Micron Technology Inc. is at the forefront of a memory industry experiencing heightened demand fueled by advancements in artificial intelligence workloads. A recent investor group meeting organized by JP Morgan featured key Micron executives, including CFO Mark Murphy, Chief Business Officer Sumit Sadana, and Senior Director of Investor Relations Samir Patodia, who provided a decidedly optimistic outlook on market conditions.
JP Morgan analyst Harlan Sur reaffirmed his Overweight rating on Micron, setting a price target of $350. He highlighted management's strong confidence in broad-based demand for DRAM and NAND memory products, driven by customers increasing their requirements in both sectors. This demand trajectory is buoyed by notable trends in Graphics Processing Units (GPUs) and eXtreme Processing Units (XPUs), with Nvidia projecting additions to a substantial multi-year backlog estimated at over $500 billion extending through the end of 2026. Furthermore, Broadcom has been demonstrating rising volumes in their Tensor Processing Unit (TPU) deployments, supported by forecasts of increased Chip-on-Wafer-on-Substrate (CoWoS) manufacturing capacity, partially attributed to higher TPU demand.
Despite the encouraging demand fundamentals, Micron faces production constraints largely linked to limited clean-room space, which restricts the company's ability to scale supply swiftly. Nonetheless, Sur noted that technological advancements—including node transitions, improvements in production efficiency, and accelerated yield ramp-ups—are expected to enable Micron to achieve at least 20% growth in bit shipments for both DRAM and NAND during calendar year 2026. While this growth improves supply prospects relative to prior expectations, it does not entirely meet the anticipated demand increase, which JP Morgan projects to exceed 30% year over year.
Pricing dynamics remain supported by the persistent supply-demand imbalance. Sur projects that DRAM and HBM demand will persistently exceed supply well beyond 2026, sustaining pricing momentum. Micron executives communicated that, in the medium term, the company can fulfill only approximately half to two-thirds of key customer demand for memory bits. Even with new clean-room capacity scheduled to commence operations starting in 2027, supply growth is expected to be gradual due to intrinsic physical constraints rather than capital limitations.
Micron has advanced the timeline for first wafer production at its Idaho 1 fabrication facility by about one quarter, targeting mid-2027; however, management anticipates a phased capacity ramp-up. This ramp-up is constrained primarily by physical space considerations, as opposed to financial factors. JP Morgan anticipates similar incremental capacity scaling and timing challenges among Micron’s competitors when they introduce new greenfield production sites in late 2027 and into 2028, paralleling ongoing demand growth.
JP Morgan’s outlook foresees continued market tightness sustaining pricing strength for at least the calendar year 2026, with projections indicating average DRAM prices could increase by nearly 60% year over year during this period.
Beyond foundational market dynamics, new AI workloads are expanding the scope of memory demand, particularly influencing NAND usage. Micron management recognizes context window memory management as an emergent application driving rapid NAND uptake. This aligns with Nvidia's announcement at the Consumer Electronics Show (CES) of its Inference Context Window Storage platform, reflecting broader industry efforts by hyperscalers and AI infrastructure adopters developing key-value cache (KV cache) management systems designed to handle growing AI context windows efficiently. These systems facilitate offloading from both HBM and system memory, effectively broadening NAND's role.
Micron expects the attach rates for Nvidia’s Bluefield-4-based systems in this domain to be robust. Importantly, KV cache offload initiatives are not anticipated to diminish HBM product roadmaps already established for the coming years.
Looking further into the future, Micron foresees substantial memory demand arising from physical AI applications, with robotics highlighted as a particularly significant growth vector. Management described plans from their clients as extraordinarily scaled, pointing to memory requirements that could become massive. For instance, a state-of-the-art humanoid robot may utilize between 64 to 128 gigabytes of DRAM alongside 1 to 2 terabytes of NAND memory, figures projected to rise over time. When extrapolated across potentially millions of such units, the resultant demand for DRAM and NAND could be a pivotal contributor to overall industry growth.
Regarding market performance, Micron Technology's stock saw a decline of 3.51%, trading at $327.62 at the most recent close. The shares remain near their 52-week high of $346.30, indicating sustained investor interest despite recent fluctuations.