AI’s Hidden Cost

AI, or Artificial Intelligence, is hitting all the headlines at the moment and not all for good reasons. Companies are racing to build larger, faster, smarter models and data centers but at what cost? There is so much attention put on what AI can do, but far less is paid to the other side – the resources consumed, the facilities needed, the hardware required and environmental impacts. This is having a huge impact on more than just the immediate AI spheres.

At the heart of all AI processes, are GPUs (or Graphics Processing Units), these devices were originally designed for gaming but are now an essential part of AI (required both for training and running these large language models). Over the past few years, it has got harder and harder to buy these chips at a sensible price for anything. Not only are GPUs becoming over priced, storage (hard drives) and memory are also being affected.

This increase in general consumer hardware affects everything, schools, businesses, consumers – simple devices you wanted to purchase have increased in price. Several manufacturers have started increasing base prices for hardware, others have added “memory surcharges” to try and show how much of a change the prices are having.

Looking beyond hardware, data centers are being constructed across the globe to run these new AI services, each one requires massive amounts of land, power and slightly unexpected water. The increase in demand here, also directly impacts consumers – energy prices are increasing to cover the increased demand as more power has to be generated in traditional means (fossil fuels). Water, used for cooling means there’s more pressure on finite reservoirs.

A balance has to be found between the insatiable growth of AI and its demand for resources, and the general person who has to pay for increases across everything. Otherwise, this technology push