Navigating the intersect of AI, environment and energy for a sustainable future

This session will examine the current landscape and future scenarios of AI energy consumption and environmental impact, offering a comprehensive overview of how energy, water and material are utilized, greenhouse gas emissions are generated throughout the lifecycle of a typical AI system. Resource demands during various stages, including data collection, model training, inference, and deployment, will be analyzed. Predictions from multiple studies and reports will be explored to discuss the potential escalation of energy and water usage, greenhouse gas emissions driven by the exponential growth of AI applications and data centers. This session aims to provide a holistic understanding of AI's environmental footprint.

AI has the potential to transform completely how we conduct research, education, and business. On one hand, the opportunities are endless, e.g., in climate, health, education, and practically all areas of human life. For example. which action may have the most significant impact on climate? Which subpopulations are most vulnerable to the adverse effects of climate-related stressors (e.g., heat waves, wildfires, tropical cyclones)? However, data centers are energy-intensive facilities, with computational power and cooling being the most energy-hungry processes. Depending on the task, data center servers require substantial energy to perform their computations, and this computing process can generate significant heat. Therefore, extensive energy-hungry cooling systems are often needed to avoid overheating of the hardware of computers and maximize their performance, stability, and lifespan, especially in high-performance systems. There is an increasing concern that the explosion of AI and its electricity demand is slowing down the progress of relying less and less on fossil fuel combustion for electricity generation. In this talk, I will provide an overview of the work conducted in my lab, hoping to shed some light on the controversial role of AI in the fight against climate change.

NVIDIA’s Head of Sustainability, Josh Parker, explores the growing complexity and capability of AI systems – highlighting the need to understand their environmental footprint and their potential to drive sustainable progress. From breakthroughs in energy-efficient architectures to real-world applications in environmental monitoring and resource optimization, we stand at a pivotal moment to shape AI’s trajectory with sustainability at its core. Drawing on recent efforts to quantify and reduce the carbon footprint of AI development, Josh Parker offers a forward-looking perspective on how transparency, innovation, and intentional design can unlock a more sustainable future for all.

This session will explore cutting-edge research focused on energy and environmental saving solutions for AI, highlighting advancements in new chips, innovative AI model architectures & algorithms, and development in frugal AI. The latest developments in environmental-efficient hardware, such as specialized processors and AI chips, as well as software optimizations that enhance computational efficiency, will be discussed. The session will also address green energy solutions designed to meet the growing demands of the AI sector, including the integration of renewable energy sources and other low carbon sources and sustainable data center practices. These technological and environmental innovations will be examined to provide a comprehensive understanding of how the AI industry can achieve greater energy efficiency and sustainability.

AI is advancing at remarkable speed, but behind the breakthroughs lies a growing crisis. Today’s AI systems consume massive amounts of energy and often lack the reliability needed for real-world trust. As these systems scale, current computing approaches are reaching their limits—technically, ethically, and environmentally. This talk makes the case for a fundamental shift: toward next-generation AI computing that is both sustainable and trustworthy. We begin by examining our recent findings that expose key limitations of today’s digital hardware, ranging from excessive energy consumption to fundamental issues of computability and reliability. Grounded in mathematical theory, these insights point to an urgent need for disruptive innovation beyond conventional architectures. This leads us to analog computing, and in particular to neuromorphic systems that emulate the brain’s efficiency. We will focus on spiking neural networks, a central model in this domain, and share our latest theoretical results. These findings suggest a promising path toward a new class of AI systems, which is inherently efficient, mathematically grounded, and fit for the sustainable digital age.

Artificial Intelligence (AI) is poised to revolutionize society, yet its escalating energy demands pose a formidable challenge to its long-term sustainability. The staggering gap in energy consumption between biological (Human Brain @20watts) and artificial intelligence (ChatGPT @100KWatts) is striking. In this talk, I will present a bio-inspired and integrative vision for building intelligent systems that are not only powerful but also energy-efficient and inclusive. Drawing from our recent work in neuromorphic computing and algorithm-hardware co-design, I will discuss how spiking neural networks (SNNs)—a class of AI models inspired by the brain—can offer a compelling alternative. These models leverage temporal processing to improve robustness and dramatically reduce energy consumption, latency, and computational overhead. From a hardware perspective, I’ll examine how memory and sparsity management can accelerate SNNs on general-purpose platforms.

The computations required for deep learning research have been doubling every few months, resulting in an estimated 5,000x increase from 2018 to 2022. This trend has led to unprecedented success in a range of AI tasks. In this talk I will discuss a few troubling side-effects of this trend, touching on issues of lack of inclusiveness within the research community, and an increasingly large environmental footprint. I will then present Green AI – an alternative approach to help mitigate these concerns. Green AI is composed of two main ideas: increased reporting of computational budgets, and making efficiency an evaluation criterion for research alongside accuracy and related measures. I will discuss these two ideas, presenting my latest results on the topic, which provide ways to substantially reduce the computational costs of large language models.

This session will explore the transformative potential of AI to reduce environmental impact in various sectors, focusing on both theoretical insights and practical applications. The ways in which AI technologies can optimize their operations will be discussed. The latest research and real-world implementations will be presented to provide a comprehensive overview of how AI can drive innovation and environmental sustainability in various sectors. This session aims to provide valuable knowledge on the advancements and opportunities AI presents for improving environmental impact across systems and promoting a more sustainable future.

Rising concerns over IT's carbon footprint necessitate tools that gauge and mitigate these impacts. This session introduces CodeCarbon, an open-source tool that estimates computing's carbon emissions by measuring energy use across hardware components. Aimed at AI researchers and data scientists, CodeCarbon provides actionable insights into the environmental costs of computational projects, supporting efforts towards sustainability without requiring deep technical expertise. This talk from the main contributors of Code Carbon will cover the environmental impact of IT and of AI specifically, the possibilities to estimate it and a demo of CodeCarbon.

Generative AI is rapidly emerging as a foundational enabling technology, set to drive the next wave of innovation across sectors. To date, much of the attention has focused on its extraordinary and exponentially advancing capabilities. This talk will shift the lens to the critical dimensions of AI’s future evolution—factors beyond performance that are essential to ensuring its sustainable and democratic future.

AI plays a dual role as both a driver and a challenge for energy sustainability. The exponential growth of data and AI-driven energy consumption calls for sustainable AI development. To address these challenges, a three-tier strategy is foreseen: "Efficient Infrastructure" (optimizing data center efficiency through advanced networking and cooling), "Intelligent Empowerment" (optimizing algorithms through multimodal and other techniques), and "Efficient Implementation" (promoting industry-specific AI applications via collaborative ecosystems). The balance between AI innovation and environmental responsibility requires global collaboration, pragmatic adoption, and human-centric values to harness AI as a force for green progress.

PrevisIA uses artificial intelligence to forecast deforestation in the Brazilian Amazon by analyzing satellite imagery, historical deforestation patterns, topography, water bodies, and socioeconomic data, with a focus on detecting unofficial roads—a key predictor of forest loss. The initiative integrates three pillars: AI-driven road detection via Sentinel-2 imagery, predictive modeling with geospatial dashboards, and collaborative enforcement with state prosecutors. Partnering with Public Prosecutor’s Offices in Pará, Amazonas, Acre, and Amapá, PrevisIA achieved 73% forecast accuracy (2021–2024) within a 4km radius. Its reports enable embargoes, fines, and criminal cases against illegal deforestation, demonstrating AI’s potential for proactive environmental protection.

Join us for the launch of the UN Climate Change Technology Executive Committee’s new technical paper on Artificial Intelligence for Climate Action. Discover how AI can serve as a powerful tool to accelerate climate change adaptation and mitigation efforts in developing countries — including least developed countries and small island developing States. The paper also explores the risks and challenges associated with the use of AI in these contexts. This work is part of the UNFCCC Technology Mechanism’s #AI4ClimateAction Initiative, delivered in partnership with KOICA and UNIDO.

This session will explore the critical policy and regulatory aspects of the interplay between AI and the energy sector. Policies that support both the optimization of energy systems through AI and the management of AI's own energy consumption will be addressed. Key topics include standardization priorities, innovative data center design strategies, and necessary regulatory frameworks. The roles of various stakeholders and a call for action to promote sustainable practices will also be discussed. This session aims to provide support for managing the relationship between AI and energy, ensuring efficiency and sustainability. This session will also highlight the importance of the ICT sector in managing its greenhouse gas emissions, aligning with global sustainability efforts.