(LEAD) SK hynix logs record profit in Q3 on AI chip supercycle

SK hynix reported its sixth straight quarter of record revenue and operating profit. For July through September, revenue reached 104.61 trillion won, while operating profit climbed to 83.33 trillion won. Net profit also set a record at 39.77 trillion won.

The company linked these results to rising prices for key products and stronger demand for artificial-intelligence memory. HBM was especially important because it is a high-value-added memory product used in demanding AI workloads. Other memory products also contributed to the company’s performance.

The results show how strongly the current AI chip cycle is benefiting SK hynix. The company said it will expand production capacity and strengthen product competitiveness. It also plans to return an additional 60 trillion won to shareholders after completing a 40 trillion-won program.

SK hynix’s year-over-year increases were exceptionally large across all three major measures. Revenue surged 327.9 percent to 104.61 trillion won. Operating profit rose even faster, soaring 557.2 percent to 83.33 trillion won. Net profit climbed 215.7 percent to 39.77 trillion won.

The different growth rates matter. Profit increased faster than revenue because the company benefited from higher prices for key memory products and strong demand for AI memory. When selling prices and product mix improve, more of each revenue unit can remain after production costs.

These figures reflect a powerful memory-chip upcycle. SK hynix said it had achieved record quarterly revenue and operating profit for six consecutive quarters since the second quarter of last year. It plans to expand capacity and strengthen products while the favorable market continues.

High-bandwidth memory, or HBM, is a type of memory built to provide very high data-transfer capacity. In simple terms, it helps a processor receive and supply data faster than conventional memory designs. The article identifies HBM as a key product behind SK hynix’s record results.

AI systems repeatedly process huge collections of data and perform many calculations at once. Their performance can be limited if the processor must wait for data. HBM reduces that waiting by supplying data at high speed, helping AI processors spend more time calculating rather than idle.

That combination of technical value and strong AI demand can support higher prices. SK hynix said prices of key products rose as demand for AI memory grew. It described HBM as a high-value-added product and plans to expand production capacity while improving competitiveness.

Revenue measures sales, while operating profit measures what remains after operating costs. A memory-chip company often has large fixed costs for factories, equipment, and research. Once those costs are covered, higher selling prices can add much more to profit than to revenue.

For example, if chip prices rise while production costs change little, each unit brings in a larger margin. Selling more chips spreads fixed factory costs across a greater output as well. The combined effect is operating leverage: profit grows faster than sales during a strong upcycle.

SK hynix’s figures illustrate this pattern. Revenue rose 327.9 percent year over year, but operating profit soared 557.2 percent. The company attributed the favorable conditions to rising prices and demand for AI memory, including HBM. The reverse can happen when prices weaken.

Memory chips are cyclical products. When demand is strong, companies may add factories and equipment to capture growth. If many suppliers expand at once, total output can eventually exceed what customers need. Excess supply usually puts downward pressure on prices.

The mechanism is straightforward. Lower chip prices reduce revenue per unit, while factories and equipment still create large costs. Inventories may rise, production lines may run below efficient levels, and operating profit can fall sharply. Companies could then delay investment or compete more aggressively for orders.

The article says SK hynix will accelerate production-capacity expansion because AI-memory demand is favorable. It does not describe an oversupply problem or predict one. The broader implication is that capacity must match durable demand, especially because the company’s recent record profits depend partly on higher prices.

SK hynix announced an additional 60 trillion won shareholder-return plan after completing a 40 trillion-won program the previous week. The move directs more value to investors while the company is reporting record earnings. The article does not state that the return plan directly caused the Solidigm decision.

Separately, SK hynix suspended plans to list its U.S. subsidiary, Solidigm, citing market conditions. Reports had suggested a U.S. initial public offering could come as early as next year, with a possible valuation of up to US$150 billion. The proposed listing therefore would have been a major financial event.

The decision also followed strong shareholder opposition and regulatory scrutiny. Critics were concerned that a separate listing could undermine minority shareholders’ interests. With the listing suspended, SK hynix’s announced shareholder-return program provides another way to return capital, while its shares had faced sharp downward pressure.

Processors are the components that perform calculations and execute instructions. Memory holds the data and instructions that processors need. They work together, but they serve different roles: the processor does the active computation, while memory supplies information and keeps results available.

AI applications move and process enormous amounts of data. If memory cannot deliver information quickly enough, the processor may sit idle between calculations. Faster data movement helps keep the processor busy, improving throughput and making demanding AI workloads more efficient. HBM is valuable because it is designed for especially high data-transfer capacity.

This connection helps explain SK hynix’s results. The company said demand for AI memory drove higher prices for key products. It specifically highlighted HBM and other memory products as major contributors to record revenue and operating profit, and it plans to expand capacity and strengthen product competitiveness.