GPU Supply Crunch Deepens As AI Data Centers Strain Memory

GPU makers are shifting resources toward AI data centers, deepening a memory shortage that is pushing consumer technology prices higher and leaving buyers with fewer options. A JP Morgan report says DRAM prices are expected to rise 400 percent from 2024 levels by the end of 2026, while Nvidia's GeForce RTX 5090 now sells far above its launch suggested price. Some industry leaders warn that tight supply could last into the 2030s.

Memory, not chips, is the bottleneck

At its core, the GPU shortage is a memory problem. The industry was already moving from DDR4 to DDR5, slowing supplies during the transition, while rapid AI data center growth has increased demand for high-bandwidth memory, a specialized form of DRAM. HBM demand far outpaces current supply, and it competes for the same wafer capacity as other memory products. Suppliers have steadily reoriented production toward advanced HBM chips, which sell for roughly 30 percent more than traditional DRAM, and are now moving to HBM4 and HBM4E, further limiting supply. A Samsung executive told Reuters in September that HBM could account for nearly 30 percent of total DRAM production in 2027, up from 20 percent the year before, with some expecting nearly 40 percent by 2029. AI data centers are projected to consume roughly 70 percent of global DRAM supplies this year.

Consumers feel the squeeze

The shortage has raised prices across consumer electronics. The JP Morgan report says the Consumer Price Index for software and accessories has increased by nearly a quarter over the last two years. A 2026 report by The Economist found similar rises in video game consoles and laptops, prompting an uncharacteristic fall in PC shipments. In Japan, some gaming GPUs have risen roughly 41 percent above their American counterparts. Resellers have also raised prices: GameStop markets Sony's PlayStation 5 Pro for roughly $500 above MSRP. Tom's Hardware's chips tracker found the GeForce RTX 5090 currently retails $4,920 above its $1,999 suggested price from its January 2025 launch.

Executives warn of years of tight supply

Micron CEO Sanjay Mehrotra said in a recent earnings call to expect supply to be much tighter in calendar 2027 and 2028 than in 2026. SK Hynix CEO Kwak Noh-jung told Reuters in July that next year could be the worst year in the industry's history from the supply perspective. SK Hynix chairman Chey Tae-won said at a Nvidia event that it will take at least four to five years to build adequate wafer supply, with production lagging 20 percent behind demand until at least 2030. Micron has reportedly sold 75 percent of its 2027 HBM output at prices much higher than 2026 prices, while Samsung has sold almost 80 percent of its 2027 memory output, with roughly 70 percent tied up in five-year contracts. Many chip producers have already sold much of their 2027 supplies and are focusing on 2028 and beyond.

Skeptics question how long the boom lasts

Not everyone expects the shortage to continue. Acer CEO Jason Chen told DigiTimes prices could plateau next year, citing a potential influx of cheap Chinese alternatives. A South Korean publication caused memory stocks to tumble in October 2026 after quoting experts who predicted a flood of new supply in 2029 would turn the memory market into a red ocean. Nvidia head Jensen Huang dismissed the idea of a supply surplus, telling CNBC that memory is something you can always use more of. Three companies, Micron, Samsung Electronics, and SK Hynix, account for roughly 90 percent of global DRAM production, and all three expect long-term shortages despite major capacity expansions. Micron ceased direct-to-consumer sales last year to dedicate resources to AI production.

Sources: Engadget