Citigroup has presented a very optimistic outlook for SK hynix, significantly raising its target price to 1.7 million won and forecasting a record-breaking operating profit of 39.1 trillion won for the first quarter of 2026.
This bullish forecast is rooted in the powerful 'AI-memory supercycle' narrative. The explosive growth of artificial intelligence is driving unprecedented demand for high-performance memory chips, which are essential for training and running complex AI models. This has led to a sharp increase in the prices of key products like HBM (High Bandwidth Memory) and DDR5 server DRAM, directly boosting SK hynix's potential earnings.
There are a few key factors behind this prediction. First, the supply of these advanced chips can't keep up with the surging demand. Major tech companies, known as Cloud Service Providers (CSPs), are rushing to secure their supply, leading to favorable pricing for memory makers. Second, there's a structural shift in how contracts are made. Instead of short-term deals, customers are now signing Long-Term Agreements (LTAs), often with price floors. This provides SK hynix with much better revenue visibility and reduces the historical boom-and-bust cyclicality of the memory market.
However, it's important to consider the potential risks. Citi's forecast is notably more aggressive than the market consensus. While SK hynix is currently a leader in the HBM market, competitors like Samsung and Micron are investing heavily to catch up. Increased competition could lead to pressure on prices and market share in the latter half of the year. The stock is also trading at a relatively high valuation, meaning it already reflects high expectations, so any failure to meet these lofty profit goals could lead to volatility.
In essence, Citi's call bets on SK hynix capitalizing on a powerful, AI-driven demand wave and a more stable market structure. The upcoming earnings report will be a critical test of whether this highly optimistic scenario is unfolding as predicted.
- Glossary
- HBM (High Bandwidth Memory): A type of high-performance memory chip that stacks multiple DRAM chips vertically to provide much faster data transfer speeds, crucial for AI accelerators like GPUs.
- LTA (Long-Term Agreement): A contract between a supplier and a customer that extends over a long period, often multiple years, helping to stabilize supply and pricing.
- DDR5: The fifth generation of Double Data Rate Synchronous Dynamic Random-Access Memory. It is the latest standard for main memory used in servers and PCs, offering higher speed and efficiency.