Lenovo says the memory market may not be heading back to the cheaper levels buyers remember from early 2025. Instead, the company believes higher DRAM and NAND flash prices could become the new normal through 2030, and possibly beyond.
The warning matters because memory sits at the core of everyday devices and digital services. If prices remain elevated, the pressure can spread across PCs, laptops, smartphones, SSDs, and cloud platforms used by both consumers and businesses.
AI demand is reshaping the market
The biggest force behind the shift is the rapid rise in AI infrastructure demand. These systems consume large amounts of memory, especially high-performance types that are now heavily contested across the industry.
Lenovo said the jump in component prices became especially sharp from late Q3 into early Q4 2025. The company also argued that supply growth is lagging far behind demand, making a return to earlier price levels look unlikely.
That view is not isolated. Micron has already acknowledged that even its most important customers still cannot get all the supply they need. Samsung and SK Hynix have also raised similar concerns about the imbalance between demand and available output.
New capacity will take time
Manufacturers are expanding production, but new wafer fabs take years before they add meaningful supply. That delay means the market will not cool quickly, even if investment is accelerating now.
SK Hynix has given a sense of how large the push has become. The company, a major player in high-bandwidth memory for AI, wants to nearly double its DRAM wafer capacity by 2030–2031.
It also aims to triple overall production around 2034, moving faster than its original schedule. Even so, many observers still question whether that scale of expansion will be enough to keep up with AI chip demand.
What higher prices mean for users
For consumers, the impact is likely to show up in the cost of finished products. Devices that rely on DRAM and NAND in large volumes can become more expensive to build, which can push retail prices higher or delay cheaper upgrade cycles.
Businesses may also feel the effect through IT spending. Higher memory costs can raise expenses for data storage, infrastructure, and computing services, especially for organizations that depend on large-scale systems.
Cloud providers are not insulated either, since modern data centers rely heavily on memory and storage components. If supply remains tight, pricing pressure can move through the stack and affect services far beyond the chip market itself.
| Company | Key Signal | Timeframe Mentioned |
|---|---|---|
| Lenovo | Expects higher DRAM and NAND prices to stay the new normal | Through 2030 and possibly longer |
| Micron | Even key customers cannot get all the supply they need | Current market condition |
| SK Hynix | Plans major DRAM capacity expansion and higher overall output | 2030–2031, around 2034 |
| Samsung | Has voiced concern over demand and supply balance | Current market condition |
A break from the old boom-and-bust cycle
Lenovo’s outlook points to a broader change in the memory industry. For years, the market was known for a boom-and-bust pattern, with spikes followed by steep declines once supply improved.
This time, the company believes the floor itself is moving higher. Rather than returning to the low-price environment seen before, memory may settle into a permanently elevated baseline.
That shift reflects how central AI has become to semiconductor investment and usage. As long as AI continues to absorb so much capacity, pressure on core components such as DRAM and NAND is likely to remain difficult to relieve.
Some relief could still come from efficiency gains and new technology over time. For the next few years, though, the message from Lenovo and major memory makers is clear: cheap memory may no longer be the market’s default assumption.
