A global shortage of memory chips fueled by the explosive growth of artificial intelligence is driving up the cost of electronics worldwide, according to an analysis by Bloomberg. The demand for chips used in AI systems has surged so rapidly that prices for certain memory components have skyrocketed, with some dynamic random-access memory (DRAM) spot prices rising by nearly 700 percent over the past year.
The rush to secure supplies is being led by major technology companies that are building vast AI data centers. Firms such as Apple Inc., Alphabet Inc., and Tesla Inc. are among those warning that the shortage is beginning to affect profitability and could slow the pace of AI development. According to Bloomberg, companies are increasingly signing long-term contracts and paying premium prices to lock in future supplies of memory chips essential for AI computing.
Memory chips play a crucial role in modern digital devices. While processors perform calculations, memory chips store data and feed it to the processor, allowing systems to run applications, process information, and deliver responses from digital assistants or AI services. Two key forms dominate the market: NAND flash memory, which stores files and data permanently in devices such as smartphones and laptops, and DRAM, which temporarily holds information being actively processed by a computer.
The AI boom has introduced a more advanced type of memory known as high-bandwidth memory, or HBM, which dramatically increases the speed at which data can move between memory and processors. These chips are essential for training and operating advanced AI models that process massive amounts of data. Because they are difficult to manufacture and in extremely high demand, HBM chips have become one of the most sought-after components in the global semiconductor supply chain.
According to Bloomberg Intelligence, demand from data centers alone accounted for roughly 50 percent of global DRAM consumption in 2025, up sharply from 32 percent five years earlier. That share is expected to rise even further as technology companies continue expanding their computing infrastructure. Major cloud and technology firms including Amazon.com Inc., Microsoft Corp., and Meta Platforms Inc. are investing heavily in AI infrastructure, helping push global spending on AI systems toward an estimated $650 billion in 2026.
As AI developers secure priority access to memory chips, manufacturers of consumer electronics are finding it increasingly difficult to obtain supplies. Companies that produce laptops, smartphones, gaming consoles and other devices are now competing for a shrinking pool of memory components. This shift has already pushed up manufacturing costs across the electronics industry.
For example, computer maker HP Inc. said memory now accounts for roughly 35 percent of the materials cost of a laptop, compared with about 15 to 18 percent only a quarter earlier. The company has begun raising prices and adjusting product configurations, including offering models with lower memory capacity. Similarly, Dell Technologies has increased prices for both servers and personal computers as component costs rise.
Smartphones could also become significantly more expensive. Research firm Counterpoint estimates that higher memory prices may increase the cost of materials for smartphones by 15 percent or more in the coming quarters. In response, some manufacturers are reconsidering low-margin entry-level devices or reducing memory specifications in certain models. The research firm International Data Corporation forecasts that the global smartphone market could shrink by 12.9 percent in 2026, potentially marking the steepest decline on record.
The supply shortage is exacerbated by the structure of the global memory industry, which is dominated by only three major manufacturers: Samsung Electronics, SK Hynix, and Micron Technology. Building new chip fabrication plants requires billions of dollars and several years before they can produce at scale, meaning supply cannot quickly expand to match surging demand.
Producing advanced high-bandwidth memory is even more complex. The chips are created by stacking multiple ultra-thin layers of silicon with microscopic connections that allow extremely fast data transfer. Because a single defect can ruin an entire stack, manufacturing yields are lower than for conventional chips, limiting how quickly production can grow.
Despite the intense demand, chipmakers remain cautious about expanding too aggressively after experiencing severe boom-and-bust cycles in the past. According to Bloomberg, companies including Micron and SK Hynix suffered billions of dollars in losses as recently as 2023 after a prolonged oversupply of memory chips following the pandemic surge in electronics demand.
As AI development accelerates, however, the industry faces a critical question: whether the current shortage represents a temporary imbalance or a long-term structural shift in computing. For now, the companies building AI infrastructure are securing the chips they need, while consumer electronics manufacturers brace for higher costs, tighter margins and potentially more expensive products for customers around the world.
