Digital health and the climate emergency: four challenges for health policy

Following a panel discussion on Digital Health and Climate Justice, part of a series of events during Right Here Right Now 2025 and involving academics working at the crossroads of health, artificial intelligence (AI) and digital technology and environmental impacts, we want to reflect on climate responsibilities and action in the context of digital health.

We know that climate change is a significant public health threat, with rising global temperatures causing more frequent and intense heatwaves and extreme weather events, with allied increases in food and water insecurity, displaced populations, wider environmental impacts and the spread of infectious diseases. What’s far less understood—and far less discussed—is how our digital choices in health care are part of the problem, and could be part of the solution.

There are many potential benefits and trade-offs in the development and use of current and new digital technologies, digital infrastructures and the wider environmental impacts these bring. We focus on four key areas that we urgently need to integrate into how we think about digital health and medical AI.

First, carbon modelling and why it is not enough. Carbon modelling, or footprinting, is the most common intersection of climate and digital health. It is important that we measure and address carbon impacts. However, it has become a kind of comfort blanket.

When sustainability is mentioned in digital health, it is almost always in terms of how a particular interventions reduces carbon. Take the example of virtual consultations that are frequently cited as carbon-reducing because they avoid emissions from travel. This is true: a recent estimate showed that one virtual appointment can save between 3 and 10 kg of CO₂e, depending on the journey avoided - but consider what is not included in this analysis. The energy used by cloud-based video platforms, the manufacturing footprint of the devices used, the water and power used by the data centres that process the calls, the fact that remote care does not always replace in-person appointments (in-person visits are frequently needed in conjunction with a remote call), and the often significant work that goes into adjusting care pathways and staff training and skills to support the development of such services. In other words, we are cherry-picking carbon savings while ignoring broader systemic impacts.

We need to go further. Carbon is only one lens. While we do need to ensure our carbon calculators are more sophisticated, we also urgently need to think about resource use, biodiversity loss, pollution, lifecycle costs, and crucially, climate justice: who benefits, who bears the cost, and what assumptions are being made about access and equity. Many of the vulnerable communities who have historically experienced health inequalities are disproportionally affected by the climate emergency.

Second is the hidden footprint of digital infrastructure and the need to address this as part of our wider digital transformation efforts. Infrastructure is the part of the digital ecosystem but, as Susan Leigh Star put it, it is ‘sunk into and inside of other structures’, making it invisible to most of us. We do not see it, we rarely think about it and when we do, we tend to think of it as clean – there is no smoke, no fumes, no waste bins - but digital infrastructure has a very real and growing environmental footprint.

Take the example of data centres, the International Energy Agency estimates that data centres consume around 1–1.5% of global electricity, and this is rising rapidly. Many health apps and platforms are hosted in commercial cloud services (e.g. Amazon Web Services, Microsoft Azure, Google Cloud) which rely on massive server farms. Some are powered by renewable energy, many are not. Even when the electricity comes from renewable sources, we need to consider – for instance - the embodied energy in the hardware, water used for cooling, the environmental and human costs of mining rare earth minerals or the short lifespan and disposability of devices.

Smartphones are a good example here – central to many digital health initiatives, they generate 85–95% of their carbon emissions before they are even turned on, through raw material extraction, manufacturing, and shipping. We need to be honest about these costs. We should be choosing partners and platforms that are transparent about their infrastructure (favouring those with credible, science-based net-zero strategies); and digital health solutions that are designed with repairability, longevity and end-of-life recycling in mind.

Third, health systems need to lead by example in aligning digital transformation and environmental sustainability.  Across the UK and globally, health systems are undergoing massive digital transformation, but how often is environmental sustainability included in that transformation agenda?

Take the example of a large-scale electronic health record implementation in a major hospital system. Planning can involve hundreds of people, tens of millions of pounds, and years of procurement, configuration, and training. But when is the environmental impact of the digital infrastructure discussed? Typically not during procurement (though this is starting to change in some settings), not during implementation, and not in on-going monitoring or evaluation. There are exceptions, but typically these kind of digital transformation projects do not ask questions about data storage energy use, device lifecycle and e-waste, server cooling systems, end-user electricity usage, or sustainable disposal or repurposing of outdated equipment.

In short, sustainability is typically seen as a bolt-on, not a built-in. Yet the NHS, for example, is one of the largest institutional carbon emitters in the UK. It has committed to a long-term net zero goal, and we have leadership in this space. However, this is often siloed, sitting aside from mainstream health policy and practice. In many health systems, sustainability roles are under-resourced, and digital teams are rarely trained or expected to think about environmental impact. Let alone lead on it.

Fourth, we need to incentivise developers and suppliers. We need to talk about the companies that drive much of digital health and medical AI. Currently, there are few incentives for these companies to prioritise environmental sustainability. Procurement frameworks typically prioritise cost, speed, and functionality. Regulation focuses on safety. Environmental impact is often not a required field in tender documentation. Designing for low power use, repairability, or carbon transparency requires time and money so, as a health tech start up or an established digital health supplier, why would you prioritise that if no commissioner or investor is asking about it?

Until we change the rules, the market will not reward sustainable choices, and indeed may disincentivise them. We need procurement criteria that require environmental performance data, regulatory and evaluation frameworks that set minimum sustainability standards, research funding calls that prioritise environmentally responsible innovation and system-level guidance that positions green health technology as the future of healthcare innovation.

Health systems across the world are now at a crossroads: digital health and AI technology can be part of the solution to the climate crisis, but only if we stop pretending it already is—and start being honest about the full picture.

Decision makers, providers, patients and the public need to engage with the full story of the impacts of digital technologies, beyond carbon modelling; confront the realities of digital infrastructure, from data centres to device waste; expect more from health systems, embedding environmental thinking into every digital transformation programme; and build the right incentives and policies so that developers, suppliers, and innovators are rewarded and not penalised for putting sustainability first.

This is not just about carbon. It is about justice, responsibility, and the future we are building with our tools. This is not a technical issue, it is a moral one.