This article was featured in the World Economic Forum, written by Schellman’s Jerrad Bartczak, Senior Associate AI, and Stu Block, Sustainability Practice Director.
Conversations around the use of artificial intelligence (AI) in business typically center around data security, ethical uses, and the risk of over-reliance on the emerging technology. But there is also growing attention and concern around AI’s environmental footprint, and rightly so. With AI having a significant impact on energy and water consumption, as well as global greenhouse gas (GHG) emissions, it is vital that the topic of environmental sustainability is included in discussions around responsible AI.
Simultaneously, there’s an undeniable sense of renewed hope that AI can positively contribute to the global transition to a lower carbon economy. As AI’s impact rises alongside its growing use, we must reflect on how it contributes to – and fights against – climate change. In this article, we’ll detail AI’s energy and water demands and the impact it has on GHG emissions, what prominent organizations are doing to mitigate the impact of their AI use, and what you can do about your AI energy consumption.
AI’s Growing Appetite for Energy and Water
Over the past decade, the demand for cloud computing and digital services has focused minds on energy optimization, with modern data centers increasingly adopting energy-efficient hardware and advance cooling systems leading to improved Power Usage Effectiveness (PUE) averages across the industry. PUE measures the ratio of total data center energy use compared to the energy consumed by the computing equipment alone, with the goal being peak effectiveness. However, the reality is that the energy demands of AI, powered by data centers, significantly outpace current efficiency gains.
A single ChatGPT inquiry consumes about five times more electricity than that of a web search, while it is also estimated that training a single language model such as GPT-3 uses electricity equivalent to the power consumed annually by 130 US homes. Then, there’s the energy consumption of AI image generation. While an AI model generating text in response to 1,000 prompts requires as much as 16% of a full smartphone charge, generating an image requires the equivalent of 100% of a smartphone charge.
Beyond electricity needs, AI operations also come with high water withdrawal demands as data centers require cooling systems to maintain optimal operating temperatures. GPT-3 is estimated to utilize roughly one 16-ounce bottle of water for every 10-50 responses it prepares, and this quickly adds up when factoring in billions of queries – with AI’s projected annual water withdrawal to reach 6.6 billion cubic metres by 2027. Lastly, operational logistics, including construction and maintenance, account for up to two-thirds of a data center's lifetime emissions. Emissions from building infrastructure, manufacturing IT equipment, and chip manufacturing requirements of data centers contribute significantly to AI’s environmental footprint.
AI’s Global Environmental Impact
As AI workloads expand, so do data center power demands. The World Bank estimates that AI’s wider industry category of Information and Communications Technology (ICT) currently generates at least 1.7% of global emissions. While this percentage may seem modest, current figures only reflect today’s reported AI usage and not tomorrow’s actual consumption. AI adoption will continue to rapidly accelerate – and considering increased global digital penetration, expansion of cloud storage, internet of things uptake, and the growing prevalence of cryptocurrencies and blockchain technologies – the collective impact on climate change could become significantly larger, even as environmental efficiencies are factored in.
Under current growth trajectories, the International Energy Agency (IEA) predicts global data centers may consume up to 1,000 TWh of electricity in 2026 – an increase of 400% from 2022, driven largely by AI-related workloads. Meanwhile, reports from major tech giants support these projections with Google and Microsoft recently disclosing year-over-year increases in GHG emissions, warning these trends challenge their climate commitments. That said, many businesses are still innovating creative ways to respond to the impacts of their energy use.
Mitigating the Impact of AI Energy Consumption
Many prominent organizations associated with cloud computing and AI development have taken strong stances on tackling emissions and climate change. Notable examples include:
- Amazon has matched 100% of its operation’s global energy consumption with electricity generated by renewable energy systems.
- Microsoft established a supplier code of conduct, requiring suppliers to transition to 100% carbon free electricity by 2030, and has committed to doing the same.
- Google remains committed to their goal to rely on carbon-free electricity by 2030 and has launched efforts to improve their AI model efficiency and data-center energy consumption.
- OpenAI founder Sam Altman invested $20 million in Exowatt, which uses solar power to help meet the needs of OpenAI data centers.
- Salesforce announced that it is lobbying for new regulations to compel companies to report AI emissions data and efficiency standards as part of its Sustainable AI Policy Priorities initiative.
Amid continued commitments and investments in computing infrastructure and AI development, looking to industry leaders can provide inspiration for opportunities to improve energy management and reduce emissions.
What You Can Do About Your AI Energy Consumption
Through their aspirations of carbon neutrality, enterprises may lead the way toward more environmentally conscious practices, yet they’re not doing it alone. As we consider the need to address the environmental impact of AI use as part of global efforts to build a sustainable future, organizations of all sizes can pull their weight in the following ways:
1. Measure emissions and reduction efforts
- Leverage tools to measure, analyze, and report your emissions over time
- Work with suppliers to maintain awareness of how your AI use translates to energy consumption
2. Use efficient AI models and optimize AI workflows
- Leverage pre-trained models and APIs
- Compress models or use batch processing wherever possible
3. Comply with ISO 42001 requirements
- When determining which ISO 42001 standards and risks affect your organization’s AI management system, include climate change as a relevant factor
- Conduct an AI system impact assessment, which evaluates ethical and societal impacts of your AI system, including the effects on environmental sustainability
4. Partner with sustainable vendors
- Utilize AI and cloud platforms committed to renewable or carbon-free energy and ones that publish sustainability metrics and goals
- Seek vendors who are ISO 14001 and ISO 42001 certified, which demonstrates their commitment to improving sustainability performance and using ethical AI systems
5. Foster a company-wide sustainability culture
- Educate and train employees on GHG emissions, energy demands, and water use related to business and AI operations
- Source ideas for sustainability related efforts from employees and create an environmental sustainability employee resource group to lead the implementation of such measures
- Engage with customers and suppliers to collaborate on ways to mitigate energy use and emissions
While these practices help mitigate AI’s environmental footprint, it’s equally important to recognize that AI isn’t just part of the problem — it can also be part of the solution.
AI’s Positive Environmental Impact
Amid the public discourse around AI and its undeniable environmental impact, many argue that it can also be a powerful tool in the fight against climate change – and they’re right. AI can quickly and effectively analyze significant volumes of data to provide insights into climate trends and assess the effectiveness of emissions reduction strategies that can lead to faster action.
Additional capabilities for AI to positively contribute to fighting climate change include:
- Providing greater understanding of extreme weather climate risks
- Modelling climate scenarios
- Improving resource utilization and management
- Expediting the development of new sustainable, energy-efficient materials
- Reducing carbon footprints by forecasting demands and optimizing operations
- Developing product life-cycle assessments to improve the accuracy of supply chain emissions calculations
These factors, amongst others, help scientists, researchers, and experts better understand and address climate change and its implications more than ever before.
Getting Started with Combating AI-Driven GHG Emissions
It’s best practice to maintain awareness of the sustainability expectations of investors, suppliers, and consumers, and to track and report your energy use and GHG emissions. You can start by publishing an annual GHG emissions inventory that is subject to third-party assurance or becoming ISO 14001 certified, which comes with increased stakeholder confidence that your organization is committed to sustainability. Becoming ISO 42001 certified also demonstrates your commitment to using ethical, trustworthy, and accountable AI systems. Most notably, it’s essential that you take appropriate measures to ensure you’re keeping up with evolving and emerging sustainability and AI regulations.
AI’s energy and water consumption and GHG emissions remain a hot topic – be it in the media or with large ICT corporations seeing it impact their ambitious climate goals. From startups to global enterprises, everyone must do what they can to curb the resource consumption and emissions driven by AI use and make efforts to preserve our planet. By learning about the measures being taken from leading businesses and adopting practical strategies, there are actionable steps your organization can take today to align your AI strategies and sustainability goals. The time to act is now.
If you’re ready to learn more about ISO 14001 or ISO 42001 compliance, including the certification requirements, process, and benefits, Schellman can help. Contact us today and we’ll get back to you shortly. In the meantime, discover additional sustainability and AI insights in these helpful resources:
- What are the ISO 14001 Requirements?
- Should You Get ISO 14001 Certified?
- A Global Snapshot of AI Laws and How Compliance with ISO 42001 Can Help
- WEF AI Governance Alliance Briefing Papers: An Overview
- Artificial Intelligence and Cybersecurity: What to Know Right Now
About the Authors:
Jerrad Bartczak, Senior Associate, AI
Jerrad Bartczak is a Senior Associate – AI within the AI Practice at Schellman, based in New York. He specializes in AI assessments including ISO 42001 and HITRUST + AI, while staying current on worldwide AI compliance and governance developments. He also possesses in-depth compliance knowledge cultivated through years of experience conducting HITRUST, SOC 1, SOC 2, DEA EPCS and HIPAA audits. Jerrad maintains CISSP, CISA, CCSFP, CCSK and Security+ certifications.
Stuart Block, Director, Sustainability Practice
Stuart Block has over 12 years of experience advising on sustainability and climate reporting processes, preparing external-facing disclosures, developing impact measurement methodologies, accounting for GHG emissions, setting climate- and human capital-related targets, and implementing environmental sustainability strategies. Prior to joining Schellman as the Sustainability Practice Director, he founded Sustas LLC, a sustainability accounting advisory firm, and was the Director of ESG Reporting and Analytics at Cardinal Health where he was responsible for the ESG reporting strategy as well as the design and implementation of the ESG reporting process in preparation for climate and human capital disclosure requirements. Stu was also a Manager in EY’s Climate Change and Sustainability Services practice where he provided clients across the healthcare, real estate, financial service, manufacturing, oil and gas, and technology sectors with ESG reporting, goal setting, and assurance services in accordance with GRI, SASB, and TCFD frameworks. Stu earned his Fundamentals of Sustainability Accounting credential from the Sustainability Accounting Standards Board (SASB), graduated from Northeastern University's MSA/MBA program, and carries an active CPA license in Idaho.