Major-Company-Funded Drilling at This Sub-$1 Copper-Gold Explorer

There is a specific number that reframes everything you think you know about Micron’s HBM ambitions right now. Industry sources have reported that Micron plans to add up to 60,000 wafers per month in HBM production capacity by end of 2026, against last year’s output of roughly 40,000 to 50,000 wafers per month. That means the net new capacity coming online this year exceeds what Micron was producing in total just twelve months ago.

The urgency is obvious. Industry reporting has put Samsung’s HBM processing at around 150,000 DRAM wafers per month, and SK hynix is often discussed in the same general range, multiples of Micron’s recent run-rate. Unless Micron closes that gap, it remains a permanent third player. If it hits its targets, the production gap with the Korean giants could narrow materially. That is not dominance, but it is optionality.

What makes this expansion worth a trader’s attention is not the volume, it’s the product mix. Micron is targeting roughly 100,000 wafers a month while sharply increasing 12-high HBM4 output for NVIDIA’s Vera Rubin platform. Micron says it began volume shipment of its HBM4 36GB 12-high in the first quarter of calendar 2026, designed for NVIDIA Vera Rubin. The revenue attached to each wafer going forward is meaningfully higher than what the prior HBM3E mix delivered.

The financial logic is already visible in the numbers. In fiscal 2025, Micron’s revenue grew from $25.1 billion to $37.4 billion, while the Cloud Memory Business Unit alone grew 257% to $13.5 billion as the company redirected DRAM supply toward HBM and higher-value data center products. Now the question traders need to ask is whether that mix shift continues to protect margins as capacity comes online, or whether more supply eventually compresses the premium.

The structural argument for tightness holds for now. Every AI accelerator generation requires more HBM to avoid idle GPU cores waiting on data. For buyers, the signal is clear: memory bandwidth continues to limit AI system performance, and when bandwidth is insufficient, expensive GPU cores sit idle waiting for data. That constraint could support pricing power into 2027, but it remains a moving target as all suppliers add capacity.

The real trade here is about market share trajectory, not just revenue. Research tracking HBM share by revenue showed Micron at about 21% in Q1 2026, but closer to 18% in Q2 2026. Micron has publicly highlighted power-efficiency gains for its HBM4 versus its own HBM3E, but the widely circulated “30% less power” claim applies to Micron’s HBM3E versus competing offerings, not to its HBM4. Doubling capacity without sacrificing yield is where execution risk lives. Competitive qualification progress from Samsung and SK hynix will show how durable Micron’s current position inside AI infrastructure actually is.

Disciplined traders don’t buy the capacity announcement. They buy the yield data. Micron’s ability to ramp 12-high HBM4 to 100,000 wafers a month while maintaining the power efficiency edge is the real test. Watch the next earnings call for any guidance revision on HBM margins. That number will tell you far more than the wafer count ever will.