Editor’s Note: Today’s post is by Rachel Martin. Rachel is the Global Director of Sustainability at Elsevier, where she is driving transformational change in the priority areas of sustainability and SDGs, with a focus on climate action. This post summarizes the discussion in the recent EASE/STM webinar, exploring the digital carbon footprint of scholarly publishing.

From the laptops used to write and submit journal articles to the servers that host and preserve the content online, digital technology is woven into every facet of modern academic publishing. Yet, the environmental impact of such activities can often be overlooked. This may be attributed in part to our largely invisible online scholarly information infrastructure that, compared to physical paper, makes it much harder to see the impact of digital publishing on the planet.

Last week, global leaders met in Belém, Brazil (November 10 – 21, 2025), for the 30th UN Climate Change Conference (COP 30), to find a pathway forward in attempting to address climate change, especially as concerns grow over the long-term environmental impact of digital technologies like generative AI. While facilitating and communicating scientific research remains essential for empowering communities and governments to act, we must also take responsibility for our own footprint. Understanding and measuring the digital carbon footprint of scholarly communication is a crucial first step in addressing how the production of digital knowledge contributes to climate change and in taking meaningful action to reduce its impact.

Publishing’s core purpose has always been to communicate scholarly research. From around the 17^th Century onward, this has primarily been through physical forms such as books and printed journals. The real shift came in 1989 with the internet’s arrival, enabling digital publishing. Publishers have been trailblazers in adopting and embracing new technologies. The first online-only journal, Postmodern Culture, was launched back in 1990, years before Google launched their search engine. These early journals started building the foundations of our digital infrastructure, such as persistent DOIs, integrating new content types such as audio and video, and, of course, enabling wider access, including to those with visual impairments. In our present day, all journals have a digital format with cloud platforms, search engines, and content delivery networks forming a vast, invisible infrastructure supporting scholarly communication.

As our sector continues to transform, there’s a growing awareness that our digital publishing systems have a carbon footprint. In simple terms, the infrastructure powering academic communication consumes energy that, in turn, generates carbon emissions. That means every click, upload, and server request draws energy from power grids that are often still reliant on fossil fuels. The more energy used, the more emissions. But, compared to a physical copy of a journal, an online article can be accessed by thousands, spreading the energy impact over many readers and making the overall footprint more efficient over time. This is one of the reasons why print is still more impactful on the climate than digital. It is also the reason why the carbon footprint of physical books and journals from most academic publishers is one of the highest emission categories.

Calculating the exact carbon impact of an online journal is complex, as it requires us to draw a boundary around what exactly to include or exclude in a digital carbon footprint. For instance, would we include the energy used by the Large Hadron Collider at CERN when publishing the article detecting the size of the Higgs Boson? Or would we include the electricity use of a smartphone or laptop that was used to access and read the article? This is important as this data can help inform actions that be undertaken to improve energy efficiencies and reduce emissions. Here, we need to ensure we take a pragmatic approach and include activities which generate emissions from publishing activities. These would include activities such as energy used when organizing and conducting peer review, but would exclude activities that would have occurred even if publishers didn’t exist, such as the energy used in a lab.

There is already an industry-wide, science-based methodology for calculating the environmental impact of digital media products. Through a coalition called DIMPACT, a comprehensive methodology was published that outlined the sources of emissions and assumptions needed to make such calculations back in 2022. Yet, while this works well for services such as streaming content, it was complicated to apply to our specific styles of digital products — journals.

To address this challenge, a working group of academic publishers under STM created and launched the first Digital Carbon Calculator tool for journals. Based on the DIMPACT methodology, the free tool simplifies the DIMPACT calculation to offer publishers and editors an overview of the digital carbon footprint of their journal or portfolio of journals (see Figure 1 below). By highlighting areas of greatest impact, the tool supports smarter, more sustainable decisions that enable publishers to deliver on their climate commitments. While we are still a ways off from having individual journal carbon reporting standards, this is a first step towards a shared methodology for measuring and reporting on digital emissions.

As we begin to understand the carbon footprint of digital journals, a new demand on energy is emerging: Generative AI (GenAI). Increasingly integrated into scholarly workflows for tasks like summarization, this too will have an associated carbon cost as it integrates into our invisible scholarly infrastructure. Initial research is showing that training large AI models demands substantial energy and water, and as technology companies scale their GenAI capabilities, we see broader biodiversity impacts around land use for new and expanding data centres also emerging. For publishing specifically, the current emissions from GenAI remain and are currently relatively small compared to those from digital and traditional print publishing. However, its rapid growth poses a significant risk of increasing energy and water consumption if not carefully managed. Industry groups like DIMPACT and STM are actively exploring methods to measure this impact as the technology expands. Meanwhile, the scientific community is beginning to offer valuable insights on minimizing GenAI’s environmental footprint, such as reducing the size of the data needed to train the models.

Ultimately, the trusted research that is published has the power to empower local communities and drive meaningful climate action, and it is our responsibility to do so in ways that protect the planet. Striking a balanced approach between innovation, pragmatism, and responsibility have been the foundational principles of STM’s work. Their carbon calculator is a critical first step toward this goal, helping publishers identify where emissions are highest and make informed decisions to reduce their footprint. But sustainability in publishing is a collective responsibility that requires transparency and continuous improvement. By working together and embedding sustainability into the digital future, in line with the principles of the SDG Publishers Compact, we can create a responsible publishing ecosystem that is climate conscious. We hope that the world’s leaders returning from COP 30 discussions will embrace a similarly pragmatic approach.