The price of DDR5 RAM kits has shot up drastically in a very short period of time, effectively making them stupid expensive, costing as much as gaming consoles or a GPU. Unsurprisingly, this dramatic spike is causing a crisis among PC gamers, and is directly affecting the sales of corresponding motherboards.
According to Japanese news outlet, Gazlog, the rise in DDR5 RAM prices have also reduced the sales of motherboards by approximately 50%. This has supposedly forced motherboard brands like ASUS, Gigabyte, and MSI to revise the pricing of their components in an attempt to keep sales flowing.
The RAM Shortage is Deliberate. Refusing To Pay These Prices Now Will Save Us All A Fortune
byu/Arbiter61 inpcmasterrace
The main issue for motherboard makers is two-fold: on one hand, you’re getting PC gamers who are upgrading from older system – folks that are upgrading from DDR4 systems or older – while others are coming in fresh and have little choice but to purchase and build modern systems that basically support DDR5 RAM.
The issue with the steep price hike on DDR5 has gotten so bad, there is now a call-to-arms on Reddit, asking gamers to boycott buying RAM completely, in the hope that this will reset the pricing. Alas, this rallying cry is unlikely to do any good for the community. For one reason, many gamers tend to adopt a “bite the bullet” or “pull the trigger” attitude when it comes to PC gaming: if they really want to build a system here and now, then by the Gods, they will build that system, exorbitant price tags be damned.
Bitten By The AI Bug, Driven By The AI Craze
The reason why memory, let alone DDR5 RAM kits, has seen its prices shoot up to nearly costing as much as a PS5 or RTX 5070 in a short time is that companies with massive datacentres have already gobbled up the world’s supply of memory chips.
The sale of memory and DDR5 RAM represents a tiny fraction of total sales for memory makers. The bulk of their sales literally comes from the patronage of heavy industry, enterprise, and datacentres. As for the reason behind that, its all because of AI. Simply put, AI datacentres have slightly different requirements than traditional datacentres.
To put it simply, while AI datacentres require GPUs to run, they also require massive, stupid amounts of memory to train their Large Language Models (LLM) in parallel processing.
The pricing issue has also seemingly affected GPU brands like NVIDIA. Supposedly, the GPU may no longer bundle video memory with its GPU dies for its AIB partners. In other words, it’s leaving its partners to source for its own graphics memory modules instead.
