Micron Beats Earnings Estimates With $54.23 Billion in Sales as AI Memory Demand Surges

Micron delivered $54.23 billion in fiscal fourth-quarter revenue on September 30, beating Wall Street’s expectations as demand for AI infrastructure drove its memory business. Its forecast for the next quarter was stronger than analysts expected, too. The earnings release puts a concrete sales figure behind a shortage that is rewarding memory suppliers and raising costs for buyers.

Revenue exceeded the $51.07 billion consensus compiled by LSEG. Adjusted earnings reached $33.42 a share, also above expectations. Micron’s shares rose slightly in extended trading after the results, following a gain of more than 500% over the past year. That earlier rally reflected a worldwide memory supply crunch as demand for chips used in AI models and workloads climbed.

The completed quarter

• Adjusted earnings: $33.42 a share, versus the $31.61 LSEG consensus.
• Net income: $37.7 billion, up from $3.2 billion a year earlier.
• Year-earlier revenue: $11.32 billion, compared with the latest quarter’s $54.23 billion.

For its fiscal first quarter, Micron expects about $61.5 billion in revenue and adjusted earnings of $38.15 a share. Both targets exceed the LSEG consensus, which called for $57 billion in revenue and $35.40 in adjusted earnings. The outlook extends management’s growth expectations beyond the completed quarter, but remains a forecast rather than sales or profit already recorded.

$39.8 billionFiscal Q4 DRAM revenue

Micron’s DRAM revenue increased 343% from a year earlier.

73%Share of total sales

DRAM accounted for nearly three-quarters of Micron’s quarterly revenue.

Hendi Susanto, a portfolio manager at Gabelli Funds, described the release as another strong earnings beat and higher outlook. In an email after the results, he said he had not heard negative data points suggesting the memory cycle would turn downward soon. That is an investor’s assessment of the cycle, not a guarantee that today’s supply pressure will persist.

High-bandwidth memory, or HBM, is made from stacks of dynamic random-access memory, known as DRAM. Advanced graphics and central processors from Nvidia and AMD need increasing amounts of HBM to handle AI workloads. Micron is the only U.S.-based HBM maker, alongside South Korean suppliers SK Hynix and Samsung. The leading producers cannot currently make enough to meet demand.

On the earnings call, CEO Sanjay Mehrotra said Micron has a strong roadmap for future HBM products. He also said the company is working with Nvidia on what he called the industry’s first custom HBM implementation. That collaboration describes work underway; the statement did not include a release date or specifications for a finished product.

Micron is investing $250 billion to build two campuses for HBM production, CNBC reported. The larger site broke ground in Clay, New York, in January. Its first new fabrication plant in Boise, Idaho, is scheduled to come online next year. SK Hynix and Samsung are also building substantial new HBM factory capacity in South Korea.

The shortage’s effects already extend beyond AI servers. Higher memory costs have contributed to increased prices for consumer electronics, including Apple’s iPads and MacBooks, according to CNBC. Micron’s earnings therefore capture two sides of the same supply squeeze: a sharp increase in revenue and profit for a supplier, and more expensive hardware for consumers. The planned factories are a response, not capacity available today.