A Strange Phenomenon Created by the AI Boom: DDR5 Memory Becomes More Expensive than the PS5
Editor
A few days ago, I was truly surprised by a report from Tom’s Hardware. It was about the G.Skill Trident Z5 Neo RGB DDR5-6000 64GB kit soaring to $600 on Newegg. Can you grasp how expensive that is? Sony’s PlayStation 5 Digital Edition is priced at $399, and Microsoft’s Xbox Series X Digital Edition is $569, yet a single memory kit costs more than these consoles.
Even more shocking is the rate of this price increase. As of September 20, this product was around $220, but in just two months, it skyrocketed by approximately 190%. As someone who enjoys building PCs, such price fluctuations were completely unexpected. While it’s common in the industry for memory prices to fluctuate, I’ve never seen it to this extent.
The main cause of this phenomenon is identified as the global ‘AI boom.’ As the competition in AI development intensifies, the demand for memory from big tech companies has surged. Semiconductor manufacturers are focusing their production lines on high-margin AI high-bandwidth memory (HBM) or server memory instead of consumer DRAM. Consequently, the supply shortage of general PC memory has worsened.
To understand this situation, one must be aware of the current structure of the memory market. The global DRAM market is dominated by three companies—Samsung Electronics, SK Hynix, and Micron—holding over 90% of the market share. As they shift their production lines to AI HBM, the supply of consumer DDR5 memory has decreased. Notably, Samsung Electronics announced that by the third quarter of 2025, HBM sales are expected to account for about 40% of their total memory sales.
The Impact of the AI Market on the Memory Industry
The current growth of the AI market makes it clearer why this phenomenon is occurring. The global AI chip market is projected to grow from approximately $71 billion in 2024 to $347 billion by 2030, with an annual growth rate of over 30%. Memory is a key component of AI systems. For instance, NVIDIA’s H100 GPU is equipped with 80GB of HBM3 memory, which alone costs around $30,000.
For memory manufacturers, this was an obvious choice. The price per GB of HBM is more than ten times that of standard DDR5. For example, SK Hynix announced that the average selling price of their HBM3E products is 8-10 times higher than that of regular DRAM. By the third quarter of 2025, SK Hynix’s HBM sales recorded approximately 7 trillion won, a 330% increase from the same period the previous year.
Samsung Electronics is no exception. The company announced plans to expand its HBM production capacity by more than three times in 2025 compared to 2024. They are expanding HBM-dedicated production lines in Pyeongtaek and Hwaseong, aiming for mass production of HBM4 by the end of 2025. Such investments negatively impact the production of consumer memory.
Micron is also not an exception. The company forecasts that its HBM sales for the fiscal year 2025 (September 2024 – August 2025) will increase by more than 400% compared to the previous year. They are expanding HBM production in Idaho and Singapore and establishing HBM-dedicated lines in Japan.
Supply Shortages Spreading Across the PC Component Market
The issue is not limited to memory. According to reports, the wait time for large-capacity hard disks (HDDs) has extended up to two years, indicating a widespread shortage across the PC component market. This is directly related to the boom in AI data center construction.
In fact, storage companies like Seagate and Western Digital have announced a surge in orders for large-capacity enterprise HDDs and SSDs. For Seagate, sales of HDDs with capacities of 20TB or more increased by 45% in the second quarter of 2025 compared to the same period the previous year. As production lines focus on such enterprise products, the supply of consumer products has decreased.
The graphics card market is experiencing a similar trend. NVIDIA is adjusting its gaming GPU production due to the surge in AI chip demand. This is one reason for the unstable supply of high-end models like the RTX 4090. AMD is also focusing on the production of the MI300 series for data centers, affecting the supply of the Radeon series.
Industry experts predict that this supply shortage will not be resolved in the short term. As companies continue to invest in developing artificial general intelligence (AGI), constraints on the supply of DRAM and NAND flash memory are expected to persist until 2026. Personally, I find this forecast quite realistic, as the AI investment frenzy seems likely to continue for the foreseeable future.
Particularly noteworthy are the capital expenditure (CAPEX) plans of memory companies. SK Hynix plans to invest about 18 trillion won, and Samsung Electronics about 25 trillion won in the semiconductor sector in 2025, with a significant portion focused on HBM and server memory production lines. Expanding consumer memory production capacity is inevitably a lower priority.
In this situation, what should consumers planning to build PCs do? Personally, I believe there is a high possibility that memory prices will rise further, especially for large-capacity kits like 64GB or 128GB. Currently, a 32GB kit seems to be the optimal choice in terms of cost-effectiveness.
Another point to consider is the generational performance differences of DDR5 memory. The high-performance DDR5-6000 and above memory currently available on the market are primarily designed for gaming and content creation. For general use, DDR5-4800 or DDR5-5200 should suffice. Considering price versus performance, there’s no need to insist on the highest specifications.
However, despite all this analysis, the market situation is subject to many unpredictable variables. Developments in China’s AI chip industry, the strengthening of U.S. semiconductor export regulations to China, and the emergence of new memory technologies can all affect prices. As of November 2025, the AI boom seems to be in its early stages, and this supply shortage is likely to continue for several years.
Ultimately, from a consumer perspective, it is important to closely monitor component price fluctuations and make purchasing decisions at the appropriate time as needed. Above all, this strange phenomenon of a single memory kit being more expensive than a gaming console seems to be a snapshot of the changes brought about by the AI era, which is both intriguing and concerning.
This article was written after reading the Untitled article, with personal opinions and analysis added.
Disclaimer: This blog is not a news outlet, and the content written reflects the author’s personal views. Responsibility for investment decisions lies with the investor, and no liability is accepted for investment losses based on the content of this article.
