Why RAM Prices Are Rising: The AI Boom Is Changing the Memory Industry
Artificial Intelligence is transforming far more than software. While companies like OpenAI, Nvidia, and Microsoft dominate headlines, one of the biggest beneficiaries of the AI revolution is an industry most consumers rarely think about: memory chips.
Krish Gupta
7/1/20263 min read
Artificial Intelligence is transforming far more than software. While companies like OpenAI, Nvidia, and Microsoft dominate headlines, one of the biggest beneficiaries of the AI revolution is an industry most consumers rarely think about: memory chips.
Over the past year, prices of DRAM and NAND memory have risen significantly after years of decline. Consumers upgrading laptops, building gaming PCs, or purchasing enterprise servers are increasingly paying more for memory, while manufacturers of AI hardware continue placing record orders for high-performance chips. The reason lies in a major shift in how semiconductor companies are allocating their production capacity.
Memory manufacturers such as Samsung Electronics, SK hynix, and Micron Technology produce different categories of DRAM. Traditional DRAM is widely used in laptops, desktops, smartphones, and consumer electronics. High Bandwidth Memory (HBM), however, is a far more advanced product designed specifically for AI accelerators and high-performance computing.
HBM has become one of the most valuable components inside modern AI systems. Nvidia's H100 and Blackwell GPUs, AMD's MI300 series, and several AI accelerators rely heavily on HBM to process massive AI workloads. Unlike conventional memory, HBM provides significantly higher bandwidth while consuming less power, making it essential for training large language models and running advanced AI applications.
The economics are simple.
HBM sells at significantly higher prices than standard DRAM and generates substantially better profit margins for manufacturers. At the same time, demand has exploded as cloud providers such as Microsoft, Amazon, Google, Meta, and Oracle continue investing billions of dollars in AI infrastructure.
Faced with finite manufacturing capacity, memory companies have increasingly prioritized HBM production over conventional DRAM.
This shift has reduced the supply available for laptops, smartphones, and consumer electronics. Even if consumer demand remains relatively stable, lower supply naturally places upward pressure on prices. It is a classic example of supply allocation rather than purely demand expansion.
The memory industry has always been cyclical.
Unlike software companies, semiconductor manufacturers cannot instantly increase production. Building fabrication facilities requires billions of dollars in capital expenditure and several years of construction. As a result, supply often reacts slowly while demand can change rapidly.
When demand weakens, prices collapse. When demand exceeds available production, prices rise sharply.
The AI boom has accelerated this cycle. Industry analysts expect HBM demand to grow by well over 50% annually for the next several years, encouraging manufacturers to dedicate increasing portions of their production lines to premium AI memory.
For consumers, the consequences are becoming visible.
Laptop manufacturers face higher component costs. Smartphone makers must manage tighter memory supply. PC builders are paying more for DDR5 modules than expected. Enterprise customers deploying conventional servers are also experiencing pricing pressure despite not directly participating in the AI race.
Interestingly, the companies benefiting most from AI may not necessarily be those building AI models.
Samsung, SK hynix, and Micron effectively supply one of the most critical inputs for AI infrastructure. As demand for GPUs rises, demand for HBM rises alongside it. Every major AI data centre requires enormous quantities of advanced memory, positioning memory manufacturers at the center of the AI value chain.
The industry has experienced controversy before.
During the early 2000s, several major DRAM manufacturers faced investigations and paid substantial fines after authorities in the United States and Europe found evidence of price-fixing. Those cases highlighted how concentrated the global memory market had become. Today, however, regulators have not alleged similar collusion regarding the current price increases. Instead, today's market dynamics are largely driven by commercial decisions regarding production allocation and capacity constraints.
This concentration nevertheless gives manufacturers considerable influence.
The global DRAM market is dominated by just three companies—Samsung, SK hynix, and Micron—which together account for more than 90% of worldwide production. Such a highly concentrated industry means that strategic production decisions by a handful of companies can significantly influence global pricing.
Looking ahead, the market will largely depend on one question: how long will AI demand continue outpacing new manufacturing capacity?
If investment in AI infrastructure continues accelerating, manufacturers are likely to keep prioritizing HBM because it delivers superior returns. Traditional DRAM supply could therefore remain constrained, keeping prices elevated. Conversely, if new fabrication capacity comes online faster than expected or AI demand moderates, pricing pressure may gradually ease.
The broader lesson extends beyond memory chips.
The AI revolution is not only creating winners among software companies. It is reshaping entire semiconductor supply chains, changing investment priorities, and altering how manufacturers allocate scarce resources. Sometimes, the biggest beneficiaries of a technological revolution are not the companies building the final product, but those supplying the critical components that make it possible.
For the memory industry, AI has transformed an ordinary semiconductor into one of the world's most strategically valuable technologies. And for consumers, it explains why upgrading a laptop's RAM has suddenly become more expensive than it used to be.