Micron shatters records: FY Q3 $41.5B revenue, 84.6% gross margin, $50B Q4 guide at 86%
Micron Technology reported fiscal Q3 2026 results on June 24 that shattered all company records: revenue of $41.46 billion (up 346% YoY), non-GAAP EPS of $25.11, and gross margin of 84.6%—a 46.9 percentage-point expansion from 37.7% a year prior. Operating cash flow reached a quarterly record of $25.39 billion, with adjusted free cash flow of $18.3 billion. The massive outperformance was driven by high-bandwidth memory (HBM) demand from AI accelerator makers, sold out through all of 2026 under multi-year fixed-price Strategic Customer Agreements.
Management issued forward guidance of $50 billion revenue for fiscal Q4 2026 at 86% gross margins—a single quarter larger than Micron's entire fiscal 2024 revenue. CEO Sanjay Mehrotra stated that memory supply will remain "tight beyond calendar 2027" due to 18-24 month lead times for new fabrication capacity. Goldman Sachs estimates the 2026 DRAM supply-demand gap at 4.9%, the tightest in 15 years. HBM capacity ramps have accelerated: Micron's HBM4 production ramps at twice the pace of HBM3E.
What distinguishes this cycle from prior memory booms: Micron's HBM revenue is now locked into multi-year contracts at fixed prices with hyperscalers (AWS, Google, Microsoft, Meta), eliminating spot-price volatility and shifting Micron's business model from cyclical commodity to quasi-recurring revenue. The company raised capex guidance to over $25 billion for fiscal 2026 (up 25% from prior guidance) to accelerate new fab throughput, but even with all three major memory makers (Micron, SK Hynix, Samsung) spending $58+ billion combined in 2026, new fab output will not arrive until fiscal 2028.
For practitioners evaluating AI infrastructure costs: Micron's sold-out HBM capacity and 84.6% margins underscore a structural supply constraint that will persist through 2027, anchoring GPU and accelerator costs at elevated levels. The HBM market is projected to reach $100 billion by 2028, up from $35 billion in 2025. The risk: if AI model efficiency improves faster than compute demand grows, or capex slows, new supply arriving in 2027-2028 could compress pricing simultaneously. Market rewarded the beat with a +16% post-earnings jump; stock trades at forward P/E below 12x on 2027 earnings, despite $1T+ market cap.