Powering the AI revolution will require a dramatic increase in global energy consumption. Can we scale up sustainably? Our COP29 team investigate below.

Entering the age of AI

Artificial Intelligence (AI) continues to drive the global conversation in the tech sector and increasingly features in our day-to-day use of technology. AI has immense potential to improve our lives, but also demands vast amounts of power through the data centres that support the cloud-based technology. By 2030, it is projected that data centres will consume 4% of the global electricity supply.

AI systems have a myriad of different uses. ‘Generative’ programs with the ability to learn from vast quantities of data could have a wide array of benefits. From assisting in research and data analysis, transforming our transportation system towards automation, and potentially detecting and monitoring diseases quicker and more efficiently, are just a small sample. However, there are challenges. 

A step change in data centre energy demand

Already an energy-intensive sector, the advent of AI has caused a surge in the amount of energy needed to sustain the digital economy.

For example, a single request to an AI chatbot is estimated to consume about 10 times more energy than a regular Google search. This requires large-scale build-out.

Sectoral leader OpenAI recently shared ambitions of building several 5-GW data centres throughout America, with each unit consuming a similar amount of power as the entire country of Ireland.

This enormous power demand, as well as the fear of falling behind on their carbon reduction targets, have forced architects of hyper-scale data centres to revisit old ‘new’ ways of procuring carbon-free electricity. In short, they are going nuclear. 

AI attracts finance and innovation

Microsoft recently spearheaded the embrace of nuclear power with their decision to fund the revival of the ‘Three Mile Island’ nuclear power plant in Pennsylvania. Amazon has also recently signed an agreement with ‘Energy Northwest’, a consortium of public utilities in Washington state, to develop four ‘Small Modular Reactors’ (SMR) by the early 2030s, as well as investing in the X-Energy startup dedicated to developing new advanced nuclear reactor designs.

To strike while the iron is hot, tech giants have raised spending levels by 63% since last year. Despite investors raising concerns that their big payoff could be further down the line than previously expected, executives seem to prefer being too fast rather than too late to the nascent AI boom. 

Ireland’s role in the AI revolution

The emergence of AI systems has profound implications for Ireland, which is well-established as a European hub for data centres. There are currently 82 data centres in the Republic, with 14 under construction and another 40 approved. Their impact on the electricity grid has been dramatic and continues to be contentious.

Data centres accounted for about 5% of Ireland’s electricity consumption in 2015 but has now risen to over 20% and is expected to reach 30% by the end of the decade. As a result of the ensuing strain on the electricity grid, an “effective moratorium” was placed on new data centre development in Ireland in 2021.

In recent months, there have been persistent calls for Ireland to lift these planning restraints, amidst a fear that Ireland could lose out to competitors across the water, as several countries see opportunities to capitalise on the burgeoning AI transformation.

Data centres can continue to bring economic prosperity to Ireland, but not unless the country solves its fundamental problems of electricity generation and transmission. How will Ireland balance a national grid currently experiencing a significant “generation deficit”, heavily reliant on gas power for over half of its generation, with the need to increase grid capacity to cater for the fledgling AI revolution?

Secondly, how can this be done sustainably, driven by zero emissions energy technology, and allows the country to achieve the targets laid out in the Climate Action Plan?

Ireland has the potential to lead decarbonisation efforts in the data centre sector. Noting our venture capital ecosystem, STEM university talent pool and the already established data centre-specific construction and engineering knowledge base in the country. Ireland was the first to the party in the of building “traditional” data centres.

But together, with state and private sector finance, can we pave a path forward which allows us to scale up our grid and generation capacity, grasp the AI moment, and meet net-zero by our 2050 target? 

Insights from COP29

Saturday, 16 November, was the thematic day for science, technology, innovation and digitalisation at COP29. As expected, the topic of AI was central to much of the day’s discussion.

A major highlight was the first-ever Digitalisation Day at a UN climate conference, which underscored the transformative potential of digital technologies in combating climate change.

The COP29 Declaration on Green Digital Action received endorsements from over 1,000 stakeholders, emphasising the role of AI and digital innovations in reducing greenhouse gas emissions and enhancing climate resilience.

Discussions also focused on optimising energy consumption in data centres through AI algorithms, which can significantly lower their carbon footprint. The declaration highlighted the need for resilient digital infrastructure and the importance of digital inclusion, particularly in developing countries.

Furthermore, the AI for Climate Action Innovation Factory and the ITU AI/ML in 5G Challenge showcased innovative AI applications to enhance sustainability and resource efficiency. Additionally, the development of advanced AI tools for the agricultural community was announced to improve farming practices and sustainability.

Other discussions featured the role of tech to combat key climate concerns, in particularly the urgent need to reduce methane emissions with UNEP's head emphasising the importance of the Methane Alert and Response System (MARS) to identify and address major leaks.

Leaders also endorsed a declaration to leverage digital technologies for accelerating climate action while minimising the carbon footprint of tech manufacturing and addressing e-waste. It is more than likely that AI will be a key feature also in these technologies.

Given Ireland’s position as a European hub for data centres, Saturday’s discussions were of particular importance to Irish markets. The emphasis on reducing methane emissions and leveraging AI for sustainability aligns with Ireland's goals to decarbonise its energy-intensive data centre sector.

The innovative solutions presented at COP29, such as platforms optimising cloud storage efficiency and AI tools for agriculture, offer potential pathways for Ireland to enhance the efficiency of its data centres and digital infrastructure, thereby reducing their environmental impact and supporting the country's climate action targets.

Furthermore, the focus on digital technologies and AI for climate action underscores the importance of Ireland's tech sector in contributing to global sustainability efforts.

In conclusion

The discussions and initiatives at COP29 highlight the critical role of AI and digital technologies in addressing climate change. For Ireland, these insights provide valuable guidance on how to balance the growth of its data centre industry with sustainability goals. By adopting innovative AI solutions and prioritising emission reductions, Ireland can continue to lead in the tech sector while meeting its climate commitments.

For further insights and to discover how KPMG can assist your business in adopting cutting-edge sustainable solutions, contact our award-winning sustainability team today.

Contact our COP29 team

Discover more in ESG