AI-driven surge in storage demand tightens enterprise SSD supply through 2030

The explosion in AI workloads is causing a global shortage of enterprise storage drives, pushing manufacturers to innovate with higher-density SSDs and advanced cooling solutions amid forecasts of sustained supply constraints until 2030.

The same surge in demand for artificial intelligence that has raised alarms about shortages of high-bandwidth memory is now straining supplies of enterprise storage drives, industry executives say, creating a fresh bottleneck in the data-centre supply chain as operators race to keep pace with larger models and heavier data pipelines. According to reporting by Reuters, suppliers and system-makers are warning that capacity and delivery times for solid-state drives (SSDs) could be tight for years as AI workloads bite into the market for NAND flash and other storage components.

Speaking at Nvidia’s developer conference, the company’s chief executive voiced alarm at the pace of change in data movement from storage to processors, saying “The storage system is going to get pounded.” Reuters coverage of the event noted that software and hardware developments unveiled by Nvidia aim to accelerate transfers from persistent storage into memory, intensifying the upstream demand for higher-capacity drives.

Executives from Solidigm, the U.S.-based storage unit spun out of SK hynix, told Reuters that next‑generation AI systems coming later this year are expected to demand roughly 35% more storage than prior deployments. Greg Matson, Solidigm’s senior vice-president, warned that while the company plans denser drives and capacity expansion, “Can we keep up? No, we can’t. I could sell twice as much as I am today.” He said suppliers expect supply to be constrained through to 2030 in line with the longer-term tightness SK Group’s leadership has forecast for memory markets.

Manufacturers are responding by introducing higher-density SSDs and shifting product mixes to favour larger QLC devices optimised for capacity rather than peak memory bandwidth. Solidigm highlights enterprise QLC SSDs with capacities up to 61.44TB as part of a strategy to accommodate AI datasets that grow both in size and in the frequency with which they must be accessed. Industry reports also describe a broader shift from the high-bandwidth memory used adjacent to GPUs toward high-capacity flash storage as a complementary layer in AI infrastructure.

Thermal design and rack‑level engineering are evolving to match the performance and density of these new drives. Solidigm and others are developing SSDs with cold‑plate interfaces and designs intended for liquid-cooled racks, reflecting a move by data centres from air cooling toward liquid or hybrid solutions to manage the heat generated by dense storage and compute stacks. Vendors say such innovations are intended to enable higher sustained throughput and reliability in modern AI deployments.

Market watchers report that the imbalance between demand and supply is already inflating enterprise SSD prices and forcing customers to rethink storage architecture. Analysis of the flash market shows that, unlike past cyclical shortages, this wave is driven by fundamentally higher baseline demand from AI training and inference workloads, leading some organisations to explore software-level strategies to squeeze more value from existing capacity while providers ramp production.

Looking ahead, suppliers say the near-term trajectory is clear: greater production of high-capacity flash, continued product innovation around density and cooling, and persistent tightness in availability through much of this decade unless capacity additions accelerate beyond current plans. According to reporting from manufacturers and market analysts, that combination will shape procurement strategies and infrastructure design as enterprises build systems to feed ever larger AI models.

Source Reference Map

Inspired by headline at: ^[1]

Sources by paragraph:

Source: Fuse Wire Services