UBS' $1,625 price target for Micron Technology is again drawing market attention. The issue is not simply upside potential, but whether artificial intelligence (AI) demand can ease the memory chip industry’s chronic cycle of oversupply and undersupply.

On Aug. 18, blockchain media outlet Decrypt reported that UBS is maintaining a buy rating and a $1,625 price target for Micron. That is about 67 percent above the early trading price of $970.67 on Aug. 18.

The $1,625 price target was not newly presented that day. On May 26, UBS sharply raised its Micron price target to $1,625 from $535 under the name of analyst Timothy Arcuri (티머시 아르쿠리).

What UBS is focusing on is how memory demand is changing as AI data centres spread. The traditional memory chip market has repeatedly seen manufacturers aggressively expand capacity when shortages and rising prices appear, only for prices and profitability to plunge when oversupply follows.

UBS judges the situation could be different in the AI server market. As AI computing performance increases, the speed at which data can be exchanged between processors and memory becomes more important. In particular, as demand rises for high-bandwidth memory (HBM) and high-performance server memory, bandwidth and performance themselves, not just capacity, are becoming key elements of AI infrastructure.

The expansion of long-term supply contracts is another factor underpinning UBS’ bullish view. UBS estimated that up to 30 percent of the industry’s total DDR volume could be tied up in contracts that include customer commitments and partial fixed pricing. It was also reported that hyperscale data centre operators have secured about 60 to 70 percent of their server DDR5 volumes through longer-term contracts.

As such contracts expand, memory makers can forecast future demand more accurately. Customers can also secure needed volumes more reliably when memory shortages occur. UBS said it sees this structure as lowering sharp swings in memory market conditions compared with the past.

The cycle in the memory industry does not disappear, however. Oversupply could occur again if producers expand capital spending on overly optimistic AI demand, or if AI infrastructure investment slows faster than expected.

UBS’ long-term bullish outlook is also premised on such risks. UBS expects Micron can maintain annual earnings per share (EPS) of more than $100 even in 2029. The $1,625 target price was reportedly calculated by applying a price-to-earnings ratio (PER) of about 15 to that long-term earnings outlook and then discounting by 1 year.

Micron’s recent results supported expectations for AI memory demand. Micron’s fiscal third-quarter revenue hit a record $41.46 billion. That marked a sharp increase from $23.86 billion in the previous quarter and $9.3 billion a year earlier.

Over the same period, GAAP net profit was $28.24 billion and operating cash flow was $25.39 billion. EPS was $24.67 on a GAAP basis and $25.11 on an adjusted basis.

Its fourth-quarter outlook was also strong. Micron projected revenue of about $50 billion, a gross margin of about 86 percent and adjusted EPS of about $31.

HBM4 supply is also ramping up in earnest. Micron said it is shipping HBM4 in volume to leading customers and supplying qualification samples to additional customers. It added that all of its 2026 HBM supply volumes have been allocated.

The stock did not immediately move toward UBS’ target price. Micron shares were down 4.06 percent at $970.67 in early trading on Aug. 18. The decline reflected weakened investor sentiment across technology and AI-related stocks as U.S. Treasury yields rose. Key semiconductor stocks including SanDisk, Nvidia and Marvell also fell.

Higher interest rates can weigh on valuations of AI and semiconductor stocks that have risen sharply recently because they reduce the present value of future earnings.

Micron’s tokenised stock, MUB, also weakened. Tokenised stocks can temporarily diverge in price from Nasdaq-listed shares depending on trading hours, liquidity and bid-ask differences.

Ultimately, the key that will shape Micron’s future share price is whether AI can truly change the memory industry’s traditional cycle. If demand for HBM4 and server DDR5 continues to expand and long-term supply contracts also increase, Micron could secure a more stable profit structure than in the past. If a slowdown in AI investment coincides with large-scale capacity expansion, the traditional cycle of falling memory prices could reappear.