share_log

Micron surged 11.7% overnight—has the 'bull market thesis' for memory chips really remained unchanged?

Golden10 Data ·  Jun 12 13:48

Micron Technology rose 11.7% on Thursday, helping the memory chip sector stabilize after a series of consecutive declines. However, this does not mean that losses have been fully recovered. Morgan Stanley believes the memory cycle is still accelerating, and AI demand along with long-term agreements could extend the current upswing.

$Micron Technology (MU.US)$Its 11.7% surge at Thursday’s close directly fueled renewed buying interest in the memory chip sector. After consecutive pullbacks, the stock has returned to levels seen about a week earlier—prior to the broad sell-off in semiconductor shares.

This rebound does not mean memory stocks have fully recovered their losses. From recent highs,$Micron Technology (MU.US)$it remains nearly 10% below its June 3 closing peak; meanwhile, the South Korea-listed$Samsung Electronics (005930.KR)$and$SK Hynix (000660.KR)$shares also closed Thursday down 18% and 13%, respectively, from their recent closing highs.

The core of this volatility in memory stocks lies not in the single-day price rebound itself, but in the market’s repricing of cycle sustainability. Investors have recently grown concerned that the memory upcycle may be nearing its end—especially after the sector’s substantial gains this year, with the subsequent correction intensifying such concerns.

Morgan Stanley analyst Shawn Kim does not view the recent correction as signaling the end of the cycle. He believes that for the memory sector—which has posted significant gains this year—a pullback is “inevitable, and ultimately healthy if the current memory bull market is to extend through year-end.”

In a report published Wednesday, Shawn Kim stated: “The cycle is still accelerating, earnings revisions remain robust, and the upswing is more sustainable than most believe.”

AI Bottlenecks Bolster DRAM Demand

Shawn Kim’s conviction that the cycle remains on an upward trajectory hinges primarily on sustained DRAM demand driven by AI infrastructure build-out. Dynamic Random-Access Memory (DRAM) continues to be a critical bottleneck in AI infrastructure expansion, leaving the world’s largest DRAM producers—SK Hynix, Samsung Electronics, and Micron Technology—with significant profit potential.

If the current DRAM cycle follows historical patterns, it “should approach its peak by year-end.” However, Shawn Kim argues that new demand stemming from the rise of agent-based AI could delay the cycle peak by several quarters.

In his view, a reassessment of memory pricing could prolong the current upcycle. Wall Street continues to raise earnings expectations, “validating the rationale behind this rally.”

Long-term agreements between memory chip suppliers and customers could also drive a re-rating of related stocks. Shawn Kim noted that since February, memory prices have nearly doubled and lead times have lengthened, with long-term contracts already locking in a portion of supply.

According to Dow Jones Market Data, Micron Technology trades at a forward price-to-earnings ratio of 9.4 times expected earnings, placing it in the cheapest decile among S&P 500 constituents.

Long-term agreements could reshape industry valuations.

Chris Caso, an analyst at Wolfe Research, also highlighted the impact of long-term customer agreements on industry valuations. He believes such agreements could alter investors’ valuation frameworks for the memory sector.

Chris Caso stated that long-term agreements imply supply expansion will be underpinned by “realistic forecasts,” creating the prospect of “higher valuation multiples.” Since such agreements are still novel in the memory industry, investors may not yet have fully priced in this shift.

However, higher valuations will not materialize automatically. Sean Kim argues that memory stocks can achieve higher valuation multiples only if companies maintain supply discipline.

Over a longer horizon, “demand must translate into sustainable economic returns.” Historically, memory cycles have typically been driven more by supply than by demand, often leading to product oversupply.

Supply shortages boost price expectations.

Memory chip manufacturers currently benefit from supply shortages, granting them pricing power. Chris Caso forecasts DRAM prices will rise 200% this year and 17.5% next year.

In a report issued Thursday, he noted that memory production—measured by bit shipments—could remain constrained through next year due to shortages of cleanroom space required for chip manufacturing.

Chris Caso also projects that by the end of 2026, DRAM prices could be 200% higher than at the end of 2025, while NAND memory prices could increase by 216%. By 2027, prices for both types of computer memory are expected to rise another 17%.

Wolfe Research's view has driven a rebound in Micron Technology's share price. Despite investor concerns about the artificial intelligence (AI) chip market and the health of the high-bandwidth memory (HBM) market used to help AI systems answer queries, the firm raised its price target for Micron Technology to $1,250 per share.

Chris Caso believes that demand for computer memory from AI data centers remains difficult to satisfy, with little noticeable increase in near-term supply to curb rising prices. He forecasts that Micron Technology’s sales next year could reach as high as $226.5 billion, with earnings per share potentially hitting $135.

A price correction does not necessarily undermine the underlying market thesis.

Regarding the outlook beyond 2027, Chris Caso notes declining visibility, with the supply-demand gap likely beginning to narrow by the end of that year.

However, if Micron Technology and its peers remain disciplined in expanding capacity, 'higher prices could persist longer'—potentially even 'into the 2028 calendar year.'

Shawn Kin also expects memory prices to eventually decline as deployments ramp up later next year. However, he argues this may not be negative for memory companies.

A decline in DRAM prices could lower the operating costs of AI models, thereby reducing deployment costs and stimulating additional AI demand. In other words, a price correction does not necessarily signal the end of the cycle; instead, it may unlock new demand by lowering usage costs.

Editor/melody

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to EleBank. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.