Micron CEO Says Memory Supply Tightness Will Extend Beyond 2027
In a CNBC interview, Micron CEO Sanjay Mehrotra said the global shortage of memory chips will persist into 2028, citing artificial intelligence workloads and expanding fabrication timelines as primary drivers. Mehrotra noted that AI systems deployed across data centers, automotive, and consumer devices require significantly more memory at higher performance levels to operate efficiently. He added that the supply-demand gap is expected to widen through 2027, reinforcing a structural shift away from cyclical spot pricing.
Operating margins for memory suppliers have climbed to approximately 80%, prompting major buyers to proactively request three-to-five-year contracts that lock in capacity through 2028 or 2030. Micron and peers Samsung and SK Hynix are implementing these long-term agreements, which include advance payments and price floors to guarantee volume. To support the extended demand, Micron is accelerating a $200 billion investment plan across facilities in the United States and abroad. While cloud data centers currently represent Micron's largest revenue driver, the company emphasized that consumer, automotive, and industrial applications account for 40% of its sales.
From the sources (4 posts)
@jukan05KIS: On Memory LTAs With suppliers' commodity DRAM operating margins already at 80% and gross margins above 90%, prices cannot be raised endlessly on supply and demand dynamics alone. What matters is that at price levels where supplier ope
@dnystedtRT @mingchikuo: The memory supply-demand gap will keep widening through 2027. That is the real reason Apple is lobbying the White House to…
@dudette1508The memory shortage isn't ending. Executives from SK Hynix, Micron, and Samsung agree: the supply-demand mismatch will intensify through 2027, with shortages lasting into 2028. Why? Massive context windows and the rise of robotics. 🧵 #memo
@dnystedtMicron CEO Sanjay Mehrotra sat down with Jim Cramer on CNBC to talk chips. Key comments: “Micron is firing on all cylinders.” “Every AI system, regardless of the device it is in, requires more memory at higher performance in order to re