Six months ago, DRAM cost $0.43 per gigabit. Dell’s COO says it recently hit $2.39. That’s a 5.5x increase in half a year, and it shows no sign of letting up.
This isn’t a single-product blip. DRAM, LPDDR5x, and NAND flash are all getting hammered with 90–95% quarter-on-quarter price surges in Q1 2026. The structural driver is AI infrastructure — and that’s what makes this fundamentally different from the 2021 COVID-era shortage and much harder to plan around.
We’ve pulled together every major analyst forecast on memory pricing in one place. Bookmark it, send it to your stakeholders. For the broader memory shortage story, we’ve put together a full overview of how AI’s appetite for memory is reshaping the tech industry.
How much have DRAM prices actually increased in 2026?
The headline number: TrendForce’s February 2026 survey revised its Q1 2026 forecast for conventional DRAM contract prices to 90–95% quarter-on-quarter. That’s the largest quarterly increase on record. Their earlier January estimate was 55–60% QoQ — the forecast nearly doubled within weeks.
PC DRAM specifically is tracking over 100% QoQ increases in some contract categories.
And it’s not just conventional DRAM. LPDDR5x contract prices surged roughly 90% QoQ — what TrendForce calls “the steepest increases in their history.” NAND flash prices rose 55–60% QoQ, with enterprise SSD pricing up 53–58% QoQ. That’s three simultaneous price shocks across different memory types, all landing at once.
Gartner’s research director Ranjit Atwal put it plainly: “The speed at which the memory pricing has increased has shocked everybody.” Gartner projects a 130% year-on-year DRAM price rise for 2026.
Here’s where things get genuinely strange. Counterpoint Research reports that DDR4 spot prices have hit $2.10 per gigabit — which now exceeds advanced HBM3e at $1.70 per gigabit. Old memory costs more than the cutting-edge stuff. That’s a legacy price inversion, and it tells you just how distorted this market has become.
The gap between contract and spot pricing matters here too. Enterprise buyers locked into quarterly contracts have some insulation. Those buying on the spot market are wearing the full brunt. If your procurement is on spot, you’re paying more for DDR4 than hyperscalers pay for HBM3e.
For why prices rose so sharply — the production mechanics — we’ve covered that in detail separately.
What does the full memory market revenue picture look like?
TrendForce projects the global memory market at $551.6 billion in 2026, rising to $842.7 billion in 2027. DRAM revenue alone is up 144% year-on-year, hitting $404.3 billion. NAND flash revenue is up 112% YoY at $147.3 billion.
Those numbers tell a specific story — revenue is surging because of price, not because more memory is shipping to consumer markets. The volume going to PCs and smartphones is actually shrinking. It’s the price per unit that’s driving everything.
Samsung regained the number one DRAM revenue position in Q4 2025. SK Hynix leads in HBM production. Micron’s HBM and cloud-related memory revenue grew from 17% of its DRAM revenue in 2023 to nearly 50% in 2025. Micron CEO Sanjay Mehrotra projects the HBM market will grow from $35 billion in 2025 to $100 billion by 2028.
Here’s the thing to keep in mind: what’s profit for Samsung, SK Hynix, and Micron is cost for everyone else. Every record revenue quarter for memory makers is a record bill-of-materials quarter for OEMs.
What does this mean for PC and smartphone prices?
When IDC, Gartner, and Counterpoint Research all converge on the same direction, the signal is pretty clear: the PC and smartphone markets are contracting.
IDC projects the PC market will shrink 11.3% in 2026. Gartner forecasts 10.4%. For smartphones, Counterpoint Research projects shipments down 12% — “the sharpest decline on record.” IDC has smartphones down 12.9%; Gartner at –8.4%.
HP reported in its Q1 earnings call that memory now accounts for 35% of the cost to build a PC, up from 15–18% the previous quarter. That makes memory the single largest cost driver in PC manufacturing. Let that sink in — one component, 35% of the build cost.
At the product level, Lenovo, Dell, HP, Acer, and ASUS have all flagged 15–20% price hikes and are shortening quote validity windows. Some OEMs are shipping lower-SSD configurations just to keep headline prices down — less storage for more money.
The timing couldn’t be worse for enterprise IT. Windows 10 end-of-life in October 2025 triggered refresh cycles that are now colliding head-on with the shortage. And AI PCs require a minimum 16GB RAM — that’s Microsoft’s Copilot+ standard — so devices need more memory at precisely the moment memory is most expensive.
Not everyone is equally exposed, though. Apple and Samsung have long-term supply agreements that secure memory 12–24 months ahead. Budget Android OEMs don’t have that luxury and have no choice but to pass costs straight to consumers.
For a deeper look at how these price increases are hitting the PC and smartphone markets, we’ve covered the device-level impact separately.
How does 2026 compare to the COVID-era chip shortage?
The 2021 shortage was a supply-side disruption. Factories closed, logistics chains broke, and pandemic demand for work-from-home gear spiked. Hyperscalers bought up huge memory inventories, prices surged, then supply normalised.
The 2026 shortage is demand-driven. AI infrastructure demand for HBM is consuming wafer capacity that would otherwise produce conventional DRAM and NAND. That distinction matters: supply disruptions resolve when logistics normalise. Demand-driven shortages persist as long as the demand source keeps growing — and AI demand isn’t slowing down.
The setup traces back to the 2022–2023 memory bust. Samsung cut production by roughly 50%. After that bust, all three major memory companies pulled back on investment. IEEE Spectrum’s Thomas Coughlin put it bluntly: “There was little or no investment in new production capacity in 2024 and through most of 2025.”
Then the AI demand wave hit, and there was simply no spare capacity to absorb it.
By the numbers, 2026 is worse. Q1 2026 is delivering 90–95% QoQ increases — there’s no comparable single-quarter spike in the historical record. IDC characterises the situation as “not just a cyclical shortage driven by a mismatch in supply and demand, but a potentially permanent, strategic reallocation of the world’s silicon wafer capacity.”
McKinsey projects $7 trillion in data centre spending through 2030, with $5.2 trillion AI-focused. That demand concentration is why AI is driving this price surge.
What do analyst forecasts say about the memory market in 2026 and 2027?
Here are the data points from each firm that we haven’t covered above.
IDC: DRAM supply growth in 2026 is projected at just 16% YoY and NAND at 17% YoY — well below historical norms. IDC uses the phrase “structural reset” to describe what’s happening. That’s not the kind of language you want to hear from analysts.
Gartner’s Atwal: “Price is not only increasing in the short term… it’s going to remain high almost through to the end of 2027.”
Counterpoint Research places the earliest inflection point at Q4 2027.
Intel CEO Lip-Bu Tan, speaking at the Cisco AI Summit, was blunter: “There’s no relief until 2028.”
The fab construction timelines explain why. New capacity is coming — Micron Singapore (2027), SK Hynix Cheongju (2027), Samsung Pyeongtaek (2028), SK Hynix West Lafayette (end of 2028), Micron’s Onondaga County megafab (full production 2030). But building a fab and getting it to volume production takes 18 months or more. That structural lag is why new capacity can’t relieve the current shortage quickly.
For the full timeline analysis of when relief arrives, we cover the fab-by-fab breakdown in detail.
Is the shortage going to get worse before it gets better?
Yes. The data points in one direction for the next 18 months.
The only near-term supply lever is incremental process node upgrades within existing fabs. TrendForce notes that “gains in output can be achieved only through incremental process upgrades, making short-term capacity tightness difficult to resolve.” And it gets worse — NAND flash manufacturers are reallocating production lines to DRAM because it’s more profitable, which further squeezes NAND capacity. The two shortages are feeding each other.
Meanwhile, demand keeps climbing. NVIDIA’s B300 GPU uses eight HBM chips, each containing 12 DRAM dies. AMD’s MI350 follows the same architecture. Every new GPU generation ratchets up per-unit memory consumption.
There’s also a supply lock-up problem. North American hyperscalers — Meta, Google, Microsoft, Amazon — are negotiating long-term DRAM agreements with memory suppliers. Counterpoint Research’s Tarun Pathak put it directly: “A lot of these memory companies are asking smartphone vendors to stand in line behind the hyperscalers.”
And here’s the part that matters for your budget planning: memory prices historically fall more slowly than they rise. Economist Mina Kim of Mkecon Insights told IEEE Spectrum: “In general, economists find that prices come down much more slowly and reluctantly than they go up. DRAM today is unlikely to be an exception.”
Where can you track memory prices and analyst forecasts?
Given the outlook, you’re going to want to keep tabs on this. Here are the primary sources we’ve referenced throughout.
TrendForce (trendforce.com) publishes quarterly DRAM and NAND contract price forecasts through its DRAMeXchange platform. Their February 2, 2026 press release is the definitive Q1 2026 pricing source.
IDC (idc.com) publishes quarterly PC and smartphone market trackers. Their December 2025 “Global Memory Shortage Crisis” report is the key public reference document.
Counterpoint Research (counterpointresearch.com) covers smartphone and memory market analysis with public-facing research summaries.
Gartner (gartner.com) publishes enterprise IT spending and semiconductor forecasts. The February 2026 data on shipment declines was reported through Computerworld.
IEEE Spectrum (spectrum.ieee.org) provides technical analysis and expert commentary. Samuel K. Moore’s “How and When the Memory Chip Shortage Will End” is the most comprehensive single source for the COVID-era comparison and fab timeline data.
NPI Financial (npi.ai) offers enterprise IT procurement advisory with pricing guidance for hardware buyers navigating the shortage.
For a complete picture of what the AI memory shortage means for tech companies — the root causes, who is winning and losing, and what to do about it — see the broader memory shortage story.
FAQ Section
Why is RAM so expensive right now in 2026?
AI infrastructure demand for High-Bandwidth Memory (HBM) is eating up wafer capacity at Samsung, SK Hynix, and Micron — capacity that would otherwise produce the conventional DRAM that goes into PCs and smartphones. TrendForce reports DRAM prices surged 90–95% QoQ in Q1 2026 as a result.
Is the memory shortage going to get worse in 2026?
Yes. Counterpoint Research puts the earliest inflection point at Q4 2027. Intel CEO Lip-Bu Tan stated “there’s no relief until 2028.” New fab capacity from Micron and SK Hynix doesn’t reach volume production until 2027 at the earliest.
When will DRAM prices go back down?
Analyst consensus points to gradual relief beginning in late 2027 as new fab capacity comes online. But memory prices historically fall more slowly than they rise, so normalisation will lag behind supply additions. Gartner expects prices to remain high through to the end of 2027.
Why is my new laptop or phone going to cost more this year?
Memory is now the single largest cost component in many devices. HP reports memory represents 35% of PC bill of materials. With DRAM prices up 90–95% and NAND up 55–60%, OEMs including Lenovo, Acer, and ASUS have warned of 15–20% device price increases.
How does the 2026 memory shortage compare to previous semiconductor shortages?
The 2021 COVID-era shortage was supply-side — factory closures, logistics breakdowns, pandemic-driven demand. The 2026 shortage is demand-driven — AI infrastructure is consuming wafer capacity. Demand-side shortages are harder to resolve because the demand source keeps growing.
What is HBM and why is it causing a DRAM shortage?
HBM (High-Bandwidth Memory) stacks multiple DRAM dies vertically. Each NVIDIA B300 GPU requires eight HBM chips with 12 DRAM dies each. Manufacturing HBM consumes the same wafer capacity as conventional DRAM — every wafer allocated to HBM is a wafer that doesn’t go to a smartphone or laptop.
How long is the AI chip shortage going to last?
Based on fab construction timelines, meaningful new capacity arrives in 2027–2028 (Micron Singapore, SK Hynix Cheongju, Samsung Pyeongtaek). Full-scale production at major new fabs like Micron’s Onondaga County facility isn’t expected until 2030.
What is the difference between contract and spot DRAM pricing?
Contract prices are negotiated quarterly between memory suppliers and large buyers through long-term agreements (LTAs). Spot prices reflect what’s available on the open market. During shortages, spot prices spike faster and higher. Counterpoint Research noted DDR4 spot prices ($2.10/Gbit) now exceed HBM3e contract prices ($1.70/Gbit).
How much has the global memory market grown in 2026?
TrendForce projects the global memory market at $551.6 billion in 2026, with DRAM revenue up 144% year-on-year. The market is forecast to reach $842.7 billion in 2027.
Will cloud computing costs increase because of the memory shortage?
Cloud service providers are themselves major memory buyers — North American hyperscalers are negotiating LTAs as of January 2026 — so memory cost inflation is baked into their infrastructure costs. Whether AWS, Azure, and GCP pass those costs through directly is an open question worth keeping an eye on.
Should enterprises buy server hardware now or wait for prices to drop?
NPI Financial advises locking in pricing early, as OEMs are shortening quote validity windows. Gartner’s Atwal recommends buying now: “Whatever you’re getting at the moment is going to be the best price.” No meaningful relief is expected before late 2027.
What does NAND flash pricing look like alongside DRAM?
NAND flash prices rose 55–60% QoQ in Q1 2026, with enterprise SSD pricing up 53–58% QoQ per TrendForce. NAND market revenue is up 112% YoY. And NAND manufacturers are reallocating production lines to DRAM because it’s more profitable, which further limits NAND capacity.
