Micron executives explain: Why is there huge investment, and why is the memory shortage unlikely to improve before 2028?

Micron executives stated that despite massive investments, the memory shortage issue will not be resolved before 2028.

Christopher Moore, Vice President of Marketing for Micron Technology's Mobile and Client Business Unit, one of the world's largest memory manufacturers, said in an interview that the current tight situation is not caused by manufacturers proactively adjusting customer portfolios, but rather by the explosive demand from AI data centers combined with the increasing complexity of memory manufacturing, which together are rapidly tightening industry supply limits.

Against this backdrop, even if industry capital expenditure remains high, substantial improvement in memory supply will still take a considerable amount of time, and this tight situation is likely to persist until around 2028.

AI devours DRAM capacity as data center TAM expands rapidly

The structural change in current DRAM demand is key to understanding the shortage problem. The proportion of data center and AI-related demand in the entire DRAM market has quickly risen from 30%–40% in the past to 50% or even 60%. In this context, the whole industry faces the reality of "inadequate capacity," rather than a single manufacturer's resource allocation issue.

Moore stated that this is not a challenge faced solely by Micron, but rather an industry-wide constraint that affects all major memory manufacturers. In the phase of concentrated construction of AI data centers, no manufacturer can ignore this shift in demand. He said:

"This is not Micron's problem, but the problem of the entire industry. We and our peers or competitors are doing our utmost to serve these market segments, but supplies are far from enough. It is indeed a regrettable situation."

Consumers have not been abandoned, only "channel forms have changed"

Regarding market doubts about Micron's withdrawal from the consumer market after exiting the Crucial brand, Micron emphasized that its consumer-side layout is mainly accomplished through OEM methods. The company continues to directly supply LPDDR and DDR modules to major OEMs such as Dell and ASUS and enters the terminal market through integration in finished machines.

In Moore's view, the prioritization of AI demand does not equate to marginalization of consumers, but is a result of expanded TAM (Total Addressable Market) and the temporary inability of the supply side to keep up.

Expansion is not about "buying more equipment," but is a system constraint of manufacturing complexity

Micron explicitly denied the idea that "simple capacity expansion can alleviate the shortage." The main bottleneck in current DRAM production is not equipment quantity, but the loss of production efficiency caused by product specification fragmentation.

As mobile and PC OEMs increase their parallel demand for various capacities such as 8GB, 12GB, and 16GB, foundries must frequently switch between different designs (DID), which directly reduces effective output per unit time. After the explosion in AI demand, manufacturers need to reduce the number of product specifications and extend the continuous production cycle of single products to maximize bit output.

Moore stated:

"You can imagine, if a foundry has many different machines making one type of chip, and you have to stop those machines to switch them to make another type of chip, then output will certainly decrease. Of course, the actual situation is more complicated, but this is the best explanation I can think of. What we're doing now is minimizing the number of chips, minimizing the number of different DID chips, to maximize output, you see? So we're working together with customers."

Slow pace of new capacity release, key turning point points to 2028

Moore revealed that Micron’s ID1 fab in Idaho, USA, which began construction three years ago, has moved its production launch up from the originally planned end of 2027 to mid-2027, but it will still take time for new capacity to ramp up from equipment installation, process maturity, to customer certification before large-scale shipments can truly begin.

In his view, only after the new production lines have fully completed certifications and achieved stable operation will there be any fundamental changes in memory market supply and demand, and this timing is more likely to fall in 2028. Moore further stated:

Memory manufacturers are scrambling to build new production lines, but process-related constraints ultimately force them to accelerate their schedules by several quarters. This means that, for ordinary consumers, the DRAM shortage may last for quite a long period, or at least until AI demand begins to fade.

Risk warning and disclaimerThe market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether any opinion, viewpoint, or conclusion in this article is appropriate for their specific situation. Investment made accordingly is at your own risk.