NVIDIA, citing its deep visibility into global AI demand, anticipates the memory shortage will continue for multiple years, a critical constraint on compute supply. The company is exploring architectural approaches like CXL and memory pooling to mitigate these bottlenecks, potentially shifting its view on SRAM-based technologies like Groq's.
AI-driven memory demand will continue to outpace supply for years, forcing architectural shifts and increasing costs for consumer device makers like Apple.