Key Facts
- Revenue: $41.8 billion in Q3 FY2025, up from $9.1 billion a year ago (quadrupled).
- Adjusted EPS: $25.30, beating analyst estimates of $21.40.
- Gross margin: 81.2%, up from 69% in Q2 and 27% a year ago.
- HBM (high-bandwidth memory) demand is the primary driver; Micron can only fulfill 50-67% of customer demand.
- Entire 2026 HBM supply is sold out; $22 billion in customer prepayments received.
- HBM4 revenue has already exceeded $1 billion.
- Q4 guidance: Revenue of $50 billion (±$1 billion), vs. analyst consensus of $44 billion.
- FY2025 capex raised to >$25 billion from $20 billion to expand HBM and advanced DRAM capacity.
- Market cap crossed $1 trillion on May 26.
Micron Technology, the Idaho-based memory chipmaker, delivered a fiscal third-quarter earnings report on Tuesday that shattered all expectations and confirmed that the AI-driven memory supercycle is accelerating at an unprecedented pace. The company reported revenue of $41.8 billion for the quarter ended May 29, 2025, more than quadrupling the $9.1 billion it recorded in the same period last year. That figure also handily beat the consensus Wall Street estimate of roughly $36 billion, marking the third consecutive quarter in which Micron has posted upside surprises of more than 15%.
The headline revenue number was powered almost entirely by surging demand for high-bandwidth memory (HBM), the specialized stacked DRAM chips that sit alongside graphics processing units (GPUs) inside artificial intelligence accelerators built by Nvidia, Google, and other hyperscalers. HBM has become the binding constraint on AI infrastructure expansion, as the memory bandwidth it provides directly determines how fast models can be trained and how efficiently inference can be deployed. Micron is one of only three companies in the world capable of producing these chips, alongside South Korea’s SK Hynix and Samsung Electronics.
Adjusted earnings per share came in at $25.30, well above the analyst consensus of $21.40. On a GAAP basis, net income exceeded $28.3 billion, or $24.95 per diluted share, compared with just under $2 billion in the year-ago quarter. Gross margins skyrocketed to 81.2%, up from 69% in the prior quarter and 27% a year earlier. For a memory company, which historically operates in a cyclical commodity market where gross margins often fluctuate between 20% and 50%, an 81% margin is extraordinary and reflects acute shortage economics rather than normal pricing dynamics.
CEO Sanjay Mehrotra, during the earnings call, revealed that Micron can currently fulfill only between 50% and 67% of customer demand for HBM. The company’s entire 2026 production capacity for HBM is already sold out under multi-year contracts, and it has collected $22 billion in customer cash deposits — essentially prepayments from hyperscalers desperate to lock in supply. “We are operating in an environment where demand far outstrips supply, and we expect this imbalance to persist through at least 2027,” Mehrotra said.
Micron’s next-generation HBM4 chips are ramping at twice the speed of the previous HBM3E generation. HBM4 revenue has already exceeded $1 billion in the current quarter, a milestone that underscores the rapid adoption of these chips in the latest accelerators from Nvidia (the Blackwell Ultra series) and Google (the TPU v6). The technology is critical because memory bandwidth, rather than raw compute, increasingly determines inference throughput as models grow larger and require more data to be moved quickly between memory and processing units.
The forward guidance was equally aggressive. For the fiscal fourth quarter ending August 28, 2025, Micron projected revenue of approximately $50 billion, plus or minus $1 billion, against analyst estimates of roughly $44 billion and a year-ago figure of $11.2 billion. The company also raised its full-year capital expenditure forecast to more than $25 billion, up from a previous target of $20 billion, to expand production capacity for HBM and advanced DRAM nodes. This capex increase is necessary to add cleanroom space at its facilities in Boise, Idaho; Taichung, Taiwan; and Hiroshima, Japan.
Micron’s market capitalization crossed the $1 trillion threshold on May 26, 2025, making it the latest memory chipmaker to reach that milestone as the AI-driven memory supercycle reshapes valuations across the semiconductor industry. The stock has gained roughly 700% over the past 12 months, reflecting a market that is now pricing memory not as a cyclical commodity but as a structural component of AI infrastructure — alongside GPUs and networking equipment.
The company provided long-term growth expectations for the HBM total addressable market (TAM), projecting a compound annual growth rate (CAGR) of approximately 40% through 2028. The TAM is expected to rise from roughly $35 billion in 2025 to around $100 billion by 2028. Micron plans to return 100% of excess free cash flow to shareholders in the form of dividends and share buybacks, a commitment enabled by the $22 billion cash deposit program, which significantly reduces the capital risk associated with its massive expansion plans.
However, there are caveats worth examining. Despite its impressive results, Micron remains the smallest of the three HBM suppliers, with an estimated market share of around 15-20%, compared with SK Hynix’s 45-50% and Samsung’s 30-35%. Its share of Nvidia’s HBM4 allocations is the thinnest of the trio, as Nvidia has historically favored SK Hynix for early production volumes. Micron’s advantage lies in its leading-edge manufacturing processes and its ability to deliver higher memory density per stack, but it still lags in overall scale.
The broader memory market is also undergoing a structural shift. Chinese manufacturers, particularly ChangXin Memory Technologies (CXMT), are expanding aggressively into consumer DRAM segments — such as DDR4 and LPDDR5 — that the Big Three have largely deprioritized in order to focus on AI-related HBM and DDR5. This could lead to margin pressure in the commodity memory space, though Micron’s exposure to those segments is decreasing as it pivots toward HBM and high-value products.
Memory pricing is inherently cyclical, and the current supercycle depends on hyperscaler capital expenditure continuing at its current pace. Companies like Microsoft, Amazon, Alphabet (Google), and Meta have collectively announced over $300 billion in AI infrastructure spending for 2025, with a significant portion allocated to memory. If that spending slows — due to an economic downturn, regulatory changes, or a shift in AI model efficiency — the demand imbalance could reverse quickly. The 81% gross margin reported this quarter is historically extraordinary and reflects shortage economics as much as product superiority. In a balanced market, margins would likely compress to 50-60%, still high but not unprecedented.
For now, the numbers speak for themselves. Revenue that quadruples in a year, margins that triple, and a guidance print that exceeds estimates by more than $6 billion are not normal results for any company, let alone one that was losing money just two years ago. Micron’s earnings confirm that the AI memory shortage is intensifying, not easing, and that the companies making the chips inside AI accelerators are capturing value at a rate the market is still recalibrating to price. The HBM supply chain remains the bottleneck for AI infrastructure, and Micron, despite its smaller size, is now an indispensable part of that chain.