Micron, SK Hynix gain memory pricing power
- Micron and SK hynix are using the AI memory shortage to raise prices and lock in supply, after both posted record 2026 results. - Micron says its entire 2026 HBM output is already contracted, while SK hynix just reported a 72% operating margin on ₩52.6 trillion revenue. - The backdrop is industrywide scarcity — chip sales hit $298.5 billion in Q1, and suppliers are steering capacity toward HBM and servers.
Memory is having a power moment. Not glamorous memory — the chips that sit next to AI accelerators and feed them data fast enough to keep those expensive GPUs busy. That market tightened hard over the past year, and now the companies that matter most in it, especially Micron and SK hynix, are starting to look less like commodity suppliers and more like gatekeepers. The news isn’t one headline so much as a cluster of fresh proof points from the past few weeks: record results, sold-out HBM supply, and another leg up in contract pricing. (news.skhynix.com) ### What changed? The clearest shift is that the big memory makers are no longer talking like companies trapped in a boom-bust cycle. Micron said in its fiscal Q1 2026 call that it had already completed agreements on price and volume for its entire calendar 2026 HBM supply, including HBM4. Then in fiscal Q2 2026 it posted record revenue of $23.86 billion(news.skhynix.com)owed with its own huge quarter on April 23, reporting record revenue of ₩52.5763 trillion and record operating profit of ₩37.6103 trillion. (investors.micron.com) ### Why does HBM matter so much? HBM — high-bandwidth memory — is the premium memory stacked next to AI GPUs. It is not interchangeable with generic PC or phone DRAM. It needs advanced packaging, tight qualification with GPU platforms, and years of process know-how. That makes the market much narrower than “memory” sounds. (investors.micron.com)a very small club. Micron now says it sees the HBM market growing from about $35 billion in 2025 to around $100 billion in 2028. (investors.micron.com) ### So where does pricing power come from? Basically, from scarcity plus concentration. TrendForce said on March 31 that DRAM makers were reallocating capacity toward HBM and server products in 2Q26, while cloud customers were using long-term agreements to secure supply. It projected conventional DRAM contract prices would(investors.micron.com)demand.” It is demand running into constrained supply, with suppliers choosing where capacity goes. (trendforce.com) ### Why can’t customers just wait it out? Because AI infrastructure is now a race. If a cloud company delays server deployment, it is not just postponing a purchase — it is giving up training capacity, inference capacity, and product rollout time. Memory becomes the choke point. Micron’s CEO put it bluntly in Q1: memory has become a strategic (trendforce.com)n buyers could squeeze vendors and wait for the cycle to turn. (investors.micron.com) ### Is this just an HBM story? No — that is the catch. HBM is the spear tip, but it pulls the rest of the market with it. When suppliers divert wafers and engineering attention toward HBM and server DRAM, less supply is available for other categories. That is why TrendForce is also talking about catch-up pricing in convent(investors.micron.com) products deserve the capacity first. (trendforce.com) ### How big is the broader wave? Big enough that the whole chip industry is feeling it. SIA said on May 4 that global semiconductor sales reached $298.5 billion in the first quarter of 2026, up 25% from Q4 2025, with March alone at $99.5 billion. Not all of that is memory, obviously. But memory and AI logic are at the center of the current ups(trendforce.com)the economics. (semiconductors.org) ### What’s the bottom line? Micron and SK hynix are not just benefiting from AI demand. They are gaining leverage over it. When supply is sold out, pricing is rising, and only a few vendors can deliver the right memory at scale, buyers stop setting terms. Suppliers do. (news.skhynix.com)
