Memory chip supplier Micron Technology Inc. almost tripled its revenue in its latest quarter as it stormed past analysts’ expectations, and also provided blowout guidance for the current quarter — yet somehow the stock still fell more than 4% after-hours.

The company reported earnings before certain costs such as compensation of $12.20 per share, crushing Wall Street’s target of $9.31 per share by a huge margin. Revenue for the period rose by 194%, to $23.86 billion, easily surpassing the $20.7 billion consensus estimate.

Micron’s bottom line was boosted even more. With all that extra revenue, the company was able to post a net profit of $13.78 billion in the quarter, up from a profit of just $1.58 billion in the same period a year earlier.

The company’s fortunes have been aided by soaring demand for its memory chips, They’ve become fundamental components of Nvidia Corp.’s graphics processing units, which power artificial intelligence workloads. Of course, memory is also used in almost every other type of computing device too, including personal computers, smartphones, tablets and autonomous cars.

Micron is just one of three memory chip manufacturers in the world and the only one based in the U.S. The others are Samsung Electronics Co. Ltd. and SK Hynix Inc., both based in South Korea. With each new generation of Nvidia’s GPUs needing greater amounts of memory, the world’s supply has become extremely tight, driving up prices. Like Samsung and Hynix, Micron has scrambled to try to increase its production capacity to overcome growing shortages.

For the current quarter, Micron expects to continue benefiting from the voracious demand for memory. It said it’s anticipating earnings of $19.15 cents per share on revenue of $33.5 billion, far ahead of the Street’s forecast of $12.05 in earnings and $24.3 billion in sales.

Chief Executive Sanjay Mehrotra (pictured) said the company set new records across revenue, gross margin, earnings per share and free cash flow in the quarter, and expects to break those records in the current quarter. “In the AI era, memory has become a strategic asset for our customers, and we are investing in our global manufacturing footprint to support their growing demand,” he added.

Micron’s stock has been on a tear. Since the turn of the year, its shares have gained more than 61%, even including today’s after-hours drop. In the last 12 months, the stock is up an incredible 351%. Of the top 10 most valuable technology companies in the U.S., Micron is the only one that has seen its stock increase this year. Oracle Corp. is down 22%, while Microsoft Corp. and Tesla Inc. have also endured double-digit declines.

On a conference call with analysts, Mehrotra said that AI and conventional computer servers are struggling with a “lack of adequate DRAM and NAND supply,” referring to dynamic random-access memory and flash memory chips.

Micron has shifted much of its production capacity towards high-bandwidth memory, which is a newer product that’s able to store data more efficiently and is used almost exclusively by AI servers. It’s easy to see why Micron is doing this, for HMB chips have higher margins than traditional memory products. During the quarter, Micron’s gross margin – profit left over after accounting for the cost of goods sold – rose from 37% a year earlier to 74% at the end of the quarter, and rose 56% from the previous quarter.

Revenue from Micron’s cloud memory business soared 160%, to $7.75 billion. Its mobile and client division saw even faster growth, with revenue rising from $2.24 billion a year earlier to $7.71 billion today.

Traditionally, memory chips have been viewed as a commodity business, with lower margins than other types of silicon products. Memory producers such as Micron typically used to sign short-term contracts with customers, but in the last few months they have begun taking advantage of the shortages to sign much longer deals.

Its customers are scrambling to do this and lock-in a supply of memory for the future. “As AI evolves, we expect compute architectures to become more memory-intensive,” Mehrotra said. “This is why we strongly believe Micron is one of the biggest beneficiaries and enablers of AI.”

The company kicked off volume production of its newest HBM4 memory products specifically for Nvidia’s upcoming Vera Rubin chips in the first quarter, and it’s planning to ramp up production of its next generation HBM4e products in 2027. The company will then pivot to manufacturing “custom HBM” in 2028 for Nvidia’s future Feynman GPUs, which are set to hit the market in 2028.

Mehrotra told analysts that the company expects its capital expenditures to increase “meaningfully” in fiscal 2027, with construction costs set to increase by more than $10 billion as it scrambles to build more memory-making facilities to meet AI’s high demand. The company is currently building new fabrication plants in Idaho and New York to increase its U.S. production capacity. The Idaho site is expected to begin initial production in mid-2027, while the larger $100 billion New York campus will come online about a year later.

Photo: SiliconANGLE