KEY POINTS
- Samsung reported Q1 2026 operating profit of ₩57.2 trillion ($31.8 billion), an eightfold increase year-over-year, with the semiconductor division accounting for 94% of total profit.
- HBM4 mass production began in February 2026 and Samsung expects HBM sales to more than triple this year, but both Samsung and SK Hynix are warning that AI-driven memory shortages could persist through 2027.
- Traders should monitor conventional DRAM pricing alongside HBM, as Samsung disclosed that standard server DRAM is now delivering higher margins than HBM in some product lines — a sign the AI memory shortage is spreading beyond specialty chips.
Samsung Electronics reported the most profitable quarter in its 57-year history, posting operating profit of ₩57.2 trillion — roughly $31.8 billion — on revenue of ₩133.9 trillion for the first quarter of 2026. The semiconductor division generated ₩53.7 trillion in operating profit, a 48-fold increase from the same period last year and accounting for 94% of the entire company's earnings. The numbers are so lopsided that Samsung is functionally a memory chip company with a consumer electronics side business.
The driver is straightforward: every AI data center in the world needs memory, and there is not enough of it. High-bandwidth memory, the specialized DRAM stacks that sit directly on AI accelerators from Nvidia, AMD, and Google, has been the focus of investor attention for two years. Samsung became the first manufacturer to begin mass production of HBM4 chips in February, shipping to unnamed hyperscaler customers. The company expects HBM sales to more than triple in 2026.
The Shortage Spreads Beyond HBM
The more surprising revelation from Samsung's earnings call was that conventional server DRAM — the standard memory used in every data center rack, not just AI accelerators — is now more profitable per gigabyte than HBM in certain product lines. That is a telling signal. It means the AI infrastructure build-out has tightened supply across the entire memory stack, not just the specialty chips at the top. Server DRAM contract prices rose 18% quarter-over-quarter in Q1, the steepest increase since the 2021 cycle.
Samsung and rival SK Hynix both warned during their respective earnings calls that AI-driven memory shortages could persist through 2027 and beyond. Hyperscaler customers are now reserving memory supply two to three years in advance, locking in capacity for data centers that have not yet broken ground. That kind of forward commitment is unusual and suggests the largest AI buyers see compute demand growing faster than the semiconductor industry can add fabrication capacity.
Samsung vs. SK Hynix: The HBM Race
Samsung's HBM4 ramp represents a strategic catch-up play. SK Hynix seized the early lead in HBM3 and HBM3E, securing pole position as Nvidia's primary memory supplier. Samsung lost ground during that cycle due to yield issues that delayed qualification with Nvidia. The HBM4 generation is Samsung's opportunity to reset the competitive dynamic.
Early indications are mixed. Samsung has the manufacturing lead — it was first to mass production — but SK Hynix retains the qualification advantage with Nvidia's Rubin platform. Both companies are running at effectively full capacity, and the market is large enough that share dynamics matter less than total available supply. Deloitte estimates the AI chip market will reach $500 billion in 2026, and memory represents roughly 25% of the bill of materials for a typical AI server rack.
What to Watch Next
The memory cycle is entering a phase where pricing power shifts from buyers to sellers. Contract prices for HBM are fixed for full-year 2026, but spot premiums on conventional server DRAM are widening, and Samsung has signaled it will prioritize high-margin AI memory over commodity products if capacity remains constrained. For traders, the actionable level is Samsung's guidance for Q2 memory division margins — if the company signals further tightening, the entire memory supply chain, from Micron to equipment makers like Applied Materials and Lam Research, reprices higher. The next major data point is Micron's fiscal Q3 report in late June, which will show whether the DRAM shortage Samsung flagged is hitting U.S. memory producers with equal force.
