📊 Full opportunity report: Seoul Highlights Memory As The Major Bottleneck In AI Development on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
SK hynix’s chairman warns that AI memory demand will outpace supply by 2027, with no significant new capacity coming online. This shortage could reshape geopolitical dynamics and increase costs for AI development.
SK hynix’s chairman, Chey Tae-won, warned during a press briefing at the Korea Chamber of Commerce and Industry’s Jeju Forum that global AI memory demand is expected to increase by 60 to 100 percent in 2027, with no meaningful new capacity scheduled to come online next year. This statement underscores a critical supply-demand imbalance that could impact AI development and geopolitical stability, as memory access becomes a strategic resource.
According to Chey Tae-won, AI now accounts for more than half of total semiconductor consumption, and demand is projected to grow by at least 50–60 percent. Despite this, he emphasized that no significant new memory capacity will be available in 2026, creating a looming bottleneck.
The supply constraints are most acute in high-bandwidth memory (HBM), where SK hynix holds a dominant 58 percent of global revenue in Q1 2026, with Micron and Samsung each holding roughly 21 percent. Chey warned that this oligopoly, combined with soaring demand, is fueling intense lobbying and raising concerns about access as governments begin to treat memory as a matter of economic security.
In response, SK hynix announced a series of capacity expansions, including accelerating the Yongin mega-cluster’s first clean room to February 2027 and investing over $14.5 billion in new facilities. However, these projects will not impact the 2026 supply gap, which is already locked in, highlighting an inevitable shortage next year.
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
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications of Memory Shortage for AI and Geopolitics
This shortage could significantly hinder AI development and deployment, especially for training large models that rely heavily on HBM. It also elevates the strategic importance of memory supply chains, potentially leading to geopolitical tensions and export controls, as countries seek to secure critical infrastructure.
Moreover, the warning from Chey Tae-won indicates that memory prices may remain elevated, impacting device costs and potentially fueling inflation in consumer electronics and enterprise hardware. The concentration of memory capacity among a few firms amplifies these risks, making supply security a national security concern.
Memory Industry Concentration and AI Demand Growth
The global memory market is highly concentrated, with SK hynix controlling over half of the HBM revenue, a situation that has persisted through 2026. Demand for HBM and high-bandwidth memory has outstripped supply guidance for two consecutive years, driven by AI’s rapid adoption and expansion.
Chey Tae-won’s remarks follow broader industry concerns about capacity constraints, pricing abnormalities, and geopolitical considerations. SK hynix’s recent investments aim to address future capacity needs, but these will not impact the immediate shortage forecasted for 2026.
Historically, supply bottlenecks have led to increased prices and strategic reallocations, and the current situation suggests similar risks are intensifying, with governments starting to intervene more actively in memory access issues.
“No company has meaningful new capacity coming online next year.”
— Chey Tae-won, SK hynix Chairman
Uncertainties Surrounding Capacity Expansion and Demand
It remains unclear whether SK hynix and other suppliers can accelerate capacity additions enough to meet the projected demand surge beyond 2027. The timelines for new fabs and their actual output are subject to delays and geopolitical disruptions, which could alter the forecasted shortage severity.
Additionally, the full geopolitical implications, including potential government interventions or export restrictions, are still evolving and could reshape supply chain dynamics.
Next Steps in Addressing Memory Supply Challenges
Industry players are likely to accelerate capacity investments, with SK hynix and others reviewing additional fab-site options. Monitoring government policies and international trade measures will be critical, as they could influence supply chain stability.
Further announcements on capacity expansions, pricing trends, and geopolitical developments are expected over the coming months, shaping the landscape for AI hardware deployment and global supply security.
Key Questions
Why is memory supply critical for AI development?
Memory, especially high-bandwidth memory (HBM), is essential for training and deploying large AI models. Insufficient supply can limit model size, speed, and overall progress.
What are the risks of a memory shortage for the tech industry?
Potential risks include increased hardware costs, delays in AI deployment, and heightened geopolitical tensions over access to critical infrastructure.
Can existing memory hardware be repurposed to ease shortages?
While some inference tasks can run on less specialized memory like LPDDR, large-scale training and high-performance AI workloads remain dependent on HBM, which is in short supply.
How might governments intervene in memory supply issues?
Governments could impose export restrictions, incentivize capacity expansion, or establish strategic reserves, all of which could influence supply and pricing.
When will new memory capacity become available?
According to SK hynix, the earliest significant capacity additions are expected in early 2027, with some projects accelerating to February 2027, but full supply normalization may take longer.
Source: ThorstenMeyerAI.com
