Nvidia's Memory Strategy Poses Long-Term Risk to Micron Despite Near-Term Strength

Micron Technology's stock has climbed more than 500% in a year, reaching $975 per share. That kind of ascent raises a different question than whether the business is sound. It asks: what would need to shift for this story to unravel, and who holds the power to make that shift? The answer, for Micron, is Nvidia.

Nvidia's second-quarter filing revealed something worth examining closely. The company's supply commitments for memory jumped from $119 billion to $279 billion, a figure tied directly to its memory purchasing plans. That commitment schedule then drops sharply after fiscal 2029, according to reporting from The Wall Street Journal. On the surface, this looks like Nvidia is locking in massive future demand. But a commitment is a plan, and plans change. Nvidia is betting on memory consumption, but that spending could be revised if the company's strategy shifts.

What matters more for Micron's long-term prospects, though, is what Nvidia is doing with its chip designs. Design choices made today lock in memory requirements for years. Consider the Rubin CPX, a processor built to handle the initial stage of an AI prompt—when the system processes everything fed to it before generating a response. Nvidia originally planned to use GDDR7, the same memory type found in gaming graphics cards. That memory costs roughly one-fifth as much per gigabyte as the premium stacked memory that generates Micron's profits, and it avoids expensive packaging steps. In August, analyst Ming-Chi Kuo reported that Nvidia had revived the project, but this time with stacked memory—though at 168 gigabytes per chip rather than the 288 gigabytes in Nvidia's flagship processor. Read that sequence as a company testing how little premium memory it can use in workloads that don't demand it. Nvidia has every incentive to keep experimenting. The real risk for Micron lies in how Nvidia designs its chips going forward, because those decisions can shape memory demand for years or even permanently.

Micron's own expansion strategy adds another layer of risk. The company plans to double its monthly output of stacked memory to roughly 100,000 wafers by year's end. But its market share in that category slipped to 18% in the second quarter, while Samsung Electronics surged to 33%, according to Counterpoint Research. Samsung reached its first billion dollars in stacked memory revenue within four months of beginning shipments. So Micron is adding substantial capacity to a market where it appears to be losing ground. In June 2026, Nvidia CEO Jensen Huang confirmed that the company had qualified all three major memory suppliers—Samsung Electronics, SK Hynix, and Micron—to provide HBM4 memory for its next-generation Vera Rubin platform. This cuts both ways. It keeps Micron in the running and gives Nvidia three competing suppliers to pit against each other, which benefits Nvidia's negotiating position and cost structure.

For Nvidia, the direction is clear. Having three qualified suppliers competing for its business is a cost advantage, especially when its memory bill is enormous. Nvidia can design chips around whatever memory makes economic sense and use multiple suppliers to manage both costs and supply risk. For Micron, the near term remains solid. Management has guided that tight supply conditions will persist beyond calendar 2027, and SK Hynix CEO Kwak Noh-jung has stated the shortage could extend through 2030. But the inflection point arrives when Micron's new capacity comes online at the same moment Nvidia's designs require less premium memory per chip and its commitment schedule thins out. That convergence—excess supply meeting reduced demand—is where the risk crystallizes. Investors holding Micron should watch two signals closely: Micron's quarterly share of the stacked memory market and whether Nvidia extends its supply commitments into fiscal 2030. That date may seem distant, but it will determine whether Micron's current valuation holds.