Elon Musk has identified memory chips rather than graphics processing units (GPUs) as the biggest potential constraint on the continued expansion of artificial intelligence, highlighting the growing importance of the semiconductor memory industry.
Speaking during a discussion of SpaceX’s second-quarter 2026 earnings, Musk said the current limiting factor for AI development is memory. His comments have drawn particular attention to Micron Technology, one of the world’s major memory-chip manufacturers.
AI Demand Is Growing Faster Than Memory Supply
Musk suggested that global memory supply is increasing by roughly 20% a year, a rate he described as extremely fast for a mature industry.
However, he said demand driven by AI is growing much more rapidly, at around 200% annually.
“If you’ve got demand increasing much faster than supply,” Musk said, basic economic principles would suggest that prices should rise rather than fall.
His comments underscore the growing pressure on memory manufacturers as AI companies build increasingly large data centres and deploy more sophisticated models.
Why High-Bandwidth Memory Matters
The issue is particularly significant for high-bandwidth memory (HBM), a specialized type of memory used alongside advanced AI processors.
HBM allows large amounts of data to move quickly between memory and computing processors, making it an important component of modern AI systems. As AI models become larger and require more computing power, demand for high-performance memory is also increasing.
Micron is one of the leading global producers of HBM and has become an important supplier to the rapidly expanding AI infrastructure market.
The company’s position has attracted increased investor attention as the semiconductor industry prepares for continued growth in AI-related demand.
Musk Says Power Is Not the Main Constraint
Musk also pushed back against the idea that electricity and cooling capacity are necessarily the biggest obstacles to SpaceX’s AI ambitions.
He suggested that the company aims to have substantially more power, cooling capacity and electrical equipment available than the number of GPUs it can actually deploy.
Elon Musk’s AI Film Plans Criticised by Doctor Strange Director Scott Derrickson
Discussing Nvidia’s GPU supply, Musk indicated that SpaceX expects to receive only a relatively small percentage of the chips it wants next year.
His comments suggest that, from SpaceX’s perspective, having enough memory and other supporting infrastructure could become just as important as securing powerful processors.
AI Agents Could Drive Huge Memory Demand
The growing use of autonomous AI agents could add another layer of pressure on memory infrastructure.
Analysts have forecast that AI agents capable of independently carrying out tasks could process around 120 quadrillion tokens per month by 2030, roughly 24 times the current level cited in the report.
Such systems require substantial resources not only for computation but also for storing and accessing information, maintaining context and executing programs.
As AI becomes more capable of handling complex tasks independently, demand for memory could therefore expand alongside demand for GPUs and other computing hardware.
Micron Could Benefit From AI Memory Demand
The potential increase in memory demand has also led analysts to raise their expectations for Micron.
According to Bank of America, stronger AI-related memory demand could potentially push Micron’s earnings per share to $236 in fiscal 2030, substantially above current consensus forecasts.
The forecast highlights how closely the future performance of memory-chip manufacturers is becoming linked to the growth of artificial intelligence.
For now, the AI industry continues to face intense demand for computing infrastructure. Musk’s comments suggest that while GPUs remain essential, the availability of memory could increasingly determine how quickly AI companies can expand their systems.
If AI demand continues to grow faster than memory production, manufacturers could face sustained pressure to increase capacity, while investors will be watching companies such as Micron closely as the next phase of the AI infrastructure boom develops.
