Siemens Q&A: How Artificial Intelligence can be Sustainable

Artificial intelligence (AI) is rapidly transforming industries and society, but its environmental impact is a growing concern that demands attention. 

Globally, data centre electricity consumption is projected to more than double by 2028, reaching 857 TWh – equivalent to the entire electricity consumption of some countries.

Siemens Smart Infrastructure, a division of tech conglomerate Siemens, creates technologies that aim to transform the world.

The €19bn (US$20bn) business is focussed on building technologies, electrification and eMobility. 

Thomas Kiessling is Chief Technology Officer at Siemens Smart Infrastructure, an expert in areas including AI, analytics and clean energy.

Since joining the company in 2021, he has focussed on developing businesses in commercial building decarbonisation, energy and transport as a service.

Thomas shares his expertise on AI with Sustainability Magazine.

What is the environmental impact of AI?

While AI offers immense potential to optimise operations and reduce costs, its growing energy demands have the potential to undermine the very sustainability goals that many industries, particularly those in renewable energy, strive to achieve.

From predictive maintenance in manufacturing to optimising energy consumption in smart grids, AI-driven solutions are transforming industries. However, these advancements come with energy and resource costs. The computational power required to train and run AI models is substantial, and as the complexity of these models increases, so too does their environmental footprint. 

For instance, deep learning models, which are among the most advanced forms of AI, require extensive computational resources that lead to higher energy consumption. This is particularly true for AI models that process large datasets or perform complex tasks, such as natural language processing or image recognition. 

Data centres are high energy consumers, not only for computing, but for cooling systems to maintain optimal operating temperatures. The reliance on data centres powered by non-renewable energy sources exacerbates the environmental impact, leading to increased carbon emissions.  

Beyond the immediate energy consumption, there is the issue of resource depletion. The hardware that powers AI—servers, processors, and storage devices—relies on rare earth elements and other non-renewable materials. AI’s water consumption is also an often-overlooked aspect of its environmental footprint.  

What’s the most important action that can be taken to mitigate AI's environmental impact?

First, and foremost, organisations should look to leverage renewable energy. One of the most effective ways to mitigate the environmental costs of AI is to decarbonise the grid and power data centres with renewable energy. 

Companies can either source renewable energy directly or invest in carbon offset programmes to balance their energy consumption. This approach reduces carbon emissions and aligns AI deployment with broader sustainability goals. 

What other measures are key for implementing AI sustainably?

To minimise the environmental impact of AI, industries need to adopt a more strategic approach to its deployment. First by prioritising efficiency. Not all AI applications are created equal. Some require vast amounts of computational power, while others can achieve similar outcomes with much less. By prioritising efficiency in AI deployment and choosing algorithms and models that are less resource-intensive, businesses can reduce energy consumption and environmental impact. 

Optimised data centre design is also essential for improving their energy efficiency. This can be achieved through advanced cooling technologies, better server utilisation and the use of energy-efficient hardware. Additionally, data centres should be in regions where renewable energy is abundant, further minimising their carbon footprint.

Continued investment in AI research and development is also crucial to create more energy-efficient models and algorithms. This includes exploring new approaches to AI that require less computational power and developing hardware that is both powerful and energy efficient.

Finally, organisations should be looking to implement AI governance. AI governance frameworks can help them to monitor and manage the environmental impact of their AI initiatives. By setting clear guidelines and performance metrics, businesses can ensure that their AI deployments are both effective and sustainable. 

How can AI be used to support the environment?

It’s important to recognise that AI also has the potential to drive positive environmental change. For instance, in the context of transportation (think electric vehicles (EVs)), AI can be leveraged to manage the complexities of e-mobility systems, enabling more efficient and environmentally friendly operations. By optimising charging strategies based on real-time data, AI can reduce energy consumption during peak hours and contribute to grid stability. This lowers operational costs and supports broader sustainability goals by reducing the carbon footprint of EV fleets. 

When AI is used to manage the broader aspects of energy management in e-mobility services it can predict potential issues in EVs earlier by analysing historical data and recognising patterns. This then leads to more timely maintenance and less energy-intensive operations. Additionally, AI can analyse vast amounts of traffic and environmental data to optimise traffic patterns, thereby reducing congestion and the associated emissions. 

In industries with complex supply chains and significant energy demands—such as manufacturing, logistics and transportation AI can streamline operations, reduce waste and improve resource efficiency. By analysing real-time data, AI can identify inefficiencies in production processes, suggest adjustments and monitor compliance with environmental regulations. This leads to a reduction in both energy use and waste production, directly contributing to environmental sustainability. 

As industries increasingly rely on unstructured data such as images, text and videos, AI becomes essential in making this data accessible and actionable. In the context of industrial transformation, AI’s ability to process and analyse large volumes of unstructured data allows businesses to uncover insights that can drive innovation in sustainability.

For example, AI can be used to categorise service tickets based on text, allowing for more efficient routing to engineers and quicker resolution of issues, which in turn reduces the overall environmental impact of maintenance operations.

The key to leveraging AI while maintaining environmental responsibility lies in balance, requiring a nuanced approach that considers both the benefits and the costs of AI deployment. AI has the potential to revolutionise the way energy is produced, managed and consumed, but it must be deployed in a way that supports and not undermines environmental objectives.

Learn more about this topic in a workshop on Technology and AI in the Low Carbon Economy at Sustainability LIVE Net Zero on 5 March 2025, tickets available here.

Make sure you check out the latest edition of Sustainability Magazine and also sign up to our global conference series - Sustainability LIVE

Sustainability Magazine is a BizClik brand