemic concern, with a forecast shortage of more than 10mn healthcare workers globally by 2030.11
HCA Healthcare, the largest hospital operator in the US by revenues, has worked with Google to develop a ‘Nurse Handover’ app that uses AI to analyse electronic health records and create concise summaries for nurses. By mitigating the need for time-intensive verbal handovers at shift changes, nurses’ administrative burdens are reduced and more time can be devoted to patient care.
Healthcare system costs are meanwhile being lowered – alongside better patient outcomes – through the integration of AI in medical technologies. US surgical robotics company Intuitive Surgical, for example, incorporates AI across its product suite, enabling better visualisation for surgeon training and learning, and helping hospitals better manage their fleets of robots.
AI is also routinely used by diagnostic imaging companies, such as German-listed Siemens Healthineers, to enable more accessible interpretation of CT and MRI scans. This helps to blunt the operational and financial impacts of an undersupply of radiologists.
Leveraging AI’s transformational potential
AI undoubtedly has the potential to drive a step change in healthcare sector productivity, from unlocking drug discovery to more efficient hospital workflows.
Despite this promise, the pace of progress could be relatively slow. Regulators will be thinking about the implications of overreliance on AI, potential biases in patient care from training data, and where liabilities lie if something goes wrong. These risks undoubtedly need to be managed.
Overall, though, early applications of AI indicate how it can demonstrably advance the quality and accessibility of healthcare. Though disruptive for some business models, we believe that opportunities will be created for innovative companies that leverage AI’s transformative potential.
1 OECD, 2025: Health Statistics 2025
2 Mullard, A., January 2025: 2024 FDA approvals. Nature
3 AlphaFold has been cited in more than 35,000 academic papers and earned its founders the 2024 Nobel Prize in Chemistry for protein structure prediction
4 Absci, 2023: Absci First to Create and Validate De Novo Antibodies with Zero-Shot Generative AI
5 Lilly TuneLab is a collaborative platform created to offer access to AI / machine learning tools leveraging Eli Lilly’s own drug discovery models
6 Eli Lilly, 28 October 2025: Lilly partners with NVIDIA to build the industry’s most powerful AI supercomputer
7 Deloitte, March 2025: Be brave, be bold – Measuring the return from pharmaceutical innovation
8 IQVIA, 2024: Global Trends in R&D 2024
9 IQVIA, 2026: AI engineered for life sciences & healthcare
10 KFF, 2025: Key Facts About Hospitals
11 McKinsey, 2025: Closing the gap on the healthcare workforce shortage
References to specific securities are for illustrative purposes only and should not be considered as a recommendation to buy or sell. Nothing presented herein is intended to constitute investment advice and no investment decision should be made solely based on this information. Nothing presented should be construed as a recommendation to purchase or sell a particular type of security or follow any investment technique or strategy. Information presented herein reflects Impax Asset Management’s views at a particular time. Such views are subject to change at any point and Impax Asset Management shall not be obligated to provide any notice. Any forward-looking statements or forecasts are based on assumptions and actual results are expected to vary. While Impax Asset Management has used reasonable efforts to obtain information from reliable sources, we make no representations or warranties as to the accuracy, reliability or completeness of third-party information presented herein. No guarantee of investment performance is being provided and no inference to the contrary should be made.