Key takeaways
Rising memory costs
Memory supply gap
Memory supply reprioritization
Long-term supply pressure
Why is AI driving memory demand so quickly?
AI workloads require significantly more memory than traditional computing. Memory demand has increased as artificial intelligence systems require large volumes of high-speed memory to train and run models. A single AI server can use as much advanced memory as dozens or hundreds of laptops (Northeastern, 2026). This difference reflects the higher computational requirements of AI workloads, including model training and inference processes.
The difference in usage is substantial. A typical laptop uses approximately 16GB of RAM, while AI systems can require 200GB or more per unit (CBC, 2025). In large-scale deployments, hyperscale data centres may contain thousands of servers, each requiring significant memory capacity. This results in a rapid increase in total memory consumption across the industry. Reports indicate that AI infrastructure expansion is occurring across multiple regions, with major technology companies increasing investment in data centre capacity to support AI workloads. As a result of this growth, global demand for memory chips exceeds supply by around 10 percent (NPR, 2025). This imbalance reflects both the rapid increase in AI-related demand and the existing limitations of global semiconductor production capacity.
Why can’t the supply keep up with demand?
Supply growth is limited, and production is being prioritised toward AI-related memory. The global memory market is concentrated, with Samsung, SK Hynix, and Micron accounting for most global production (Northeastern, 2026). These companies operate large-scale semiconductor fabrication facilities, but increasing capacity requires significant capital investment and extended construction timelines. Reports indicate that new fabrication facilities require multiple years to become operational and involve multi-billion-dollar investments (ACS, 2026).
Manufacturers are allocating production toward higher-margin memory products used in AI systems, including high-bandwidth memory (IDC, 2025). These products are designed to meet the performance requirements of AI workloads and are priced at a premium compared to standard Dynamic Random Access Memory (DRAM). As production is redirected, the volume of memory available for other applications, including personal computers and smartphones, is reduced. Industry data indicates that suppliers are focusing on higher-value segments rather than expanding output across all categories (Sourceability, 2026). This allocation approach results in reduced availability of standard memory products and contributes to supply constraints across consumer and enterprise markets.
Why are prices rising so sharply?
Prices are increasing as demand exceeds available supply. Memory prices have risen significantly since late 2025. In early 2026, prices increased by up to 90 percent compared to the previous quarter (Northeastern, 2026). In some cases, prices increased by up to five times within a few months (BBC, 2026). These increases reflect both reduced supply availability and increased demand from AI-related applications. Retail pricing data shows similar trends. A 32GB DDR5 memory kit increased from approximately $100–200 in late 2025 to $350 or more, with limited availability reported in multiple markets (Tom’s Hardware, 2026). Price tracking data also indicates that DDR4 memory, which typically declines in price over time, has instead experienced increases, with some reports showing doubling or tripling of costs (Tom’s Hardware, 2026).
Memory has also become a larger share of total system cost. In some cases, memory increased from 15–20 percent of total PC cost to 30–40 percent (BBC, 2026). This shift reflects the combined effect of rising unit prices and increased memory requirements in modern systems. Suppliers have also adjusted pricing conditions, including shorter quotation validity periods and allocation-based supply (Sourceability, 2026).
The new reality of the memory market
The memory market is experiencing sustained demand growth and constrained supply. AI adoption continues to increase memory requirements, while production capacity expansion remains limited. Industry forecasts indicate that DRAM and NAND supply is growing at approximately 16–17 percent annually, which is below the level required to meet current demand growth (IDC, 2025). The impact is observable across multiple device markets. Smartphone prices are expected to increase by approximately 14 percent, while global shipments may decline by nearly 13 percent (CNN, 2026). PC vendors are also expected to increase prices by 15–20 percent, reflecting higher component costs (IDC, 2025). Reports indicate that some lower-cost devices have experienced significant price increases as a result of rising memory costs (IEEE Spectrum, 2026).
The market is also shifting toward higher-performance memory technologies. AI workloads require memory types such as high bandwidth memory (HBM) and newer standards like DDR5, which provide higher performance but also increase cost per unit. This transition contributes to higher baseline costs for computing systems and infrastructure. Industry sources indicate that supply constraints and elevated pricing are expected to continue into 2027 and beyond, with some describing current conditions as a long-term structural shift in the memory market (IDC, 2025; SCMP, 2026).
Partner with us to control costs and keep your systems scalable in an AI-driven market!
Let’s connect!
