By any measure, the increase in power demand as a result of AI is staggering, creating complications for digital infrastructure investors and technology companies’ decarbonisation strategies. The International Energy Agency forecasts that data centre-related electricity consumption will roughly double from 2024 to 2030, a surge that is creating localized pockets of grid strain that is leading everyone from hyperscalers (i.e., Microsoft, Google, and Amazon) to the fast-growing AI and technology companies that depend on their services to reevaluate how to reconcile their consumption of energy with their ambitious decarbonisation targets. In many grid regions, data centre appetite cannot be readily met through available resources, and there are yearslong queues to bring new renewable resources online.
With intense interest among private markets investors in companies powered by AI, it’s as important as ever to understand the implications of the current AI surge in setting company strategy.
For infrastructure investors, the coming years are poised to see significant growth in digital assets. Staying ahead of the curve in offering digital services powered by clean energy stands to differentiate one provider’s data centres in a competitive market. The standards for sustainability leadership in data centres continue to become more rigorous; whereas a data centre whose electricity consumption was covered under a renewable virtual power purchase agreement (VPPA) may have been viewed as leading in the past, hyperscalers and others are increasingly moving toward 24-hour carbon-free energy, or clean energy that matches needs on an hourly basis. Achieving this in a constrained market has led to some surprising headlines, such as Microsoft signing a 20-year power purchase agreement (PPA) for nuclear energy at the notorious Three Mile Island site in Pennsylvania. However, there are myriad approaches at the disposal of digital infrastructure developers to meet clean energy goals, whether through clean energy microgrids, efficient computing, or others.
For private market investors investing in AI and technology companies, it’s not too early to begin considering how companies’ platforms are powered. While tech companies’ Scope 1 & 2 operational emissions may generally be at the lower end compared with other industries, their Scope 3 value chain emissions may be driven in large part by their digital footprint. Some of the hyperscalers they depend on for digital services have validated near-term decarbonization targets under the Science-Based Targets Initiative (SBTi), but not all do. For example, Amazon’s target was dropped in 2023, complicating technology companies’ abilities to lean on Amazon or other digital services suppliers to support them in meeting their own decarbonisation target-setting ambitions. Given the patchy coverage on decarbonisation target setting, tech companies should be considering digital services providers' decarbonisation and renewable energy approaches alongside other factors during the procurement process. In addition, there may be other levers for tech companies to consider, such as efficient code or carbon optimisation applications (as offered by Microsoft Azure). A holistic view of these considerations throughout the investment cycle stands to futureproof investments and differentiate against peers as awareness of AI’s greenhouse gas implications rises.
Written by Eric Bloom,
COO & Head of Client Solutions, Re:Co