gaming

Micron: AI HBM demand and slow fab builds mean no near-term fix for memory crunch

Micron executive Sumit Sadana warned that surging demand for HBM used in AI servers is consuming wafer supply and squeezing availability for other memory products. Building the new leading-edge fabs needed to increase wafer output requires greenfield construction and long ramp-up times, far more complex than data centers. Micron has no timeline for when wafer supply will meet demand, creating continued uncertainty for device makers and consumers.

• Sumit Sadana is Micron’s executive vice president and chief business officer and made the remarks at a tech and finance forum.
• Rapid growth in AI server-critical HBM is pressuring the wafer supply left for other memory markets.
• Expanding capacity requires greenfield fabs—building entire fab clusters on empty sites—which are among the most complex engineering projects and take a long time to come online and ramp.
• Micron does not have line of sight on when wafer supply will intersect demand, leaving uncertainty for buyers of memory modules and electronics.