Memory prices drop, Citi slashes Micron target price!

Citing the apparent recent weakness in the spot price of DDR5 DRAM, Citi willMicron TechnologyThe target price was significantly lowered by 17%, but the Buy rating and all earnings forecasts were kept unchanged, believing that the long-term logic of AI-driven storage demand has not been fundamentally shaken.

According to Chasing Wind Trading Desk, Citi analyst Atif Malik lowered Micron's target price from $510 to $425 in a research report released on March 31, a decrease of 17%. At the close of trading on March 30, Micron shares were trading at $321.80, with about 32% upside from the newer price target.

According to the reported data, the spot price of mainstream DDR5 16GB DRAM has recently declined by about 6%. This round of decline is mainly attributed to the marketGoogleConcerns with TurboQuant technology. This technology is believed to potentially compress the memory consumption of AI inference, thus raising market concerns about the outlook of storage demand.

Micron and its peers have commenced negotiations with ultra-large-scale cloud computing vendors for a three-to five-year strategic long-term agreement with a framework covering locked-in base purchase volume, prepayment arrangements and quarterly price adjustment mechanism based on market conditions, which is expected to effectively support the contract price.

Accordingly, Citi analysts lowered their valuation benchmark for Micron from 6x to 5x cycle bottom P/E, and based on the expected peak EPS in 2027, they arrived at a new price target of $425, which is in line with the historical valuation bottom range of 5x to 6x in DRAM upcycles.

Notably, the bank is leaving its entire financial forecast for Micron unchanged: core EPS forecast of $58.46 for fiscal 2026 and $94.55 for fiscal 2027. The current share price corresponds to approximately 5.5 times the expected P/E in 2026 and approximately 3.4 times in 2027, which is at a historically relatively low level.

According to the report, Micron and its storage peers are negotiating with ultra-large-scale cloud computing vendors on strategic long-term agreements for three to five years, covering arrangements such as locking of basic purchase volume, setting of prepayment mechanism and dynamic adjustment of quarterly prices according to market conditions.

The above-mentioned long-term agreement framework is expected to build structural support for contract prices. Micron generates about 79% of its revenue from its DRAM business, and contract price action is critical to its earnings visibility.

TurboQuant isGoogleA model compression technology developed by the research team optimizes the KV Cache (Key-Value Cache) calculation through new quantization methods (including PolarQuant technology and QJL algorithm) to reduce the memory occupation in the AI inference process.

According to the report, the impact of TurboQuant on storage requirements is similar to that of the previous DeepSeek event: on the surface, the efficiency improvement technology will reduce the computing power and memory cost of a single AI reasoning, but the lower usage cost will further release the application volume and finally push up the total computing power and memory demand.

From the historical perspective, cheaper technologies often drive the demand for more technologies, and the AI field is no exception. From this, the bank judged that the recent correction of spot price reflects more short-term market sentiment disturbance than the trend reversal of AI storage demand.