We Can Digitise and Decarbonise
Latin America can show how to prevent the rise of AI from bringing the fall of net zero, says Rodrigo Abreu, Digital Infrastructure Operating Partner at Patria Investments.
In a much-overlooked initiative, world leaders recently attempted to bring climate change and digital technology together at COP29. But the result, the Declaration on Green Digital Action, downplayed the most significant issue putting technology in opposition with current climate goals – the rise of AI.
In a relatively short time, ChatGPT has spearheaded the adoption of a host of accessible generative AI tools for an almost endless number of corporate and consumer applications, and we are witnessing just the very beginning of it. Large tech companies have committed over US$200 billion[1] in capital investment to AI in 2024 alone, accelerating the development of AI infrastructure worldwide.
Now, AI infrastructure’s insatiable hunger for power is posing a significant dilemma for the tech giants pouring billions into AI technologies: how to reconcile their sustainability commitments with the growing energy needs of their data centres?
The choice between AI and net zero
Part of the issue is that the grids serving AI data centres in most of the large, developed countries still rely heavily on fossil fuels. This holds true even for technologically advanced nations like the US, where the majority of AI data centre capacity is being developed. These data centres already consume an estimated 4% of US electricity, a figure that could double by the end of the decade.[2]
The continuous drive for more sophisticated AI tools and capabilities puts an enormous strain on the ambitious net zero targets set by businesses, countries, and international organisations. As global pressure to decarbonise mounts, AI threatens to overwhelm existing renewable energy capacity in many regions. Microsoft’s bold decision to restart the Three Mile Island nuclear plant to power its data centres shows the magnitude of the challenge and underscores the radical lengths businesses are prepared to go to in their quest for AI expansion.
But we don’t necessarily need to choose between slowing the progress of AI and hitting net zero. It is possible to find a third way – a way in which AI continues to grow, but without significantly growing the consumption of fossil fuels.
A third way
The reality is that this needn’t be an unsolvable dilemma. Global renewable energy capacity has grown by a staggering 415% since 2000 – from 0.8 to 3.9 TW [3] – and it’s a long way away from reaching a plateau. On the contrary, we continue to better understand which regions have the most significant potential to generate vast amounts of renewable energy in solar and wind. While China has led the development of renewable energy capacity in the recent years, other leaders in solar and wind are emerging.
The problem is that the majority of our power-hungry data centres today are not built in locations which can provide easy access to renewables. And even though some methods such as green hydrogen promise a way of ‘exporting’ this capacity, in practice vast quantities of renewable energy are near-impossible to export efficiently. But while renewable energy is difficult to export, new AI data centres, which are voracious consumers of energy, can theoretically be built anywhere.
The new locations where AI can be powered sustainably will require ample renewable energy capacity and infrastructure that can readily meet – or quickly expand to meet – the inflated energy demand. To keep pace with the requirements of the digital age, the ideal locations also need reliable digital connectivity, a stable investment environment, local expertise in the sector and supportive government policies. So, where can we find these?
The Latin American exception
Latin America currently comes closest to fulfilling all of these criteria. It’s the only region in the world where several countries already source most of their electricity from renewables. Brazil sources around 87% of its consumption, and Latin America is approaching 70%[4]. Several countries in the region rely on hydropower as the backbone of their overall power capacity, and wind and solar make up more than a third of the energy capacity in many countries.
The region takes advantage of the abundant natural power sources available, and as a result it has become a global outlier in renewable power generation. Because of this, Latin America has gained extensive experience in developing renewable projects of all shapes and sizes, while also rapidly expanding its capabilities in building and operating data centres. As an investor and developer in both sectors, we can attest the region has the expertise to scale up capacity quickly.
Latin America has all the ingredients to become one of the world’s sustainable data centre capitals. But realising this potential requires significant investment. The region’s data centre capacity is projected to grow roughly three times by 2030 – while this is a significant expansion, it would still only account for less than 3% of the global capacity [5], and should ideally be further accelerated, so the world doesn’t have to rely on heavy fossil fuel consumption instead. The private sector will be vital in directing data centre investment towards green alternatives, looking beyond national boundaries to prevent the rise of AI from becoming the fall of net zero.
The AI COP?
This year will be pivotal for Latin America. Global events that have been taking place in the region, from Group of 20 leadership summit in Brazil to COP16 in Colombia, set the stage for the upcoming COP30 in Belém. This meeting presents a vital opportunity to reset the agenda. It’s a chance to show that COP is still effective at bringing about change and consensus as the urgency to address climate change intensifies and geopolitical divides make global cooperation ever more difficult.
Latin America can lead the global agenda, raise awareness to the urgent demands posed by AI and potential solutions it can offer – and continue building momentum behind data centre investment in the region. Leading the global conversation on this topic will help cement its strengthening position in relation to both technology and energy at a crucial inflection point. As our reliance on AI deepens, we must decide if our reliance on fossil fuels will deepen with it. Knowing it can be avoided, is this really a decision we want to make?
[1] According to Citi Bank’s analysis quoted in The Financial Times ‘Wall Street frets over Big Tech’s US$200bn AI spending splurge’, published 01.11.2024
[2] World Economic Forum, ‘How power-positive data centres can unlock AI’s potential while transforming communities’ published 19.09.2024
[3] Visual Capitalist, ‘Visualised: Renewable energy capacity through time (2000-2023)’ published 18.06.2024
[4] IEA World Energy Outlook 2024 (biomass accounted as renewable), Bloomberg (Latin America Outlook 1H 2024
[5] Altman Solon (2022), InfraStructure Oct-23, McKinsey & Company Report, February 2024; Conservative estimates as sources may consider higher capacities
The post We Can Digitise and Decarbonise appeared first on ESG Investor.