Right now, anyone who has been using any sort of technology, or has been on the internet has probably come across what many are calling the "RAMpocalypse." The purpose of this article is to help readers understand why RAM prices have increased so dramatically and what role Artificial Intelligence has played in it.
Before looking at the economics of the story, it is important to understand the technological core behind it. Even a year and a half ago, RAM was one of the few PC components people rarely worried about when budgeting a build. Whether DDR4 or DDR5, RAM prices were nowhere near as concerning as GPUs, despite AI already experiencing significant growth.
The core problem is that AI training models and the device you are reading this article on do not use memory in the same way. This phenomenon has been labelled the "HBM Siphon Effect." Traditional RAM connects to a CPU through the motherboard, whereas High Bandwidth Memory (HBM) stacks memory chips vertically and places them directly beside powerful AI processors such as NVIDIA GPUs. This provides the enormous data bandwidth required to train Large Language Models (LLMs) such as GPT-4.
This brings us to RAM production. The global memory market is dominated by three manufacturers: Samsung, SK Hynix and Micron. Producing HBM requires significantly more manufacturing resources than standard DRAM because of its complexity.
As AI companies began offering billion dollar contracts, manufacturers shifted a significant portion of their production towards HBM. This reduced the supply of consumer RAM, resulting in noticeable price increases for products such as DDR5 sticks.
Social media also heavily reacted to the situation. Users joked in the comments of AI-generated videos with remarks such as "RAM costs ten times more because of this, by the way..."
Analysts indicate that in the final quarters of 2026, RAM prices are expected to go further up by 30% to 40%.