The race for artificial intelligence is significantly increasing global demand for RAM.
Illustrated by the economic and diplomatic rivalry between the United States and China, RAM is becoming a real strategic issue. This exponential rise in prices is also impacting Bitcoin, for which memory and storage remain essential for nodes and mining infrastructure.
AI is creating unprecedented pressure on memory supply chains
Since September 2025, prices for RAM modules (printed circuit boards on which integrated memory circuits are mounted) have jumped from 163% to 619%. This radical change is part of a geopolitical context marked by a frantic race for artificial intelligence (AI).
Since 2024, the rise of AI has profoundly changed memory usage. This is the case with large language models (LLMs), which require high computing power and very fast data transfer capabilities between GPUs and memory.
As a result, high-bandwidth memory (HBM) is becoming indispensable, as it offers both low latency and very high throughput. Major AI players are thus mobilizing a growing share of global memory production (DRAM and NAND) to equip data centers dedicated to training and inferencing large models.
These infrastructures are extremely memory- and fast-storage-intensive, prompting manufacturers to focus on the most profitable segments, such as server memory and HBM, at the expense of consumer products. This strategic reallocation has created an artificial shortage in certain market segments.
The price increase is not uniform and particularly affects high-capacity variants. This is the case for DDR4 and DDR5 modules, the standard models used in PCs and mining rigs in particular.
The 32 GB DDR5 module, for example, has seen a dramatic increase in all markets, with US prices rising from $149 to $392 (+163%) and Japanese prices seeing the most extreme increase, rising from ¥16,000 to ¥115,090 (+619%).
Limited direct impact, but costly indirect repercussions for Bitcoin mining activities
As you know, Bitcoin mining is based on the SHA-256 algorithm, a hash algorithm used to secure and validate blocks in the network. This algorithm is essentially arithmetic, meaning it performs intensive calculations but requires only very small amounts of RAM to store temporary variables.
In practice, this means that even a significant increase in the price of RAM has virtually no effect on the cost of manufacturing ASICs (specialized circuits for mining), whose cost is mainly: silicon, nanometer etching, packaging, and heat dissipation.
However, the impact of the increase in the price of RAM is felt indirectly. This is because Bitcoin nodes store the entire blockchain on an SSD, which uses NAND flash memory, and use RAM for the mempool or UTXO cache.
If the price of RAM rises sharply, the cost of operating these nodes will increase, which could discourage some individuals or organizations from maintaining full nodes, thereby reducing the number of independent participants and the diversity of the network.
In other words, Bitcoin remains robust in the face of rising RAM prices for block production, but a prolonged increase in memory costs could affect the decentralization and resilience of the network.
It is interesting to note that the CEO of TeamGroup, a RAM module manufacturer, has indicated that memory prices are unlikely to normalize before 2027-2028, pending the arrival of new production capacity on the market.
