Behind the chip war in the AI era, there is a little-known battle for memory. From high-end GPUs to consumer devices, memory has become the most scarce and critical resource—this is not an alarmist statement.

The data is in front of us: memory costs account for 80% of the total GPU cost. As AI giants like OpenAI and Google compete to stockpile HBM and DRAM, the global chip supply chain is beginning to feel the pressure. Memory manufacturing giants such as Samsung, Micron, and SK Hynix are experiencing unprecedented capacity shortages, with orders piling up.

The ripple effects on the consumer side have already appeared. The cost of next-generation iPhones and other high-end electronic products is expected to increase by 40%—meaning the devices in your pocket could become more expensive. This is not just a pricing issue but also a restructuring of the entire industry chain's costs.

In response to this battle, industry giants have their own strategies. OpenAI has secured long-term memory capacity contracts to lock in supply early. Nvidia is exploring alternative technologies like SRAM to reduce dependence on traditional memory. Meanwhile, the three major memory giants are balancing capacity expansion and pricing strategies.

This memory race has just begun, and the ultimate winner will depend on who can seize the opportunities brought by this supply tightness more quickly.