Institute briefing
AI Infrastructure Drives Sustained DRAM Shortage Through 2027
The global DRAM market faces prolonged supply tightness until 2027, primarily due to intense demand for high-bandwidth memory (HBM) essential for AI infrastructure, creating significant implications for AI development and resource accessibility.
What happened
The global market for DRAM (Dynamic Random-Access Memory) is projected to experience sustained tightness in supply through 2027, primarily driven by the escalating demand for high-bandwidth memory (HBM) used in artificial intelligence (AI) infrastructure. This persistent demand from the AI sector is creating a significant imbalance in the memory market, impacting overall availability and pricing for various computing applications.
Why it matters
For businesses and policymakers, this prolonged DRAM scarcity underscores a critical supply chain vulnerability within the rapidly expanding AI ecosystem. The intense competition for HBM, essential for powering advanced AI models and data centres, highlights the foundational hardware dependencies of AI development. This situation necessitates strategic planning for procurement and resource allocation, as it could influence the pace of AI innovation and deployment across industries, potentially favouring larger entities with greater purchasing power. Moreover, it raises questions about equitable access to critical AI infrastructure components and the broader economic implications for sectors reliant on memory technologies.
By the numbers
- 2027: The year through which DRAM supply is expected to remain tight due to AI infrastructure demand.
The Institute take
The ongoing strain on DRAM supply, particularly for HBM, is often framed as a simple market dynamic, but it reveals a deeper challenge in AI governance: the concentration of critical resources. While the focus is typically on the ethical implications of AI algorithms, the physical infrastructure underpinning AI systems presents its own set of governance considerations.
The race for high-performance memory for AI is not just an economic bottleneck; it's a structural issue that could exacerbate existing inequalities. Responsible AI leadership demands proactive strategies to diversify supply chains and explore open-source hardware initiatives. Relying solely on market forces to resolve this could inadvertently centralise AI power, making advanced AI capabilities inaccessible to smaller innovators and potentially hindering the development of more diverse and equitable AI applications.
Briefing notes
Questions this story answers
01What is causing the prolonged tightness in DRAM supply?+
The sustained tightness in DRAM supply is primarily caused by the escalating demand for high-bandwidth memory (HBM) used in artificial intelligence (AI) infrastructure.
02How long is the DRAM supply expected to remain tight?+
DRAM supply is projected to remain tight through 2027 due to the ongoing demand from AI infrastructure.
03What are the implications of this DRAM scarcity for AI development?+
This DRAM scarcity highlights a critical supply chain vulnerability for AI, necessitating strategic planning for procurement and resource allocation, and raising concerns about equitable access to critical AI infrastructure components.
04What governance challenge does the DRAM shortage present for AI?+
The DRAM shortage, particularly for HBM, reveals a deeper AI governance challenge related to the concentration of critical resources, potentially centralising AI power and hindering access for smaller innovators.