DRAM prices surge amid AI-driven shortage

As AI infrastructure investment is projected to grow in 2026, tightened supply in the global memory supply chain looks inevitable. Component prices are racing higher, forcing buyers across industries to rethink their sourcing strategies.

Major suppliers are responding with steep contract price hikes and capacity shifts toward AI-optimized components. Meanwhile, procurement leaders face mounting pressure to secure critical memory ahead of further disruptions. The current squeeze is more than a short-term imbalance, but a symptom of a memory market in flux as it reaches a structural turning point.

Samsung’s steep memory price hike

The memory chip shortage has entered a more acute phase as Samsung, the world’s largest memory producer, has raised contract prices between 30% and 60% compared to its September levels. These hikes have been reported by major outlets and further confirmed by downstream buying behavior. For a market experiencing increased constraint driven in large part by hyperscale demand for AI infrastructure, they are a serious sign of urgency.  

Notably, Samsung has delayed its customary October pricing announcement, a move widely interpreted as a signal of tighter-than-expected inventory levels. Buyers have responded swiftly by placing larger or emergency orders in anticipation of further price hikes.  

This surge in buying has led many to pay premium rates for allocation, feeding into a cycle of volatility across the DRAM and NAND markets.  

While AI-related demand is overwhelmingly confined to high-bandwidth memory (HBM), the buildout of AI data centers is creating a secondary surge in DDR5 and LPDDR5 demand. These technologies are essential for servers, GPUs, and accelerators that support training and inference workloads. But they’re also found in many consumer electronics. As a result, non-HBM memory modules are facing heightened competition for limited production capacity.  

This puts Samsung in a unique position. Unlike Micron and SK Hynix which have aggressively prioritized HBM, the South Korean giant has maintained a more balanced focus across its memory portfolio. This slower shift may prove advantageous in the coming quarters as it gives Samsung pricing power in standard memory classes during a period of rapid supply tightening.

Market experts predict the company will raise contract prices another 40% to 50% in Q4 according to Reuters and TrendForce. Ellie Wang, an analyst for the latter, said in a recent statement, “They [Samsung] are really confident that the price is going to increase. And the main reason is that now the demand is really strong, and everyone is working on long-term agreements with the suppliers.”

While some customers may secure inventory under pre-negotiated contracts, those relying on the spot market face mounting challenges. As the market leans further in favor of suppliers, these obstacles are expected to grow in the months ahead.  

Looking forward, Samsung has announced plans to construct a new memory production line in South Korea, but this capacity will not come online for some time. For now, the market remains undersupplied. The ripple effects of that shortage cannot be ignored.  

In the current high-demand environment, Sourceability can help customers secure memory contracts and identify alternative suppliers in the face of uncertain availability. Our team of experts and network of partners, along with demand intelligence tools like Datalynq, can help your organization mitigate dependency on spot markets and secure strategic allocations ahead of disruptions.

Broader impact of the AI-fueled memory squeeze

While Samsung’s recent price hikes have become an obvious signal of stress in the market, a broader reallocation of production capacity is quietly underway. It’s setting the stage for deeper disruption across consumer, industrial, and automotive supply chains in 2026.  

Memory makers have widely adopted the pivot to AI-focused products, but the more concerning development is how this decision is altering global procurement behavior. Across the memory market, panic buying and double ordering are already taking hold. These tactics, last seen during the pandemic-era chip crunch, are reemerging, particularly among OEMs in consumer electronics and automotive.  

Buyers are attempting to secure inventory far in advance, often placing speculative orders to hedge against future availability gaps. Such behavior is distorting demand signals, leading to over-allocation in some sectors and critical shortfalls in others. No matter which side of the spectrum buyers find themselves on, information on which crucial buying decisions are made is becoming obfuscated.  

Even so, the push for aggressive buying is rational given the worsening constraints facing the market. SK Hynix has already booked its entire capacity for memory chips through the end of 2026. Other manufacturers have less plainly suggested they are nearing the same milestone.  

Worse, experts warn the conditions driving this supply crunch in 2025 are only growing stronger as 2026 draws near. Dan Nystedt, vice president of research at TriOrient told CNBC in an interview, “The AI build-out is absolutely eating up a lot of the available chip supply, and 2026 looks to be far bigger than this year in terms of overall demand.”  

SMIC and other manufacturers are already warning of potential shortages for smartphones and vehicles—a consequence of AI-centric capacity reallocation and long delays in bringing more online.  

Even short of a full-blown supply crisis, companies in non-AI verticals face escalating lead times, narrowed supplier options, and rising costs across the board. Analysts warn that this shortage could lead to a prolonged “super-cycle” of memory price hikes rather than a short-term spike.  

Buyers may also soon find themselves stuck holding aged inventory that no longer aligns with design needs while struggling to secure newer-generation DDR5 or LPDDR5 components. This fragmentation adds complexity to inventory management and introduces new lifecycle risks, especially for products with multi-year support requirements.

In this environment, forecasting based solely on historical usage isn’t sufficient. Sourceability helps customers identify and sidestep supply-side stress before it hits production lines using demand forecasting through Datalynq’s forward-looking demand models and supplier risk indicators. Combined with sourcing diversification and agile procurement frameworks, these tools can mitigate the volatility introduced by speculative buying and shifting production priorities.