The three major memory chip manufacturers — Micron, Samsung, and SK hynix — are reportedly becoming stricter when it comes to chip orders. Sources tell Nikkei Asia that the companies will conduct more stringent due diligence measures, including confirming the identity of the end-user, the quantity of the order, and even asking if the demand is real. These measures come as the memory chip supply remains tight, with AI chip makers still demanding huge amounts of HBM, taking up capacity that would otherwise be used for consumer memory.
“The three companies became stricter and asked us about who we will supply to, how much quantity, and if the demand of our customers are real,” one executive at a GPU and server supplier told the publication. “Because at times like this, some people will overbook or stockpile more than they need, which would disrupt the market later.”
The AI infrastructure build-out is the primary reason for the global memory chip shortage, especially as memory chip makers prioritize large orders from hyperscalers and AI chip makers like Nvidia. This is especially true given that these AI tech giants aren’t price-sensitive, and they’re willing to pour nearly limitless amounts of cash into whatever hardware they need to stay ahead in the AI race.
On the other hand, consumer electronics are hardest hit, with nearly all models taking the brunt of the supply crunch. This also goes beyond the usual memory modules, laptops, and GPUs that were the first to be affected by the shortage, as even smartphones, TVs, set-top boxes, home routers — practically every “smart” electronic device and even automobiles — will have difficulty sourcing memory going forward.
Relationships matter
The current situation is particularly harsh for smaller companies that cannot afford to pay higher memory chip prices or lock in more expensive long-term contracts. After all, it makes sense for any business to prioritize more lucrative products that will earn them more revenue at their current capacity.
And while it might seem logical for them to scale up and increase production to capture demand for both the enterprise and consumer markets, that is easier said than done. It takes years for a memory production line to get up and running, and given the boom and bust cycle of the memory market, chip makers are hesitant to spend billions of dollars on new fabs.
More than that, memory chip makers have long institutional memories, and they’re less likely to accommodate PC and electronics makers that didn’t act favorably during lean seasons.
“When demand from AI chip developers —companies like Nvidia and its peers — is already several times larger than that of everyone else, leading memory makers simply don’t have the capacity to focus on smaller-brand customers,” a memory chipmaker mid-level employee told Nikkei. “That’s especially true for those [customers] that brutally cut prices and slashed orders over the past few years when end markets were weak. It’s a cruel reality, but relationships with memory suppliers matter in a crunch.”
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.
