AI is at the beginning of what promises to be an exceptional growth trajectory.
Research from Foundry revealed that 41% of companies say AI is either on their radar screen or they are actively researching the technology. Twenty-one percent say they are piloting new AI initiatives.[1]
As AI takes off, questions around the sustainability of the technology are coming to the fore, given the large amounts of energy consumed by datacentres processing AI workloads. Ensuring AI deployments are sustainable over the long term is therefore a key priority for technology providers and enterprise users alike.
Foundry reached out to the CIO Experts Network, a community of IT professionals and technology industry influencers, to discuss just how AI adoption can be made to work with enterprises’ sustainability principles.
Investing in sustainable solutions
For Ramprakash Ramamoorthy (@ramprakashr), Head of AI Research at Zoho Corporation, the investment choices made by enterprises will be key.
“Sustainable AI transformation begins with selecting energy-efficient models and infrastructure to reduce carbon footprint. Organisations should leverage cloud platforms powered by renewable energy and design AI systems with the intention to minimise unnecessary computation and storage,” he says.
Kieran Gilmurray (@KieranGilmurray), Chief AI Innovator at Technology Transformation Group agrees with this assessment: “As IT leaders, you can drive AI transformation while prioritising sustainability,” he says.
“By making energy-efficient choices from the start, selecting low-impact models, reusing existing systems, and partnering with eco-friendly cloud providers, you can align AI projects with sustainability goals. This approach not only reduces waste and lowers costs but also creates future-ready systems that benefit both business and the planet.”
Technology vendors are currently working to improve the energy efficiency of AI infrastructure to help IT leaders meet their sustainability objectives. AMD, for instance, has achieved a 38x improvement in energy efficiency from 2020 to 2025 for its EPYC CPUs and Instinct accelerators, surpassing its initial 30x goal.[2] The company has now set itself the challenge of enabling a 20x improvement in rack-scale energy efficiency for AI training and inference by 2030, from a 2024 base year.[3]
An enterprise-wide approach to sustainable AI
Making the right technology choices is a primary way in which CIOs can try to make their AI deployments more sustainable.
Ramamoorthy advocates for a joined-up, enterprise-wide approach.
He says: “Embedding sustainability KPIs into AI projects ensures environmental impact is measured and managed alongside performance,” he says. “Cross-functional collaboration between IT, sustainability, and business leaders is key to aligning innovation with ESG goals and by integrating responsible practices from the start, companies can scale AI while supporting long-term environmental goals.”
Javier Campos (@javcamposz), AI Safety Researcher and Group Chief Technology Officer at Peach, echoes these sentiments.
AI-driven tools can enhance ESG strategies, enabling predictive resource management and automated sustainability reporting to cut operational costs, he says.
He adds: “Transparent governance frameworks with standardised energy and carbon metrics ensure accountability, fostering stakeholder trust in AI’s sustainable deployment.”
Tom Allen (@_thallen), Founder of the AI Journal, believes that achieving AI transformation sustainably requires embedding efficiency and responsibility into every stage of the AI lifecycle.
He explains this starts with designing energy-efficient algorithms and leveraging sustainable infrastructure—such as data centres powered by renewable energy.
“Transparency in reporting energy use and collaborating with partners committed to sustainability, like AMD, further amplifies these efforts.”
Allen gives the example of the retail sector, where AI-driven demand forecasting can be optimized to run on energy-efficient hardware, reducing the carbon footprint of large-scale data analysis.
“By adopting edge AI solutions and prioritising resource-efficient machine learning techniques, retailers can improve operational efficiency while meeting their sustainability targets—demonstrating that technological advancement and environmental stewardship can go hand in hand,” he says.
Sustainability and the evolution of AI
What’s clear is that sustainability is going to have an increasingly important impact on the evolution of AI within enterprises in the years ahead. For one, our experts agree that it will help shape the type of models used within business settings.
“Consideration should also be taken to right-size the AI model for the desired outcomes,” says Ramamoorthy. “Using an LLM (large language model) may not always be the right approach and SLMs (small language models) and MLMs (masked language models) can also be considered, which can reduce the computing load required and produce the outcome more efficiently.”
LLMs are AI models with billions of parameters trained on vast volumes of data and are capable of understanding and generating human-like text across a wide range of tasks.
SLMs are smaller, more efficient models with fewer parameters, and are optimised for faster inference and deployment on edge devices. MLMs, meanwhile, are trained by predicting masked words in a sentence and are typically used for understanding tasks rather than free-form text generation.
The rise of agentic AI is another factor that IT leaders will need to keep in mind.
Campos adds that as agentic AI proliferates, organisations can achieve sustainable transformation by adopting energy-efficient algorithms and carbon-neutral infrastructure, leveraging AI to optimise resource use and reduce emissions.
More sustainable AI practices, including modular hardware designs and renewable energy-powered data centres, minimise environmental impact while supporting scalable agentic systems.”
AI enabling a more sustainable future
Foundry’s panel of industry experts make it clear that far from being inimical to the sustainability agenda, when deployed with thought using the latest in energy-efficient infrastructure and models, AI can advance environmental objectives.
Daniel Jacobs (@DJDJaco), Founder & CEO at Starkhorn, summarises this way of thinking: “AI transformation presents a valuable opportunity to promote sustainability by optimising resource utilisation, lowering energy consumption, and fostering more informed decision-making.
IT leaders can take proactive steps by prioritising the development of sustainable data centres and designing AI models that emphasise efficiency “ultimately reducing their environmental impact,” he says. “Embracing these approaches can lead to a more sustainable future for our technology and the planet.”
The key now is for enterprises to coalesce around AI technology ecosystems that prioritise sustainability every bit as much as innovation.
Learn more about AMD’s commitment to sustainability.
[1] Foundry, “From hype to reality: AI adoption gains traction in 2025,” 2025
[2] AMD, “AMD Surpasses 30×25 Goal, Sets Ambitious New 20x Efficiency Target,” June 2025
[3] Ibid