RAM vs. ROM: Understanding the Evolution and Market Crisis of Computer Memory

The distinction between random access memory (RAM) and read-only memory (ROM) was once a straightforward technical concept, but modern computing has rendered the terminology far more complex. Fundamentally, RAM serves as the workspace for a device’s processor, holding active data that vanishes when power is lost, while ROM traditionally housed permanent startup instructions. However, true read-only memory has largely disappeared from consumer electronics decades ago.

In contemporary devices, almost all storage labeled as ROM is actually rewritable flash memory. For instance, smartphones like the Redmi 13 list “ROM” on their specification sheets, but this refers to UFS or eMMC flash storage with system partitions locked to read-only operation rather than literal read-only chips. Similarly, the BIOS firmware chip in personal computers is technically a small flash unit updated via software patches, not a static memory component.

RAM remains critical for performance, acting as the bridge between slow storage and the processor. When a user opens a browser tab, the data is copied into RAM, allowing nanosecond-speed access. Dynamic RAM (DRAM), the most common type, stores bits as electrical charges in capacitors that require constant refreshing. If power fails, this volatile memory is immediately cleared. Enthusiasts utilizing DDR5 modules or graphics cards equipped with GDDR chips understand that higher capacity allows for more simultaneous applications and demanding games without relying on slower swap files.

Currently, the memory market is experiencing a severe supply crunch driven primarily by the artificial intelligence boom. Data centers are consuming the vast majority of DRAM supply as companies race to build AI infrastructure. According to TrendForce, conventional DRAM contract prices surged between 90 and 95 percent quarter-over-quarter earlier this year. Major corporations, such as NVIDIA, are investing billions to construct massive AI data centers, further straining resources available for general consumer electronics.

This industrial shift has had tangible consequences for the consumer market. Micron terminated its Crucial consumer brand in February to redirect manufacturing toward high-margin AI data center demand. Historical parallels exist; in 2017, Huawei faced scrutiny when it was revealed that memory shortages led to inconsistent storage speeds in its P10 devices. The current crisis mirrors these past events but on a larger scale due to the unprecedented energy and compute requirements of AI workloads.

The history of ROM highlights how technology adapted to overcome physical limitations. Early ROM chips had data physically etched into the circuitry, making changes impossible without factory intervention. In 1956, Wen Tsing Chow introduced Programmable ROM (PROM), which allowed data to be written by blowing microscopic fuses, though this process was permanent. The Erasable PROM (EPROM) followed in 1971, invented by Intel engineer Dov Frohman, utilizing UV light through a quartz window to reset the chip. Later, electrically erasable EEPROMs eliminated the need for UV light, paving the way for Fujio Masuoka’s flash memory. NOR flash appeared in 1984, followed by NAND flash in 1987, which now powers nearly every SSD, smartphone, and USB drive.

While genuine factory-set ROM persists in niche applications like cheap microcontrollers running unchanging code, the term has become a shorthand for permanent storage in consumer marketing. Original game cartridges from the Atari era represent one of the few remaining examples where users interacted with true ROM chips, which is why digital copies used in emulators retain the name. Today, RAM continues to function exactly as it did sixty years ago, while ROM has evolved into a versatile, rewritable medium that defies its original name.