Skip to main content
The healthcare sector, with its vast proprietary datasets and deep domain expertise, is well positioned to emerge as a long-term beneficiary of AI. Beyond big tech: could healthcare be AI’s next beneficiary?
Lydia Norman AuthorName Associate Fund Manager
Opinion

Beyond big tech: could healthcare be AI’s next beneficiary?

lydia-norman-web.jpg


Associate Fund Manager

  • The healthcare sector, with its vast proprietary datasets and deep domain expertise, is well positioned to emerge as a long-term beneficiary of AI.
  • In drug discovery, AI could accelerate R&D, identify new candidates and reduce costly dead ends.
  • AI is more likely to expand big pharma’s pipeline of potential medicines than displace it, with clinical trials, regulation and manufacturing remaining critical barriers.

Doctor, doctor, what’s next for AI?

First, I must start with a disclaimer: I have rewritten this article many times. Every week, the AI debate evolves, with a new model, opportunity or threat on the horizon. I have wanted to write about AI’s potential in drug discovery for some time, but as the healthcare sector has traded in near-perfect reverse tandem to technology, I’ve struggled to tally my optimistic view of healthcare with the evolution of AI. However, as crowded tech trades roll over and the market wrestles with the difficult questions of monetisation and funding, I believe the time to pivot thinking to the longer-term AI beneficiaries is now.

Throughout my investment career, healthcare has been one of my favourite sectors to follow. Few industries combine discovery, innovation and commercial opportunity in the way that healthcare does, not to mention that successful innovation in the sector can have real world positive impact well beyond financial returns.

Much of the market’s attention so far has been centred on the immediate AI beneficiaries, starting with mega-cap technology and semiconductors, then expanding to power and infrastructure supporting the data centre build-out. The recent selloff of the initial AI-related winners has refocused the conversation back to more fundamental metrics. I think the key question long-term investors should be asking is who stands to gain the most from utilising AI and creating tangible value in the real world beyond the capex spree? For me, healthcare is one of the most compelling answers, and I think the reasoning for this begins with data.

A love letter to proprietary data

In my view, the resounding edge in this new AI world will be domain expertise and proprietary data. Domain expertise is the human element: professionals who can accurately interpret and contextualise the output of AI because they know their specific area of expertise. Then there is proprietary data, which is exclusive and not freely accessible to everyone. As models improve, compete and eventually become less differentiated, proprietary data will be the root cause of why AI can give unique insight and conclusions.

Healthcare as an industry accounts for an estimated 30% of the world’s data1, with vast amounts generated from researchers, medical providers and healthcare companies. Life-science tools have significantly advanced over the past decade, boosting the number of data points (genomics, proteomics etc.) gathered on human biology.

AI’s biggest impact on the healthcare industry will be its ability to sift through vast quantities of data quickly, analysing and iterating on this data at a pace that humans simply cannot. This is particularly significant in the expensive and uncertain process of discovering new medicines.

Drug renaissance or death of big pharma?

Drug discovery has traditionally been an expensive and time-consuming process with uncertain outcomes. AI could improve the process on two key fronts. The first by helping to raise probability of success by accelerating and improving parts of the research and development (R&D) process, in turn reducing the number of costly dead ends companies pursue. The second is helping to identify new drug candidates by trawling through treasure troves of proprietary data for undiscovered gems and revealing patterns that were previously difficult or impossible to identify.

The potential prize is more drug candidates, better-informed decisions about which ones to pursue and, ultimately, the possibility of getting successful medicines to patients faster. But that raises a more disruptive question: if AI reduces barriers to drug discovery, does it lower them enough to displace traditional pharmaceutical companies altogether?

Some have proposed that AI driven start-ups will be able to identify potential novel medicines, test them completely in virtual models, et voila! A blockbuster medicine disrupting the traditional pharmaceutical company is born. However, I think some expectations regarding this outcome need tempering.

Big pharma companies sit on decades of highly valuable, unique proprietary data, and hence might find gems others do not. This does not mean AI start-ups are locked out of the data race altogether. Proprietary datasets can be bought and, increasingly, companies are going beyond the datasets already available and commissioning experiments specifically to generate new ones. In this case, the valuable product of the experiment is really the data file itself. The more advanced and better-funded AI drug discovery companies are going one step further and building their own labs, creating what is referred to as a “lab-in-the-loop”. AI can generate a hypothesis, the physical lab then tests it, the resulting data is fed back into the model and off it goes again.

It’s also true that AI can identify molecules of interest without necessarily having the scientist-driven understanding or domain expertise that has traditionally guided drug discovery. And some may argue this understanding isn’t necessary. Paracetamol is an example that springs to mind (despite its discovery long predating AI) – it’s one of the world’s most widely used medicines and yet, to this day, scientists still do not fully agree on precisely how it works.

But then comes biology. Virtual models can help narrow down the search, and “lab-in-the-loop” approaches can test and refine a hypothesis, but ultimately a medicine has to work in incredibly complex biological systems in real people. This is where I think the AI advantage swings back towards big pharma. Demonstrating that something is both safe and effective requires considerable capital, clinical development and regulatory expertise. AI can make that process better and hopefully faster, but it cannot simply virtualise it away. It’s a useful tool that can help trials better select patients, improve trial design, write reports quicker and then help companies maximise the value of the data they generate, but the ultimate test remains what happens when a potential medicine reaches patients.

The real proof in the pudding will emerge over the next 5-10 years: will more AI-enabled drug candidates reach clinical trials, and does a higher percentage of them ultimately make it to market?

AI as the jet fuel to discovery

Whilst attending a recent European healthcare conference, there was one term surprisingly absent from all conversations: AI. (Winding back to spring, regardless of industry, the hot topic of AI surfaced at every single meeting, usually framed in “disruptor” or “opportunity” prose.) At the conference, the one consistent thread from management teams was that they weren’t focused on externalities and political noise but instead on the factors directly within their control – developing better treatments and building stronger pipelines for future growth. This is where I believe AI will be used within the sector, as the tool that can accelerate the work already taking place.

AI could act as the jet fuel to discovery, lowering barriers at the earliest stages of the process and leading to a rising cadence of drug candidates. But beyond that, it faces real-world limitations: it can speed up and improve the necessary clinical trials but not replace real-world evidence driven by the people it is trialled on; it can assist and automate manufacture but not replace the know-how to scaling complex drug manufacturing. As AI powered start-ups develop or identify a promising novel asset, clinical trials and manufacturing remain clear barriers to getting that drug to market. In my view, larger pharma companies will step in, acquire, commercialise and scale those innovations, rather than be disrupted by them. For big pharma, this is pipeline optionality, not displacement.

Disclaimer

This document has been prepared by EdenTree Investment Management Limited and has been produced for information purposes only. As such the views contained herein are not to be taken as advice or recommendation to buy or sell any investment or interest thereto.

These are the views of the author(s) at the time of publication and may differ from the views of other individuals/teams at EdenTree Investment Management. The views presented are as of the date published.

No forecasts can be guaranteed and there is no guarantee that the information supplied is complete or timely, nor are there any warranties with regard to the results obtained from its use. Edentree is the source of data unless otherwise indicated.

Capital at risk. The value of an investment and the income from it may go down as well as up and the investor may not get back the amount initially invested. Selecting stocks due to our ethical criteria means that the choice of stocks is limited to a subset of the stock market and this could lead to greater volatility.

This financial promotion is issued by EdenTree Investment Management Limited (EdenTree) Reg. No. 2519319. Registered in England at Benefact House, 2000, Pioneer Avenue, Gloucester Business Park, Brockworth, Gloucester, GL3 4AW, United Kingdom. EdenTree is authorised and regulated by the Financial Conduct Authority and is a member of the Investment Association. Firm Reference Number 527473. Phone 0800 011 3821. Email clientservice@edentree.com. Edentreeim.com

Sources

  1. World Economic Forum, “How data-driven digital healthcare tools cut costs and boost outcomes”, 4 January 2024, updated 3 June 2025.