AI servers
2026-07-03 20:01:01Morgan Stanley Lifts SIMO to $400 as AI Servers Reshape the NAND Cycle
Morgan Stanley has sharply raised its price targets on key storage names, arguing that AI infrastructure is fundamentally changing the NAND demand cycle. The most aggressive move was on Silicon Motion, whose target was lifted from $155 to $400, based on 23x expected 2027 EPS and expectations that revenue will hit a record in 2026. Longsys and Phison also saw higher targets, though both remain rated Equal Weight, highlighting that the AI storage upcycle is unlikely to benefit all module vendors equally.
The report’s main thesis is that AI-driven NAND demand will extend through 2027. Global NAND supply-demand is projected to remain about 2% oversupplied in 2025, then swing to a 15% shortage in 2026 and still show a 9% deficit in 2027. Total demand is estimated at 1,111 EB, 1,250 EB, and 1,484 EB for 2025-2027, versus supply of 1,128 EB, 1,058 EB, and 1,347 EB. The key driver is not a broad recovery in smartphones or PCs, but rising demand from AI servers, cloud SSD procurement, enterprise storage, and boot drives.
Pricing data already reflects this divergence. Channel checks for 3Q26 show TLC enterprise SSD prices up about 30% quarter-on-quarter, server DRAM up 20%, and legacy DRAM such as DDR3/DDR4 up 30%-40%, while consumer NAND pricing remains more constrained. Morgan Stanley sees SIMO as especially well positioned through enterprise SSD controllers and AI boot drive modules, but also warns that the bullish outlook depends on successful ramp-up, customer adoption, and sustained AI server demand. Looking further out, 2028 remains the key risk window, with YMTC capacity expansion and AI demand growth determining whether shortages persist or the market swings back into oversupply.