With the climate crisis occupying a larger part of the public consciousness, the UN’s Sustainable Development Goals (SDGs) have been created as benchmarks for sustainability and how governments, organizations and individuals can contribute toward preserving the planet for future generations. To help achieve these SDGs, many support the adoption of digital technologies, such as artificial intelligence, to increase efficiency. This is a common rationale for pushing digital transformation projects—that such a move is the answer to reducing an organization’s carbon footprint or other negative environmental impacts.
However, Elin Hauge, a Norway-based technology, AI and business strategist, believes that techno-optimism—or the idea that technology can solve all of humanity’s problems—needs to be tempered with a more realistic point of view. Based on what she has observed through her professional experience, technology can indeed solve some sustainability problems. However, there are also cases where it actually just sweeps the problem under the rug, simply pushing it away to a different location that’s out of sight and out of mind.
Hauge recalls delivering a keynote speech at an event hosted by a company in the electrical grid industry. She says that at the event’s Q&A, one of the questions for attendees was: “What is the most important action your company is taking to meet the UN SDGs?” Many expressed that the answer was the digitalization of products and services. But Hauge felt it didn’t add up because this approach still involves consumption.
“All digital processes require resources, primarily electric power. People tend to think of data as something that’s just floating wirelessly around in the air, especially with the advent of cloud technology,” the strategist says. “But in reality, cloud service providers operate huge data centers—physical buildings that consume large amounts of electricity and water for cooling purposes.”
According to a report by the International Energy Agency, the electricity consumed by data centers, cryptocurrencies and AI was around 460 terawatt-hours, which accounted for almost 2% of total electricity demand in 2022. This amount could potentially double by 2026 as technology continues to evolve rapidly. According to Hauge, these data centers are usually built in areas where operators can take advantage of readily available sources of renewable energy, such as the sun or wind.
While data centers can be beneficial to the economy, the strategist says that they can also negatively impact the surrounding environment. For example, global technology giant Meta abandoned its plan to build what would’ve been the Netherlands’ largest hyperscale data center in the agricultural community of Zeewolde in 2022 after widespread opposition to the project. The facility had been projected to span around 166 hectares of land and consume 200 megawatts of power at full build.
Hauge adds that another major issue is the water consumption of these data centers. Meta is planning to build a data center in Talavera de la Reina, Spain, that’s expected to use around 665 million liters of water annually. This amount is significant, considering the problems with drought in the surrounding region of Castilla La Mancha during recent years. The area—responsible for a quarter of Spain’s grain production—has had restrictions on water consumption and had been expected to lose between 80% to 90% of its grain harvest in 2023.
When assessing the toll that the huge demand for technology takes on the environment, including issues such as rare earth metal mining and electronic waste, the question of how this situation can be remedied looms. “Obviously, we cannot and shouldn’t roll back digitalization,” Hauge says. “But we need to start having more balanced thinking regarding these technologies and stop believing that they are the solution to all our sustainability issues.”
The strategist sees many positive applications of AI but also asserts that people should think about the cost-benefit of implementing it as well as other digital technology projects. For instance, a 2023 study found that generative AI models, such as GPT-3 from OpenAI, can consume around half a liter of water for around 10 to 50 responses. “There are many industrial use cases of AI that I wholeheartedly support, such as lifesaving medical imaging, agricultural optimization or energy grid balancing,” she says. “What I do not support is the haphazard implementation of digitalization and AI just because it sounds cool or to greenwash a business’ operations.”
Hauge believes there should be more awareness and transparency—with the environmental footprint of technologies brought to public attention and incorporated into businesses’ sustainability reporting. “If we are made aware that our choices have a tangible impact on the environment, then it’s up to us to make smarter choices around the use of digital technologies,” she says.
