Major sustainability milestones and events are on the horizon. The fall will bring Climate Week NYC and COP29, while 2025 marks a major escalation in reporting requirements, with the first CSRD deadline in July. As business leaders follow news and developments, they will note an emerging topic: the connection between AI and sustainability. It’s worth a look beneath the headlines to get a better understanding of this dynamic as well as how it will affect companies in the years ahead.
This just in: AI could destroy the world
Anyone committed to sustainability and progress in the fight against climate change has likely been unnerved by the steady barrage of stories on technology’s effect on energy demand. A sampling offers an idea of the hue and cry:
- Power consumption by data centers from AI could increase by 200 terawatt-hours per year from 2023 to 2030.
- Google’s investments in AI and generative AI have caused it to fall behind in its efforts to reach net zero by 2030.
- Every photo posted on Instagram is equivalent to turning on a light bulb forever.
Innovation—particularly in generative AI—is continuing at a breakneck pace, but each advancement requires massive energy usage. It can feel like the tech world is hurtling forward in an effort to win the AI arms race without a fulsome discussion of the broader consequences, including energy usage. These headlines may be a way of sounding the alarm and applying the hand brake. But are they accurate?
As I was starting to lose hope, I stumbled upon research from the Center for Data Innovation that provided cause for optimism. Its analysis called out some of the errors in past calculations that resulted in gross overestimates of energy demand by tech innovation. Some of these erroneous findings were picked up and republished without question. Before long, they were a foundational element of the AI narrative.
So while there’s broad consensus that AI will increase energy demand, with an added burden on power infrastructure, the magnitude of this jump is up for debate.
Check that: AI could save the world
As justification for near-term increases in energy demand, tech champions often trot out the potential for AI to actually solve the thornier challenges related to climate change. While some use cases seem realistic and scalable—for instance, AI’s ability to reduce the energy consumption of buildings by analyzing data captured from the industrial Internet of Things—others can feel more like sci-fi greenwashing.
However, this recent story about a Silicon Valley start-up that discovered a massive copper deposit in Zambia by using AI to analyze geological data shows these new technologies can indeed be a game changer. Copper is a critical mineral in the energy transition, and the Zambia discovery is enough to power 100 million electric vehicles. Notions of technology (including quantum computing) saving us all might not be so far-fetched.
How companies can respond
The uncertainty related to AI and sustainability creates strategy and communications challenges for business leaders. Beyond continuing to monitor the evolution of AI and its applications, companies can take the following actions.
Support regulations that promote transparency about the energy use of AI
Too often, generative AI and other applications are treated as shiny new toys. After ChatGPT’s release, millions of users experimented with the tool, often with frivolous queries. Some of these users may be unaware that the development and use of generative AI requires massive amounts of energy. Each ChatGPT query, for example, consumes 25 times more energy than a Google search.
Concealment—in which the impact of actions is kept out of sight—helps to lessen the perception of the broader costs. Factory farming is one example: The tacit acceptance by many in society of its tangible benefits (cheaper meat) may be made easier by the fact that the real costs (significant damage to the environment and animal cruelty, among others) are largely hidden.
A clearer accounting of AI’s impact on energy demand could better inform a range of decision-makers. One issue: Regulations on tech generally lag behind the pace of innovation, sometimes by years. So as advancements in AI hurtle forward, legislators will need to improve their speed in grasping the implications of innovation and passing the necessary safeguards. Pressure from companies could reinforce the stakes.
Get ahead of sustainability and ESG reporting
AI’s growing environmental footprint highlights the difficulties in calculating—and mitigating—scope 3 emissions. Most companies will not naturally think of the energy requirements of AI data centers as being relevant to their own emissions profiles. Yet CSRD will require companies doing business in the EU to report on scope 3 emissions, and the first submission deadline is July 2025. And with more companies falling under the remit of CSRD every year and the US Securities and Exchange Commission debating new reporting requirements, the issues related to AI energy usage are unlikely to remain hypothetical for long.
Organizations need to make sure they have a plan in place for how they will calculate their AI-related environmental footprint. Doing so will help them meet regulatory requirements and also gauge how AI is currently being used, identify risks and opportunities, and inform conversations with suppliers and vendors.
Provide workers with information and guidelines on the judicious use of AI
Whether or not companies have data about their own current AI-related emissions, leaders can promote the environmentally responsible use of AI by amplifying information to their workers and encouraging them to be more thoughtful with their queries. As an additional step, organizations could craft formal guidelines for the use of generative AI tools, with a focus on efficiency and value creation. These actions might seem like painting around the edges, but they can make a difference if adopted more broadly. Widespread behavior change is possible: Travelers in the EU have become much more conscious of the environmental impact of air travel, for instance, and this awareness has led to different transportation choices. Meanwhile, countries such as France have banned short-haul flights and promoted the use of rail as an alternative mode of transportation.
AI and its impact on climate change presents an interesting dynamic: rapid, seemingly constant advancements along with a good dose of ambiguity. As executives monitor the shifting regulatory landscape, they can do themselves a favor by looking beyond the headlines and sharing clear information and direction with their workforce to inform better decision-making. It’s also never too soon to gain a better understanding of how AI usage will affect ESG reporting.