Global Memory Shortage Emerges as Critical AI Bottleneck, DeepMind CEO Warns

Google DeepMind CEO Demis Hassabis warned that a severe memory chip shortage is hindering AI model deployment and research, creating a critical bottleneck in the race for artificial general intelligence. Hassabis stated that supply chains cannot keep pace with the surging demand for large models like Gemini. The shortage is driven by the specific requirements of High Bandwidth Memory (HBM) for AI training, distinct from consumer electronics. With the market dominated by Samsung, Micron, and SK Hynix, capacity allocation remains strained. Intel CEO Chen Liwu projected the memory deficit will persist through 2028, shifting the industry's constraint from compute power to infrastructure. Rising costs and delivery delays are already forcing price adjustments across the technology sector.