In the past two years, Open AI and its peers have accelerated internal and external growth at a surprising pace. Even the adaptation of its services and products has rapidly expanded. It may be fair to say that what happened with Google search and Facebook over five to ten years is happening to AI within months.
Rapid growth surely comes at a cost, and one of them is energy use. It is believed that a single ChatGPT query uses 2.9 watt-hours of electricity compared to 0.3 watt-hours for a Google search. In a way, as the brain develops, every thought requires eight times the energy.
In 2023, global energy demand grew by 2.2% compared to 2.4% in 2022, but it is still higher than the long-term average of 1.5%. This demand is expected to grow by 3.6% annually over the next three years.
This will be difficult to match from the supply side as sources are limited and setting up plants requires longer timelines. This has resulted in Amazon, Google and Microsoft to look at nuclear power. They have already committed billions of dollars to companies that will build small molecular reactors (SMRs) that would be closer to data centres.
If this happens and most likely will, then these data centres will be a national security site and critical asset. This would indirectly make the companies more influential in the future of the country than compared to now. Whether it will be a positive or negative overall impact on the economy is too early to call as it is in the making.
Comments