Etron Chairman Lu Chao-Chun has a blunt way of describing today’s DRAM market: he says major memory makers have effectively become the industry’s “Santa Claus.” In other words, as DRAM shortages continue to squeeze supply, customers are increasingly “grateful” simply to receive any allocation at all—an unusual shift that highlights just how much leverage top suppliers now have when deciding who gets memory, and how much.
The current memory crunch is also tied to a wider “memory supercycle,” fueled largely by explosive growth in artificial intelligence hardware. This surge hasn’t just been a talking point for chipmakers—it has rippled through the entire PC ecosystem, disrupting availability and pricing across multiple categories and putting pressure on everything from laptops to graphics cards.
According to Lu, part of the reason the supply chain is so tight today goes back a few years. During the pandemic era, demand for memory products dropped to very low levels, and manufacturers didn’t prioritize expanding production capacity. When market conditions improved, suppliers focused more on repairing margins and recovering lost profits than on building out new capacity. That decision is now colliding with today’s reality: demand has skyrocketed, but adding meaningful new DRAM output is not something the industry can do overnight. Building and ramping advanced memory manufacturing can take years.
A major driver behind the spike in DRAM demand is high-bandwidth memory (HBM), a specialized type of DRAM used in modern AI accelerators. As AI chip developers and hardware companies race to secure HBM for new products, competition for supply has become intense. Compared with conventional DRAM, HBM typically requires more memory modules per system, which amplifies overall demand even further. As a result, a large share of today’s DRAM capacity is being pulled toward AI-focused production.
The downside is that mainstream consumers often feel the pain most. When memory supply is constrained and pricing stays elevated, it can affect the cost and availability of everyday devices—PCs, laptops, and GPUs included—because many products rely on stable DRAM supply to hit target specs and price points.
As for how soon this eases up, the outlook isn’t especially comforting. Industry expectations suggest DRAM constraints could persist until 2027. If that estimate holds, the next couple of years may continue to be challenging for both manufacturers and buyers, with “allocation” remaining a key word across the memory market.
