Research · Supply Chain
DRAM prices have moved more in the past two quarters than in the previous decade combined. It's not a niche component story anymore. It's showing up in laptop prices, phone specs, and the entire on-device AI pitch tech companies have been selling you.
If you've priced out a new laptop recently and felt like the numbers looked a little higher than you remembered, you weren't imagining it. Somewhere behind that price tag is a memory chip market that's gone through the kind of price movement usually reserved for oil shocks or currency crises, and almost nobody outside the semiconductor industry noticed until it started showing up at checkout.
How bad are the numbers, actually?
Bad enough that industry trackers are running out of ways to describe it without sounding like they're exaggerating. According to TrendForce, global DRAM contract prices rose an estimated 90 to 95% quarter-over-quarter in the first quarter of 2026 alone, followed by another 58 to 63% jump in the second quarter. Year-over-year, DRAM prices are up roughly 171%, according to research from the Bloomsbury Intelligence and Security Institute, a rate of increase that's reportedly outpaced gold. DDR5 spot prices have quadrupled since September 2025.
Put plainly: the component that quietly determines how much RAM your next laptop or phone can afford to include has, in the space of about a year, become one of the most volatile materials in the entire technology supply chain.
Why this is happening: AI data centers are eating the supply
The cause isn't a factory fire or a natural disaster, the usual suspects behind past chip shortages. It's demand, specifically the enormous and rapidly growing appetite of AI data centers for high-bandwidth memory, the specialized chips that feed data to AI accelerators at the speed modern training and inference workloads require. Memory manufacturers, facing a choice between higher-margin AI memory and standard commodity DRAM, have been reallocating wafer capacity toward the former, at the direct expense of the latter.
Photo by Kevin Ache on Unsplash
The scale of that reallocation is genuinely striking. Hyperscaler capital expenditure is approaching roughly $600 billion in 2026, up about 36% year-over-year, and a meaningful share of that spending flows directly into memory procurement. The competitive shakeup inside the memory industry tells its own story here: SK Hynix has reportedly overtaken Samsung Electronics in DRAM revenue for the first time since 1992, largely on the strength of its position in AI memory, while Samsung is racing to catch up with a planned 50% expansion of its own HBM capacity this year.
What this actually does to a laptop or phone on the shelf
This is where the story stops being an abstract chip-industry issue and starts hitting anyone shopping for a device. Major PC makers, including Lenovo, Dell, HP, Acer, and Asus, have warned of price increases in the range of 15 to 20% for 2026, with some industry estimates running as high as 30% in worse-case scenarios, though more conservative forecasts from IDC put the average increase closer to 4 to 8%. That range itself tells you something: even the analysts tracking this closely can't agree on exactly how bad it gets, only that "bad" is the shared direction.
Photo by Elijah Pilchard on Unsplash
Memory isn't a minor line item in a device's cost either. According to IDC, memory can represent 15 to 20% of a mid-range smartphone's total bill of materials, and 10 to 15% for a high-end flagship. When the price of that single component doubles or triples, device makers are left with three unappealing options: raise the retail price, quietly cut the amount of RAM or storage included, or absorb the hit to their own margin. Several major consumer electronics brands have already chosen the first option in the open. Nintendo cited memory costs directly when it raised the Switch 2's price to $499.99. Microsoft said storage and memory costs rose more than 2.5 times ahead of its August 2026 Xbox price increase. Sony raised PS5 prices in April for the same underlying reason.
The display market is the canary in the coal mine
One of the clearest signals of how far this has spread comes from an unexpected corner: flat panel displays. Counterpoint Research now forecasts global flat panel display shipments will decline 4.7% year-over-year in 2026, as memory-driven price increases flow through to finished devices and suppress demand, particularly at the low and mid-range tiers where price sensitivity is highest. Mobile phone display shipments are projected to fall 9.5% year-over-year. Notebook panel shipments are down 2.7%, and monitor panels down 1.7%.
Photo by Eyestetix Studio on Unsplash
There's a telling countertrend buried in that same data, too. While budget and mid-range devices are getting squeezed, manufacturers are leaning harder into premium OLED offerings to protect margins elsewhere in the lineup: OLED monitor panel shipments are forecast to grow 60% year-over-year, and OLED notebook panels 50%. In plain terms, the mid-market is shrinking while the premium tier gets more investment, a pattern that tends to widen the gap between what a budget-conscious buyer can access and what a premium buyer gets, right at the moment affordability matters most.
Where this really bites: on-device AI growth is stalling
Here's the finding that ties this whole story back to where this publication usually spends its time. ABI Research's own mid-year consumer technology outlook names the memory shortage directly as a drag on the on-device AI category specifically, and it's a genuinely awkward moment for it to happen. The entire "AI PC" and "AI phone" pitch of the last two years has rested on convincing consumers to pay more for a device with more capable, memory-hungry on-device AI features. ABI Research has now lowered its long-term on-device AI chipset shipment outlook as a direct result of these pricing pressures.
It's a genuinely awkward irony worth sitting with: the same AI boom driving hyperscaler memory demand through the roof is simultaneously making it harder and more expensive to put capable AI hardware directly into consumers' hands. Chipset innovation hasn't slowed down, but the economics around it have gotten considerably less forgiving.
This story, at a glance
What this means if you're building a hardware startup right now
If your product roadmap depends on shipping a physical device with meaningful onboard memory, this isn't background noise you can wait out quietly.
Photo by Vitaly Gariev on Unsplash
- Rebuild your bill-of-materials assumptions now, not at your next funding round. A cost structure modeled on 2025 memory pricing is likely already out of date, and the gap between your projections and reality will only widen from here.
- Consider software-side memory efficiency as a genuine competitive lever, not a nice-to-have. If a smaller memory footprint means a meaningfully cheaper bill of materials in this environment, efficient engineering becomes a real cost advantage, not just an engineering preference.
- Diversify component sourcing where you realistically can. Analysts consistently point to supply diversification as one of the few levers device makers actually control in a shortage driven by structural demand rather than a single point of failure.
- If your pitch depends on cutting-edge on-device AI as the differentiator, price-test that assumption honestly. The broader category's growth outlook has already been revised downward once this year because of exactly this pressure.
How long does this actually last?
This is the question every device maker and hardware startup actually wants answered, and the honest response is: longer than past shortages, and nobody's fully certain how much longer. The last comparable DRAM shortage, in 2017 and 2018, roughly doubled memory prices and eased within about eighteen months once supply caught up. This cycle looks structurally different, because the underlying driver, AI infrastructure demand, isn't a temporary spike, it's a sustained, multi-year capital investment cycle. Intel has pointed to 2028 before conditions normalize. Silicon Motion has warned the shortage across DRAM, NAND, and HBM could persist into 2028. Gartner expects the broader storage crunch to run into 2027. Micron has committed $200 billion to new US production capacity, but meaningful output from that investment isn't expected until mid-2027 at the earliest.
New fabrication capacity simply takes years to build, which means the honest planning assumption for now is that relief, if it comes, is a 2027-to-2028 story, not a next-quarter one.
Frequently asked questions
Should I buy a laptop now or wait for prices to come down?
Based on current forecasts, waiting is unlikely to help in the near term. Multiple industry sources point to a price peak in the second half of 2026 with limited downside before mid-2027 at the earliest, so if you need a device for genuine work needs, delaying may not save money.
Is this shortage only affecting premium or high-end devices?
It's actually hitting budget and mid-range devices hardest, since memory represents a larger share of a lower-cost device's total bill of materials. Premium devices have more room to absorb the cost increase without becoming unaffordable.
Will this permanently change how "AI PC" and "AI phone" products are marketed?
It's too early to say permanently, but the near-term effect is clear: ABI Research has already revised its on-device AI shipment outlook downward specifically because of memory pricing pressure, which suggests the category's growth story is being genuinely tested, not just delayed.
The bottom line
The memory shortage driving up laptop and phone prices right now isn't a temporary supply hiccup waiting to resolve itself. It's the direct, structural consequence of AI infrastructure spending outbidding consumer electronics for the same manufacturing capacity, and every forecast on the table says it gets worse before it gets better. For anyone building hardware, buying hardware, or betting a product roadmap on affordable on-device AI, the planning horizon that matters now isn't next quarter. It's 2027, possibly 2028, and the decisions worth making today are the ones that hold up under that timeline.
Further reading
See our deep dive on Ramp's $44 billion valuation, and our Research coverage on how AI efficiency gains are reshaping global access to capable models.
Figures referenced in this article are drawn from public research and reporting, including Counterpoint Research, ABI Research, TrendForce, IDC, and the Bloomsbury Intelligence and Security Institute, as of the publish date, and are subject to revision as market conditions evolve. This article does not contain affiliate links; where future articles do, they will be disclosed per our Affiliate Disclosure.