Citrini Research Analyst Jukan: Micron May Become More Important Than Nvidia for Inference

Citrini Research analyst Jukan said Micron is not Nvidia but could become more important than Nvidia in the future, as inference is now directly tied to revenue. According to Odaily, Jukan wrote on X that improving inference performance is not achieved only by adding Nvidia GPUs, because GPUs can be underutilized during inference due to memory bottlenecks. Jukan said adding memory has higher value for inference and argued that the return on investment for inference ultimately depends on memory rather than GPUs. He also questioned why people remain focused on accessing Micron through Nvidia’s framework, saying the issue should be considered more comprehensively. Jukan summarized his view by stating that inference is memory.