AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Could A Memory Shortage Last Until 2030? SK Hynix Weighs In on ThorstenMeyerAI.com

Age 18–24?Offer from Amazon

Prime made for students and young adults

  • Fast, free delivery for dorm and study essentials
  • Prime Video and Amazon Music included
  • Member-only deals
Try Prime for Young Adults Free trial for eligible 18–24 year olds
As an affiliate, we earn on qualifying purchases.

TL;DR

SK hynix said on October 28 that memory demand is expected to exceed supply through 2030, citing the difficulty and long lead time of adding production capacity. The company also said higher memory costs and shortages could reduce PC and mobile-device shipments; how much prices or shipments will change remains uncertain.

SK hynix said on October 28 that it expects memory demand to exceed supply through 2030, citing the time needed to build new factories and the technical challenges of producing advanced chips. The South Korean supplier also warned on its earnings call that higher memory costs and supply shortages could reduce PC and mobile-device shipments, linking the AI infrastructure boom to potential pressure on consumer electronics.

In its official third-quarter release, SK hynix said supply was “not keeping pace with the growth of the market.” It expects the imbalance to persist through 2030 because of the “technical difficulty of leading-edge processes” and the “long lead time required to build new production facilities.” That is a company forecast, not a guarantee that every type of memory will remain in short supply for the same period.

The company’s release did not discuss PC or phone sales. On the earnings call, Chief Financial Officer Kim Woo-Hyun said shipment volumes for PCs and mobile devices were expected to decrease because of “cost burdens from rising memory prices and supply shortages.” He added that expected growth in AI infrastructure demand would more than offset that decline for SK hynix. The comment describes the company’s outlook; it does not establish how many devices will be affected across the market.

SK hynix executives said customers’ requirements for high-bandwidth memory (HBM) were more than twice what the company could supply at the time of the call. The company also said long-term agreements with 14 major customers, typically lasting five years, cover more than half of its production capacity for 2027 and 2028. Those agreements include deposits, according to the source material, indicating that some future output has already been committed.

At a glance
reportWhen: SK hynix issued its forecast on October…
The developmentSK hynix forecast that memory demand will outstrip supply through 2030, extending concern about shortages beyond the near term.
Thorsten Meyer AI
Supply

Infographic 01 · Where the memory goes

AI data centers get first call on memory.

High-bandwidth memory (HBM) for AI chips and the memory in your phone come out of the same factories. HBM needs far more wafer per gigabit, and its buyers sign years ahead.

22%of DRAM wafers go to HBM by the end of 2026 (TrendForce estimate, top three makers)
9%of all DRAM bits are HBM at the same time (TrendForce estimate)
≈3×wafer per gigabit for HBM vs. ordinary DRAM (my arithmetic from TrendForce)
>50%of SK hynix capacity for 2027 and 2028 already under long-term contracts

The bakery problem

HBM share of wafers vs. share of the memory bits they produce

End 202518%of wafers · 8% of bits
End 202622%of wafers · 9% of bits
End 202730%of wafers · 13% of bits

Same factories, same wafers: every wafer used for AI memory is missing from phone and PC memory.

Who gets first call

Long contracts lock up supply years ahead

  • SK hynix: deals with 14 major customers, usually five years, with deposits; more than 50% of 2027–2028 capacity
  • Micron: more than 75% of fiscal 2027 shipments already committed
  • HBM demand: customers want “more than double” what SK hynix can supply today
  • Cloud budgets: memory ≈ 47% of big cloud providers’ capex in 2026, 68% in 2027 (TrendForce estimates)

Three makers, three horizons, one customer’s view

  • SK hynixThrough 2030Demand above supply (company forecast)
  • SamsungInto 2028Tight supply, based on market research
  • MicronNo “line of sight”Tighter in 2027 and 2028 than in 2026
  • Acer (customer)Easing from 2H 2027Says suppliers talk up the shortage

HBM is a wedding cake, phone memory is bread. Same ovens. The cake customers pay more and order years ahead.Thorsten Meyer · my analogy

Sources: TrendForce, 2 Jun (wafer and bit shares, estimates); TrendForce, 25 Aug (capex shares, estimates); SK hynix, 28 Oct 2026; SK hynix call, 28 Oct; Micron call, 30 Sep; The Register, 1 Oct; Samsung call, 29 Oct; Acer via Tom's Hardware / DigiTimes. Wafer-per-gigabit ratio: my arithmetic. All horizons are forecasts.
Powered by Thorsten Meyer AI

AI Demand Meets Consumer Memory

Memory is used in phones, PCs and servers as well as AI systems. HBM, a form of DRAM built in stacks for use beside AI processors, is drawing strong demand from data-center customers. SK hynix says this demand is growing faster than supply, while factories and advanced production processes take time to expand.

The potential effect reaches consumers through both availability and price. The source reports that US prices for a 32-gigabyte DDR5-6000 kit were at least $447 on October 29, compared with a lowest recorded price of $72 in Tom’s Hardware’s index. That is a reported price comparison for a specific kit, not evidence that every memory product has risen by the same amount. The source also cites TrendForce estimates of sharp increases in contract prices for conventional memory during 2026.

If component costs remain high, device makers may raise retail prices or change specifications to manage costs. The source says some manufacturers have reduced memory, storage or camera specifications in mainstream phones, and that some have reused recycled memory chips. Those examples are attributed to TrendForce and Counterpoint; their scale across products and markets is not established here.

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

Ninja Kitchen System Blender and Food Processor with BlendSense Technology, 1800-Watt, 72 oz, Black (TB401)

Ninja Kitchen System Blender and Food Processor with BlendSense Technology, 1800-Watt, 72 oz, Black (TB401)

Ninja Foodi Smart Dual Heat Air Fry Oven, 13-in-1, Silver, SP351

Ninja Foodi Smart Dual Heat Air Fry Oven, 13-in-1, Silver, SP351

As an affiliate, we earn on qualifying purchases.

How HBM Pressures Chip Supply

DRAM provides temporary working memory in devices, while NAND flash stores files and applications even when power is off. HBM is also DRAM, but its stacked design moves data quickly between memory and processors, making it useful for AI computing. SK hynix, Samsung and Micron are among the companies competing to supply this market.

The source material, citing a June 2 TrendForce estimate, says HBM is expected to take about 22% of wafer starts at the three largest memory makers by the end of 2026 while accounting for about 9% of memory bits produced. For the end of 2027, it gives estimates of about 30% of wafer starts and 13% of bits. These figures are estimates, and the source’s comparison that HBM requires roughly three times the wafer per gigabit is its own calculation from those estimates.

Other suppliers have described a similarly tight market, though with different timelines. The source says Samsung expects tight supply into 2028, while Micron has said it has no “line of sight” to a supply-demand balance. These statements support the view that suppliers see constraints ahead, but do not independently confirm SK hynix’s precise through-2030 forecast.

“Supply is not keeping pace with the growth of the market.”

— SK hynix, in its October 28 earnings release

Forecasts And Market Limits

The through-2030 outlook is SK hynix’s forecast. The available material does not give a year-by-year estimate of shortages by memory type, or say when particular products might return to balance. It also does not establish how much of the company’s projected supply constraint reflects production limits, customer demand, or choices to prioritize more profitable products.

There is a commercial dispute around how shortages are described. The source reports that Acer has accused suppliers of talking up tight supply to protect margins. That is Acer’s allegation; the supplied material does not include supporting evidence or a response from the suppliers. SK hynix and other memory makers also stand to benefit from high prices, a relevant interest when evaluating their forecasts.

Shipment projections and price effects remain uncertain. SK hynix said higher costs and shortages could weigh on PC and phone volumes, but the supplied information does not quantify the expected decline from this forecast. Nor does it show how much manufacturers will pass component costs on to buyers, absorb them, or offset them by changing device specifications.

Signals From Upcoming Results

The next evidence will come from whether memory producers add capacity, how much of that output is allocated to HBM, and whether suppliers report any easing in customer demand or lead times. SK hynix’s long-term customer agreements for 2027 and 2028 mean that its reported commitments and production plans will be relevant indicators as those years approach.

Investors and device buyers will also be watching results and guidance from Samsung and Micron, as well as updates from PC and phone makers. The key questions are whether forecasts for tight supply continue, whether component costs feed into retail prices or device specifications, and whether shipments change as SK hynix expects. The supplied material does not identify a specific date when the market is expected to rebalance.

Key Questions

What did SK hynix say about the memory shortage?

On October 28, 2026, SK hynix forecast that demand would exceed supply through 2030, citing the time and technical difficulty involved in expanding advanced manufacturing.

Does the forecast mean all memory will stay scarce until 2030?

No. The statement is SK hynix’s company-wide outlook in the supplied material. It does not give a separate shortage timeline for every memory type or product.

Why is AI demand affecting memory for consumer devices?

AI processors use HBM, which is made from DRAM. SK hynix says HBM demand is strong and that expanding production takes time; the source also cites estimates that HBM uses a larger share of wafer capacity relative to the memory bits it produces.

Will phones and PCs become more expensive?

Higher memory costs could put pressure on device prices or specifications, but the supplied material does not establish how much manufacturers will pass on to buyers. SK hynix said rising costs and shortages could reduce PC and mobile-device shipments.

Is the long-shortage forecast independently confirmed?

No. It is SK hynix’s forecast. The source also reports that Samsung expects tight supply into 2028 and that Micron has no “line of sight” to balance, but those statements do not confirm the same through-2030 timeline.

Source: ThorstenMeyerAI.com

You May Also Like

10 NAS Systems To Compare For Home And Business Storage In 2027

A 10-model NAS comparison for home and business buyers, focused on bay count, drive options, redundancy, setup and storage needs.

ArXiv will ban researchers who upload papers full of AI slop

ArXiv announces a crackdown on AI-generated papers with untrustworthy content, including a one-year ban for authors submitting incontestable evidence of unchecked LLM output.

The Model Is Only 10%: The Real Lesson of the New SDLC

A new Google whitepaper reveals that the key to effective AI development is not the model size but the harness and context engineering, shifting focus from models to configuration.

Mistral Large 4: Why It Stands Out Abroad But May Not Power Your Agents

Mistral Large 4 scores 38.4 on Artificial Analysis, but trails leading US and Chinese models and costs more per benchmark task than some rivals.