Memory Component Ratio in Router BOM Rises from 3% to 20%, Telecom CAPEX Under Pressure
Global memory semiconductor prices having surged more than 600% in one year has put emergency pressure on telecommunications companies' broadband equipment procurement strategies. In particular, consumer network equipment such as routers, gateways, and set-top boxes are taking direct hits.
According to 'Memory Price Tracker' published by market research firm Counterpoint Research in February, consumer memory prices rose more than 600% over the past year as DRAM and NAND production capabilities concentrated on high-margin server applications due to the surge in AI server demand.
Since January 2025, consumer DRAM and NAND prices have risen more than 6x, and increases of approximately 90% were also tallied in Q1 2026 compared to Q4 2025. The report analyzed that price increases are likely to continue through June 2026. The main causes cited are the explosion in HBM (High Bandwidth Memory) and server DRAM demand. This is because manufacturers are prioritizing production capabilities for high-margin server products.
The most heavily affected area is broadband routers. While smartphone memory prices rose approximately 3x over the past 9 months, broadband equipment using consumer memory rose nearly 7x. Optical fiber routers, Fixed Wireless Access (FWA) CPEs, and set-top boxes are representative examples.
The particular problem is changes in BOM (Bill of Materials) structure. The memory cost proportion within routers has surged from around 3% a year ago to more than 20% currently. This means that beyond simple component price increases, the very cost structure of equipment is being reorganized. OEMs that did not preemptively secure volumes or small equipment manufacturers with low negotiating power are placed in even more vulnerable positions.
2026 is a year when global telecommunications companies are accelerating broadband infrastructure expansion. FTTH (Fiber to the Home) expansion, FWA-based coverage expansion in rural and peripheral areas, and introduction of high-performance CPEs with built-in AI computing functions are major plans. However, the surge in memory prices is causing complex risks including increased procurement costs, delivery delays, and disruption of AI CPE strategies. In particular, as AI functions are added, memory requirements increase and cost burdens inevitably grow further.
The report mentioned the possibility of prices peaking in the first half of 2026, but also forecasted that supply-demand imbalance may structurally persist given that production line conversion and investment expansion are difficult to achieve in the short term. This means supply constraints may continue even after short-term price peaks.
Telecommunications companies are expected to restructure their response strategies. Checking OEM supply chain security, renegotiating long-term purchase contracts, BOM redesign to low-capacity memory-based equipment design, and adjusting AI function phased adoption are being discussed. Large telecommunications companies can secure volume priority, but small and emerging telecommunications companies are more vulnerable to rising cost pressure.
This situation is a case showing how AI demand is reorganizing resource allocation in the semiconductor industry. As high-margin AI server-use memory is allocated first, consumer electronics and telecommunications equipment is being pushed to secondary priority. The 'chain effect' of AI infrastructure expansion pressuring other industries' supply chains is being realized.
In 2026, competitiveness in the telecommunications industry has increased likelihood of being determined not only by network technology capability but by component procurement capability. Memory is no longer a simple component but is emerging as a strategic asset that determines the pace of broadband expansion.
