I'm trying to understand the recent increase in RAM prices. I've heard that AI technologies require a substantial amount of RAM, but I thought it was primarily video RAM (vRAM) that was important—mainly for loading large models to utilize GPU acceleration effectively. From what I gather, standard RAM comes into play when loading models for CPU processing, which isn't very efficient compared to running them on GPUs. It seems like there's a consensus that the RAM shortage stems from AI, but I'm unclear on the details and would love to clarify what I might be misunderstanding.
3 Answers
Although AI primarily benefits from vRAM, regular RAM is still essential. For high-performance AI applications, it’s typical for systems to have quite a bit of RAM alongside their GPUs. Plus, many organizations hosting AI models need enough RAM to support the high demands of their operations.
It's definitely true that AI is a significant factor driving up RAM prices. Data centers specifically built for AI are consuming a huge amount of RAM, and major manufacturers are focusing their production on fulfilling contracts from these companies, rather than supplying consumer needs. This shift has left a shortage of regular RAM for desktops, pushing prices higher when it is available.
The core issue relates to manufacturing capacity. Since AI workloads are driving demand for vRAM, fabs are prioritizing that, leaving limited capacity for traditional DDR5 RAM. This imbalance is a big reason prices have skyrocketed. Many AI-focused servers require 2-3 times the amount of system RAM compared to their vRAM, further emphasizing the heightened demand in this sector.”,
Exactly! Plus, manufacturers like Micron have pretty much backed out of the consumer market, prioritizing AI contracts instead.