DRAM Shortage Until 2030: How to Position Yourself on Memory Giants?
A Memory Supercycle That Lasts
During the groundbreaking ceremony for its future factory in Indiana, SK Hynix CEO Kwak Noh-Jung stated that he sees no clear signs of a cycle reversal and anticipates a memory shortage until the end of 2030.
Beyond SK Hynix, analysts are echoing this sentiment: Morgan Stanley speaks of a "memory supercycle," Bank of America has pushed its scenario to 2030, and Deloitte mentions a "RAMageddon" where new production capacities would not arrive before 2029 or 2030.
The cause of this tension is mechanical: HBM (high bandwidth memory used in AI chips) and traditional DRAM come from the same factories and compete for the same silicon wafers. Since HBM is much more profitable and concentrates nearly three times more capacity per bit, manufacturers prioritize it, which dries up the supply of traditional DRAM.
As a result, DRAM prices are skyrocketing. In the second quarter of 2026, their contractual prices surged by 58 to 63% in just one quarter according to TrendForce, the largest increase in 10 years. A 64 GB server module thus rose from about $450 at the end of 2025 to over $900 at the beginning of 2026. The pace is now slowing, but TrendForce still anticipates increases until the end of the year.
The Three Giants That Are Winning
The DRAM market is an oligopoly: Samsung, SK Hynix, and Micron together control over 95% of global production. In the HBM segment, the most strategic part of the AI cycle, SK Hynix dominates with about 60% market share and the status of preferred supplier to Nvidia.
This surge is directly reflected in the accounts: in the first quarter of 2026, SK Hynix surpassed 50 trillion won in revenue (about $35 billion) for the first time, with operating profit multiplied by 5 year-on-year and a record operating margin of 72%. Micron, on the other hand, reported a record gross margin of over 80% in its third fiscal quarter, which ended in late May 2026.
On the stock market, prices have followed the same growth: in 2025, Micron's stock surged by about 240%, Samsung more than doubled, and SK Hynix's market capitalization nearly quadrupled. However, these values have significantly corrected since their peaks, a reminder that past performance does not guarantee future results.
Should One Expose Themselves to the Memory Sector?
SK Hynix's optimism should be taken with a grain of caution: as the world's leading seller of HBM, the company has every interest in maintaining the narrative of a shortage that boosts its prices and justifies its heavy investments. However, its diagnosis is corroborated by independent analysts, and the demand is indeed real.
For an investor looking to ride the AI wave, memory remains one of the most direct bets: unlike many stocks in the sector, orders are concrete and often contracted several years in advance.
Caution is still warranted, as memory is a historically cyclical industry, and new factories expected from 2028 could eventually ease price pressure. The sector is also very volatile: Samsung and SK Hynix lost 23 to 35% in a month this summer due to price concerns before Goldman Sachs reiterated its buy recommendation.
Additionally, there is the threat of Chinese competition. CXMT in DRAM, whose revenue surged by about 700% year-on-year in the first quarter of 2026, and YMTC in NAND are ramping up with new factories, to the point where CXMT is already catching up to Micron in production capacity.
However, the threat mainly weighs on traditional DRAM, less so on HBM, the most profitable segment, where the three giants maintain several years of lead.
-- Price
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.
You may also like

Coinhouse Acquires Tilvest and Strengthens Its Position in Crypto Management

"Technology Takes a Backseat": How Stablecoins Transition from Savings to Everyday Payments

BTC Drops 62% Against Nasdaq, Resistance at 78500

Banks Improved Their Profitability, But Concerns Over Delinquency Persist: Key Insights from Recent Financial Statements

CME Targets ETFs: The First FCA-Regulated Multi-Asset Crypto Indices Are Born

Crypto Funds: $3.2 Billion in One Week, Bank of America Hasn't Seen This Since October 2025

What Is Worth Preserving: Rupture On Remains, Decay, And The Collector's Dilemma

Fundamental Analysis of Cryptocurrencies: How to Evaluate Digital Assets Before Buying

AI Forces Diversification of Bitcoin Custody

Natura Changes CEO: Carlucci Returns After a Decade

Warren Buffett Turns 96: The Investments That Made Him a Wall Street Legend

After MiniMax's Stock Price Rebounds 92%: What Remains When a Model Company Removes the Model

Meta Faces Dilemma After AI Layoffs: Cybersecurity Incidents Up 40%, Cleanup Time Up 70%

Bull Secures €387.8 Million Contract for LUMI-AI Supercomputer

Chinese AI Through the Eyes of Silicon Valley VCs: Sanctions Didn't Lock It Down, but Instead Brought Out a 'Monster'

Consolidation in Ranges and Rate Reevaluation: Trader Assesses Bitcoin and Ethereum Movement Scenarios

Digital Ruble Fails to Generate Significant Interest Among Russians, Says Sberbank

AI Scam Hits $3.2 Million Per Incident, Crypto Security Battlefield Shifts from 'Code' to 'Scams'

Matt Damon to Keynote at Ripple Swell 2026 in New York

How Paintings Could Have Become Part of the Digital Economy, but Didn't

Shein's IPO and China's PMI: What Changes for Markets

From Pay to One-Stop Asset Management, BiyaPay Expands Global Diverse Financial Services Boundaries

Trump's Cryptocurrency Scheme Rakes in $1.4 Billion While Investors Lose $4.7 Billion

AI Creates Abundance, BTC Creates Scarcity: What Web3 Truly Changes Is Not Production Relations, But Value Relations

Cryptocurrency Arbitrage and Price Scanning: How Quants Identify Inefficiencies in Funding Rates and Spreads Using Trading APIs

Why Stablecoins Can't Serve as Credit Despite Being Transferable and Store of Value?

Bernstein Comments on Seven Memory Types: After HBM, DRAM and NAND Compete for the Next Trillion-Dollar Market

After Impacting Two Generations, Meta Ordered to Pay $18 Billion

What is Thinking Cat (HMM)? Reasons Why Ownerless Tokens Can Still Fail











