This increased energy consumption raises questions about the sustainability of AI development...
So while many of us are excited about AI, especially once inclusive data sets become the norm, let’s spare a thought for the energy that is/will be required to power AI data centres.
Regular data centres require ~30 megawatts (which can power 15,000-30,000 houses), while AI data centres require ~80 megawatts. This increased energy consumption raises questions about the sustainability of AI development, especially given the current global focus on renewable energy and reducing carbon emissions.
AI data centres require a lot more electricity than normal data centres for the following reasons:
1. AI models are far more complex than regular software thereby requiring a lot more computing power to process tonnes of data in order to make the right decisions;
2. Training and running AI models requires a lot of practice, a lot of data, and a lot of processing power;
3. AI data centres need a lot more servers than regular data centres
In other words, more powerful computers doing more complex work using more equipment means more energy consumption. Could this explain why Sam Altman (CEO, OpenAI) has expressed interest in Fusion Energy which, while still in its early stages, has the potential to provide clean and almost limitless energy? This could therefore, indicate a recognition of the need for more sustainable energy solutions to power AI development.
Subscribe to my TWO Newsletters and receive alerts as soon as new editions are published: (1) Inclusive Leadership in the era of AI and (2) Leading with Emotional Intelligence (EQ)
Comments