DeepMind CEO Demis Hassabis told CNBC that the global memory-chip supply chain is constrained, causing shortages of high-bandwidth memory (HBM) used by large language model training. He said capacity bottlenecks at a few suppliers—Samsung, Micron, SK Hynix—limit deployments and slow research, and noted Google’s TPU designs help but cannot avoid component shortages. Google expects high AI infrastructure spending into 2026.
Key Points
- 1Warns of constrained global memory supply, especially HBM, limiting AI model training capacity.
- 2Highlights choke points at few suppliers—Samsung, Micron, SK Hynix—creating competition and price pressure.
- 3Advises practitioners to plan for constrained HPC memory, prioritize experiments and long-term infrastructure spend.
Scoring Rationale
Industry-wide, credible executive warning with strategic implications, but offers limited novel technical information.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
Active Search Campaigns by BudgetEasyHigh CPC Clicks & Poor Landing PagesMediumCampaign ROAS by Attribution ModelHard
250 free problems · No credit card
See all Ad Tech problems
