March 21 (Reuters) - Micron's MU.O shares fell 8% on Friday, as its dour margin forecast took the shine off a robust quarterly revenue outlook driven by demand for its semiconductors used in artificial intelligence tasks.

Micron, one of only three providers of high-bandwidth memory $(HBM)$ chips for data-intensive generative AI applications, forecast adjusted gross margin below estimates on Thursday, as lower pricing for consumer memory chips hits profitability.

After a 1.4% drop in 2024, Micron's shares have gained more than 13% this year as investors bet on improving consumer memory chip pricing and expected the company to benefit from its essential position in AI supply chains.

"NAND Flash oversupply remains a drag on margins," Rosenblatt analysts said in a note on Friday, referring to a type of memory chip used in consumer electronics like smartphones and personal computers.

Soft end-market demand and aggressive buying by electronics suppliers during the pandemic led to oversupply of the consumer memory chips, resulting in weaker pricing.

Micron forecast third-quarter adjusted gross margin of about 36.5%, marginally below analysts' average estimate of 36.9%, according to data compiled by LSEG. The forecast represents a sequential drop of 3 percentage points.

"There has been a challenging industry environment in NAND," Micron's chief business officer Sumit Sadana said on Thursday during a post-earnings call.

Micron has been cutting back on NAND production, resulting in underutilization which spreads fixed costs over a smaller output, hurting margins.

AI DEMAND PERSISTS

The ramp-up of HBM production to meet strong demand for its AI memory chips from GPU market leaders like Nvidia NVDA.O, has also pressured margins.

Micron forecast third-quarter revenue above estimates, owing to AI-linked strength.

"We see high-bandwidth memory as a key growth driver," Morningstar analysts said, adding that they expect "continued AI and data center demand."

(Reporting by Arsheeya Bajwa in Bengaluru; Editing by Sahal Muhammed)

((ArsheeyaSingh.Bajwa@thomsonreuters.com; +91 8510015800;))