Inflection Points and AI’s Physical Impact

Beyond energy consumption, the expansion of AI infrastructure raises concerns about land use and community displacement. Reports from MIT and EY highlight that discussions around Gen AI often focus on energy use but overlook its physical footprint. “We talk about energy consumption, but we don’t always discuss the impact on local communities—the land required for data centers and infrastructure.” As AI scales, ensuring equitable land use planning and addressing displacement concerns will be a vital part of the conversation.

Advancements in Sustainable AI Innovations

Doug Ross, CTO at Sogeti, part of Capgemini, detailed recent advancements that improve the efficiency of AI models while reducing energy consumption. He introduced the concept of a ‘mixture of experts,’ a technique that activates only relevant neurons for specific tasks, significantly reducing overall energy consumption. “Imagine a brain with all synapses lit up versus only using the neurons needed for a specific task—this is how a mixture of experts can optimize AI processing,” Ross explained.

Ross highlighted applications in key industries such as energy, supply chain management, and healthcare. “In supply chains, AI is improving efficiencies by predicting and prescribing solutions that shorten cycle times and reduce transportation costs,” he explained. AI can also assist in forecasting supply chain disruptions due to weather events, ensuring that businesses can adjust their inventory and logistics operations efficiently. In healthcare, AI is enhancing early detection and response in oncology, reducing delays in patient management, and improving response times. “AI can act as a first-level responder, helping oncologists focus on more complex cases while reducing inefficiencies in patient care,” Ross noted.

Balancing Energy Needs with Efficiency

One of the central debates in sustainable AI is balancing the increasing demand for computing power with the push for greater efficiency. Ross acknowledged that while AI models require significant energy, ongoing innovations—such as small language models (SLMs) and hyperscalers that optimize processing—are reducing energy costs and improving system efficiency. “We are moving toward more powerful small models that can deliver competent answers in specific domains while consuming less energy,” he said. These solutions allow companies to scale AI operations while keeping sustainability goals in check.

AI in Sustainability Planning and Long-Term Strategy

AI is increasingly being integrated into sustainability planning and pricing, but there is often pressure for quick and singular solutions. Ross pointed out that sustainability efforts should not be approached with short-term, reactive measures. “There’s a tendency to want a ‘quick fix,’ but real change requires a long-term plan—20 years in the pipeline,” he stated. Businesses that prioritize financial targets over sustainable planning may struggle to implement AI-driven efficiency in a meaningful way.

Ross emphasized that reducing cycle times in business processes is one of AI’s most effective ways of enhancing sustainability. “It’s not just about saving energy in one sector—it’s about reducing operating costs across industries and optimizing resources at scale,” he said.

Transparency, Public Perception, and Accountability

The growing need for transparent AI reporting, particularly in ESG compliance and corporate sustainability strategies, is a key consideration. AI is increasingly integrated into ESG reporting, but ensuring accuracy and accountability remains a challenge. While AI can help process large data sets quickly, concerns about misinformation and bias persist. Ross suggested that regulatory frameworks and technical innovations, including voluntary accountability standards, could help organizations navigate these challenges. “Larger enterprises are already shifting toward more efficient AI models that align with financial incentives and sustainability goals,” he noted.

Ross also referenced the ‘FTX’ framework—Financial, Trust, and Experience—when evaluating AI adoption. “If organizations can achieve the same trust level and user experience for a fraction of the costs and energy consumption, they will choose that model,” he explained, reinforcing the importance of AI’s role in efficiency-driven decision-making.

Sustainability and the Future of AI

As global discussions around sustainability continue, upcoming events such as the World Economic Forum and the UN Global Platform will focus on generative AI’s role in addressing climate challenges. Ross expressed optimism about the potential of AI-driven solutions to benefit environmental justice communities and enhance resource efficiency. “We are seeing major innovations in energy-saving technologies, and there’s a growing awareness of AI’s role in sustainability,” he said. These discussions will be critical in shaping policies and best practices for the responsible development of AI technologies.