📊 Full opportunity report: Memory: The Unsung Chokepoint Holding Back AI Innovation on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
SK hynix’s chairman warns that AI memory demand is set to surge 60-100% in 2027, but no new capacity is expected to come online. This imbalance could hinder AI progress and escalate geopolitical tensions over memory access.
SK hynix’s chairman, Chey Tae-won, has warned that there will be no meaningful new capacity coming online in 2026 to meet the projected 60 to 100 percent increase in AI memory demand in 2027. This forecast highlights a looming memory shortage that could significantly impact AI development and trigger geopolitical tensions over access to high-bandwidth memory (HBM).
During a press briefing at the Korea Chamber of Commerce and Industry’s Jeju Forum, Chey Tae-won stated that customers are requesting 60-100% more AI memory in 2027 than they are currently receiving. For more on memory market trends, see this analysis of memory supply and demand. With AI now accounting for over half of semiconductor consumption, he estimated demand growth at a minimum of 50-60%, but emphasized that no new capacity is expected to be operational next year.
This supply-demand imbalance is particularly acute in high-bandwidth memory (HBM), which is critical for AI accelerators. Chey described the situation as leading to near-chaotic lobbying, with governments increasingly treating memory access as a matter of economic security. He warned that pressure from national governments could intensify, potentially affecting global supply chains and geopolitical stability.
In response to these challenges, SK hynix announced plans to accelerate capacity expansion, including moving the Yongin mega-cluster’s first clean room to February 2027 and investing over 21.6 trillion won (~$14.5 billion). The company also plans to convert its Cheongju M15X plant into a dedicated HBM facility and review additional fab-site options. However, none of these projects will produce significant capacity before 2027, leaving a capacity gap in 2026.
Models get the headlines.
Memory is the chokepoint.
SK Group’s chairman at the Jeju Forum, per The Korea Herald: customers want 60–100% more AI memory in 2027, governments now treat memory access as economic security — and no company has meaningful new capacity arriving next year.
The gap, in his own numbers
customer requests to SK hynix vs this year. AI already consumes over half of all semiconductors; total demand growth floored at 50–60%.
“No company has meaningful new capacity coming online next year.” The gap year is already locked in — fabs don’t move faster than physics.
Result, per Chey: near-chaotic lobbying — no longer just from companies. Foreign governments are intervening for domestic industries; next, governments pressure governments.
Tighter than the chokepoints you worry about
SK hynix’s race against its own warning
Company figures and projections as announced — none of it lands in 2026.
Half true: unified-memory Apple Silicon doesn’t queue for HBM — a fleet you own is insulated from allocation politics, and owned hardware converts supply-chain risk into sunk cost.
The other half: LPDDR and HBM share DRAM wafer economics — chipflation reaches workstation memory too, and training compute stays fully hostage. Local inference changes who feels the shortage, not whether it exists.
Week tie-in: if memory demand grows into capacity that doesn’t exist, doing the job in 3B parameters on memory you already own isn’t aesthetics — it’s engineering under constraint.
high bandwidth memory (HBM) modules for AI
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications of Memory Shortage on AI and Geopolitics
The projected memory capacity shortfall poses a risk to the accelerated growth of AI technologies that depend on high-bandwidth memory. As demand outpaces supply, costs for AI inference and training are likely to rise, impacting both industry innovation and consumer prices. Additionally, the concentration of HBM production among a few firms raises concerns over geopolitical control and economic security. Governments may increasingly intervene, potentially leading to trade restrictions or supply chain disruptions. This situation underscores the importance of diversifying supply chains and investing in capacity expansion to sustain AI progress.
AI memory upgrade RAM
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Recent Developments in Memory Industry and Demand Trends
In 2026, SK hynix held 58% of the global HBM revenue share in Q1, with Micron and Samsung each holding roughly 21%. The industry has faced persistent demand outstripping guidance for two consecutive years, driven by AI’s rapid adoption. Chey Tae-won’s remarks follow prior announcements of significant investments in capacity, including a 19 trillion won (~$12.9 billion) Cheongju packaging plant and plans to expand the Yongin mega-fab. Despite these investments, the capacity increase will not materialize before 2027, leaving a critical gap in 2026.
Additionally, the industry’s physics mean that new capacity takes years to develop. Chey’s comments also highlight the broader industry trend of consolidation and geopolitical maneuvering over memory access, with Asian exporters and governments increasingly involved in shaping supply chains and pricing.
“No company has meaningful new capacity coming online next year.”
— Chey Tae-won, SK hynix chairman
server memory modules for AI development
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Uncertainties Surrounding Memory Capacity Expansion
It is not yet clear whether SK hynix and other memory manufacturers will accelerate capacity development beyond current plans. The timeline for new capacity coming online remains uncertain, and potential geopolitical interventions could further complicate supply chain stability. Additionally, the impact of emerging memory technologies or alternative architectures on alleviating the bottleneck is still under discussion.
high-performance memory for data centers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps in Addressing Memory Supply Challenges
Industry players and governments are expected to monitor capacity expansion efforts closely. SK hynix and other firms may announce additional investments or partnerships aimed at speeding up capacity growth. Meanwhile, AI developers might adapt by optimizing models for existing memory constraints or shifting toward local inference solutions to mitigate risks. The geopolitical landscape will also influence how supply chain policies evolve in the coming months.
Key Questions
Why is memory capacity so critical for AI development?
Memory capacity, especially high-bandwidth memory like HBM, is essential for training and running large AI models efficiently. Insufficient memory can bottleneck AI performance and increase costs.
What are the main factors causing the memory shortage?
Demand for AI applications is growing rapidly, but capacity expansion has lagged due to long development timelines, high costs, and geopolitical factors concentrating production among few firms.
Could alternative memory technologies or architectures solve this shortage?
Emerging technologies could help, but current industry trends show a heavy reliance on traditional high-bandwidth memory, which remains limited in supply and slow to scale.
How might governments influence the memory supply chain?
Governments may impose export controls, subsidies, or strategic stockpiling, which could further restrict or stabilize supply, depending on geopolitical priorities.
What should AI companies do to prepare for this shortage?
Companies might optimize models for lower memory use, invest in local inference hardware, or secure existing capacity to hedge against future shortages.
Source: ThorstenMeyerAI.com
