Summary of Key Points
The surge in demand for AI computing power has completely disrupted the storage industry's decades-long “three-year cycle” pattern. Storage giants such as Samsung, SK Hynix, and Micron have shifted most of their wafer production capacity to high-end memory (HBM) dedicated for AI use, leading to a severe shortage of consumer-grade memory (e.g., DDR5), which has caused prices to skyrocket (with some products increasing by 3-4 times). This has had a direct impact on the consumer electronics industry, resulting in higher costs, price increases or reduced specifications in products, and declining shipments. However, high-end products like AI PCs are experiencing growth despite these challenges. Meanwhile, domestic storage companies have an opportunity to replace imported products, but it will be difficult to reverse the high-price trend in the short term.
1. Wild Price Increases in Consumer-Grade Memory
The price of memory modules has risen dramatically. Data from German physical retailers shows that the cost of mainstream DDR5 memory has increased by 340% compared to July 2025 (for example, a module that used to cost 100 euros now costs 440 euros). A set of 2x32GB DDR5-5600 memory went from 530 euros in December last year to 677 euros in January this year, with a monthly increase of 147 euros—more than the original price of the entire set (140 euros) in July!
Not only new memory prices have risen, but older types like DDR2 and DDR3 are also seeing increases. The contract price for DDR2 is expected to rise by 55-60% in the second quarter and another 35-40% in the third quarter. Even NAND flash memory (used in solid-state drives) is not spared, with contract prices expected to increase by 70-75% in the second quarter of 2026. This price surge is not due to traditional supply and demand fluctuations in the consumer electronics market but rather a structural shortage caused by AI’s demand for storage capacity.
2. AI Competing with Consumer Electronics for Storage Capacity
Why is there such a shortage of consumer-grade memory? Because storage giants are dedicating all their production to HBM (high-end memory) for AI use. The demand for memory in AI servers is 8-10 times that in traditional servers, making each AI server almost like a “memory-consuming monster.” Global shipments of AI servers are expected to increase by 180% in 2026, and the three major storage companies (Samsung, SK Hynix, and Micron) control over 90% of DRAM production capacity. They are allocating 70% of their new production to HBM, even reducing the capacity for consumer-grade memory.
For example, SK Hynix’s HBM orders have already stretched into the next few years, with all products sold out for 2026, and they have signed a partnership with NVIDIA extending until 2030. With HBM accounting for 23% of global DRAM production capacity, the supply of consumer-grade memory has naturally decreased—it’s like the bread that was originally intended for everyone now being taken by AI.
3. Difficulties in the Consumer Electronics Industry
The rising cost of memory is putting significant pressure on smartphone and PC manufacturers:
- Smartphones: Average prices are expected to increase by 14% to $523 (a record high), but shipments will decline by 12%. Manufacturers are hesitant to raise prices too much, so they are reducing specifications: models priced at around 3000 yuan are no longer available with 1TB of memory; some entry-level models priced under 1000 yuan have been discontinued.
- PCs: HP reports that the cost of memory has increased from 15% to 35% of the total component cost. Mainstream gaming laptops (e.g., Lenovo’s拯救者Y9000P) have seen prices rise from around 10,000 yuan to 15,000-16,000 yuan. Global PC shipments are expected to decline by 14% to 255 million units in 2026, but AI PCs are growing, with shipments expected to increase by 39% to 150 million units due to the demand from businesses and gamers for high-end configurations.
4. Strategies to Cope with High Prices
The industry is finding ways to adapt:
1. Manufacturers Adjusting Their Approaches:
- High-end models (e.g., AI PCs) will continue to use large amounts of memory and see price increases.
- Mid-range models will reduce specifications (e.g., lower refresh rates, smaller batteries) to maintain prices.
- They are also signing long-term price-locking agreements with storage companies to avoid fluctuations in spot market prices and using software optimizations (e.g., memory compression) to reduce reliance on physical memory.
2. Domestic Storage Companies Seizing the Opportunity:
- ChangXin Technology has started mass-producing a full range of DDR4 and DDR5 memories, reaching international advanced levels and capable of replacing some imported products. However, their production capacity and manufacturing processes are still inferior to those of foreign giants, so they can only mitigate price increases in the short term.
3. Exploration of New Storage Technologies:
- New storage technologies like MRAM and SCM are being developed, which could reduce reliance on DRAM in the long run. However, it will take several years before they become commercially available and provide significant help in the short term.
Conclusion
The explosion in AI computing power has transformed the storage industry from a cyclical commodity to a strategic asset for AI applications. The consumer electronics industry is now facing a “new normal of high prices.” In the past, companies relied on lowering prices and adding more features to attract customers; now, they must raise prices and reduce specifications. Although domestic storage companies and AI PCs offer some hope, the cost of buying smartphones and PCs may continue to rise for consumers in the short term.
(The entire analysis is written in plain language, making it easy for non-financial professionals to understand.)