How Sustainable Is the Demand for AI?
The demand for artificial intelligence–backed services is surging, and so is the need for computing power and data storage to support them.
Computing power—essentially how fast a machine can perform calculations—is heavily dependent on hardware, particularly electronic chips. Furthermore, the complexity of these calculations is ever increasing, surpassing what current computers can offer. This has spurred interest in the next wave of computing, commonly referred to as quantum computers, though that falls outside the scope of this article.
Furthermore, artificial intelligence remains highly dependent on the quantity of data available, as more data generally leads to greater accuracy in results. Consequently, the demand for data—and the storage infrastructure to house it, whether on-premises or in the cloud—is continuously growing.
These characteristics of AI service demand pose several real-world challenges, particularly a voracious appetite for energy to power the systems and for water to cool the hardware. This can cause significant issues for communities impacted by data centers location, as they compete for these vital resources.
It is estimated that a large data center can consume millions of gallons of water per year. Energy consumption has also sharply increased, especially with the rise of large language models that require substantial computational resources for training.
The manufacturing of computer chips also relies on natural resources, including rare earth elements. While chips are primarily made of silicon, other materials such as tantalum and gold are used in CPUs, GPUs, and capacitors. Recently, rare earths have been at the center of geopolitical tensions between nations leading AI development and have contributed to armed conflicts in the countries where these materials are mined.
There are various efforts underway to create more energy-efficient data centers. These include integrating renewable energy sources such as solar, wind, and geothermal, as well as adopting advanced cooling techniques that reduce the strain on water resources.
