Micron Technology has delivered what can only be described as a landmark quarterly report, with fiscal third-quarter revenue hitting nearly $42bn — a fourfold increase from the $9bn reported in the same period a year earlier. The results, announced on Tuesday, far exceeded Wall Street expectations and cemented Micron’s position as a primary beneficiary of the artificial intelligence boom, particularly in the high-bandwidth memory (HBM) segment.
Adjusted earnings per share came in above $25, compared with analyst estimates of roughly $21. Under generally accepted accounting principles, net income exceeded $28bn, or nearly $25 per share, a dramatic leap from under $2bn in the year-ago quarter. Gross margins surged to above 81 percent, up from 69 percent in the prior quarter and 27 percent a year earlier, reflecting the premium pricing power Micron enjoys in the AI memory market.
The HBM Revolution
The headline number — near $42bn in revenue — was largely fueled by insatiable demand for high-bandwidth memory, the advanced stacked DRAM chips that sit adjacent to graphics processing units (GPUs) inside AI accelerators built by companies like Nvidia and Google. HBM has become the primary bottleneck in AI infrastructure expansion, as the speed at which data can be moved between memory and compute cores directly impacts the performance of large language models and other generative AI workloads. Micron is one of only three companies globally capable of manufacturing HBM, alongside SK Hynix and Samsung, giving it significant leverage in a market where supply is severely constrained.
CEO Sanjay Mehrotra revealed that Micron can currently satisfy only between half and two-thirds of customer demand for HBM. The company’s entire 2026 supply of HBM has already been sold out under multi-year contracts, and it has collected $22bn in customer cash deposits — effectively prepayments from hyperscalers desperate to lock in allocation. These deposits not only de-risk Micron’s massive capacity expansion but also provide a clear signal of the structural nature of AI memory demand.
Next-Generation Products and Guidance
Micron’s next-generation HBM4 chips are ramping at twice the speed of the previous HBM3E generation. Revenue from HBM4 has already surpassed $1bn, underscoring the rapid adoption of the technology in the latest accelerators from Nvidia and Google. As AI inference workloads scale, memory bandwidth — rather than raw compute — increasingly determines throughput, making HBM a critical component of the AI value chain.
The forward guidance was equally aggressive. Micron projected fiscal fourth-quarter revenue of approximately $50bn (plus or minus $1bn), against analyst estimates of roughly $44bn and a year-ago figure of just over $11bn. To meet this demand, the company raised its full-year capital expenditure forecast to more than $25bn, up from a previous target of $20bn. These investments will expand production capacity for HBM and advanced DRAM, with new fabrication facilities coming online in the United States, Japan, and Taiwan.
Market Valuation and Shareholder Returns
Micron’s market capitalization crossed $1 trillion on May 26, making it the latest memory chipmaker to reach that milestone as the AI-driven memory supercycle reshapes valuations across the semiconductor industry. The stock’s roughly 700% gain over the past year reflects a market that is pricing memory not as a cyclical commodity but as structural AI infrastructure. The company plans to return 100% of excess free cash flow to shareholders, a commitment enabled by the cash deposit program that reduces the capital risk of its expansion.
Competitive Landscape and Risks
Despite these impressive numbers, there are important caveats. Micron remains the smallest of the three HBM suppliers, trailing SK Hynix and Samsung in total HBM market share, and its allocation in Nvidia’s HBM4 orders is the thinnest of the trio. The broader memory market is also shifting: Chinese manufacturers, such as CXMT, are expanding aggressively into consumer DRAM segments that the Big Three have deprioritized in favor of AI chips. This could create pricing pressure in non-HBM markets over time.
Memory pricing is inherently cyclical, and the current supercycle depends on hyperscaler capital expenditure continuing at its current pace. If AI infrastructure spending slows or HBM supply catches up with demand — which could happen as SK Hynix and Samsung ramp their own HBM3E and HBM4 production — the margins that Micron reported this quarter could compress rapidly. The 81% gross margin is historically extraordinary for a memory company and reflects shortage economics as much as product superiority.
Historical Context and the Memory Cycle
To understand the magnitude of Micron’s achievement, it is helpful to look back at the company’s recent history. Just two years ago, Micron was reporting losses as the memory market endured a severe downturn triggered by oversupply and weak demand from PCs and smartphones. The shift to AI has transformed the company’s fortunes, but it also introduces new dependencies. The entire memory industry is now tightly coupled to the investment cycles of a handful of hyperscale cloud providers — Amazon, Microsoft, Google, and Meta — whose combined AI spending is expected to exceed $200bn in 2025.
Analysts have noted that the total addressable market for HBM is projected to grow at a compound annual rate of roughly 40% through 2028, rising from approximately $35bn in 2025 to around $100bn. Micron’s success will depend on its ability to maintain technological parity with its Korean rivals while managing the enormous capital outlays required to build new fabs. The company is also investing in advanced packaging capabilities, which are essential for stacking memory layers in HBM.
From a technological standpoint, HBM4 represents a significant leap forward. It increases the number of stacked DRAM dies from 12 to 16, doubling memory capacity per module, and uses a new interface standard that boosts data transfer rates. These improvements are critical for next-generation AI accelerators that require hundreds of gigabytes of memory bandwidth per second. Micron’s ability to ramp HBM4 production faster than HBM3E indicates a growing manufacturing maturity that could help it gain market share.
Customer concentration is another factor to consider. While the prepayments from hyperscalers provide visibility, they also mean that Micron’s revenue is heavily tied to the spending plans of a small number of customers. If one of these customers decides to build its own HBM capabilities or switches suppliers, the impact on Micron’s top line would be significant. For now, though, the demand for HBM far outstrips supply, creating a seller’s market that benefits all three vendors.
The broader semiconductor industry is watching Micron closely. The company’s results are seen as a bellwether for the AI infrastructure buildout, and its guidance suggests that the boom has further to run. However, history suggests that memory cycles are unforgiving. The current supercycle has been driven by an unprecedented confluence of factors — the rise of generative AI, the limitations of Moore’s Law in compute scaling, and the need for specialized memory architectures. Whether this supercycle can avoid the boom-bust pattern of previous memory cycles remains an open question.
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 $6bn are not normal results for any company, let alone one that was losing money 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.