Join on WhatsApp
Join on Telegram
Rabat – Companies like Samsung Electronics and SK Hynix have had to shift their production capacity to high-bandwidth memory (HBM) chips intended for AI workloads that will respond to the explosion in the demand for AI-enabled data-centres and server infrastructure.
That shift has made it harder to find less popular forms of memory, like DRAM and DDR5, which are used in daily devices like smartphones and PCs. Experts in the memory chip business call this dynamic a “super cycle,” according to Reuters.
Because the manufacturers are moving production to HBM chips, there are fewer wafer spaces available for regular memory. Analysts report that DRAM inventories are shrinking, and buyers are responding by placing multiple orders and panic-buying to ensure supplies. This behavior with inventories is intensifying the squeeze on regular memory chips.
Prices are surging as a result. A recent report shows that increased margins are beneficial for manufacturers, but electronics companies downstream face higher costs, and some are already passing these costs on to consumers.
The impacts go beyond memory modules. The lack of components threatens industries that rely on regular memory chips, like servers, PCs, and smartphones.Â
Additionally, the supply shortage comes at a time of geopolitical tensions and a lack of investment in older node production lines, which makes it harder to immediately increase output.
Several large tech firms are investing heavily in AI infrastructure, making the bottleneck even tighter. In 2025, deals like Oracle’s partnership with AMD and Microsoft’s expanded work with Nvidia highlighted how companies are racing to secure chips for their growing AI operations.
Outlook and risks
While the current cycle favors memory-chip producers, analysts warn that the upswing may be temporary. The industry often shifts between strong demand and oversupply. Some experts warn that today’s scarcity could develop into excess supply if corporations invest too much.
On the investment side, top suppliers that have profited from the surge are being monitored to see how they plan to remain in business in the long run. For instance, Samsung’s shift to HBM is currently benefiting margins, but some investors are worried that the limits of conventional memory could make AI components less competitive in the future.
“The extreme pessimism turned into extreme optimism for Samsung. We have to wait and see,” said Albert Yong, a managing partner at Petra Capital Management, a Seoul-based Samsung investor, as reported by Reuters.
Higher memory chip prices could cut into electronics makers’ profits or raise the cost of their products. On the commercial side, companies that use memory modules for servers or high-performance computers have to deal with longer lead times and higher procurement risks.
The rise in chip demand driven by AI isn’t just for advanced AI processors; it’s changing the entire memory-chip ecosystem — from the back-end production to the devices people use every day. The pressure is strong, the stakes are great, and the way producers and consumers react now, will change the chip market for years to come.
Read Also: Nvidia Commits up to $100 Billion in Partnership With OpenAI to Build Massive AI Infrastructure
