SK Hynix Posts Record Q3 Earnings as AI Chip Demand Drives Revenue Past 100 Trillion Won

South Korean memory chip giant SK Hynix has delivered a blockbuster set of financial results for the third quarter, shattering company records on the back of insatiable global demand for artificial intelligence hardware. According to the company’s official earnings release published on Wednesday, SK Hynix reported a staggering quarterly revenue of 104.6 trillion won—equivalent to nearly $78 billion—alongside an operating profit of 83.3 trillion won, or roughly $65.5 billion.

This financial milestone marks the first time in the company’s history that its quarterly sales have crossed the monumental threshold of 100 trillion won, translating to approximately $74.5 billion. The sheer scale of the growth underscores how deeply entrenched memory manufacturers have become in the global AI infrastructure boom, as tech companies worldwide race to secure the specialized hardware required to train and deploy advanced machine learning models.

The year-over-year expansion was historic by any metric. SK Hynix’s revenue skyrocketed by 328% compared to the same period in the previous year, while its operating profit surged by an eye-watering 632%. This performance yielded an astonishing operating margin of 80%, meaning the South Korean chipmaker retained approximately 80 won of profit from its core operations for every 100 won of sales generated. Wall Street and regional analysts surveyed by Bloomberg had anticipated strong numbers, but the company’s operating profit comfortably outpaced expectations, which had centered around 77.5 trillion won.

Despite the record-breaking top-line and operating figures, SK Hynix’s net income told a slightly more nuanced story. Net income for the quarter came in at 39.8 trillion won. While this figure represents a robust 216% increase from the same period a year ago, it also reflects a 58% decrease compared to the immediately preceding second quarter.

Addressing the gap during the earnings call, Chief Financial Officer Kim Woo-hyun explained that the divergence was driven primarily by net non-operating expenses totaling 32.6 trillion won. These charges, which occurred outside of core chip sales, were heavily impacted by an 11.6 trillion won currency loss alongside a 22.9 trillion won valuation loss on external investments.

Looking ahead, SK Hynix executives struck a remarkably bullish tone regarding the structural supply-demand imbalance in the global memory market. The company informed investors that demand for memory components is projected to comfortably outrun available supply through at least 2030. Furthermore, the manufacturer revealed that its production volumes for standard DRAM and high-bandwidth memory (HBM) are effectively sold out through 2029.

DRAM serves as the foundational standard memory utilized across servers, smartphones, and personal computers. In contrast, high-bandwidth memory, or HBM, represents the specialized stacked memory architecture that sits in tight physical proximity to AI processors, feeding them massive streams of data with ultra-low latency.

"Given the complexities involved in implementing leading-edge processes and the lead times required to construct new production facilities," Kim stated during the earnings briefing, "a situation where supply cannot keep up with demand is expected to persist for some time, a constraint projected to continue through 2030."

The intensity of this demand has manifested in subtle operational shifts among major tech buyers. SK Hynix noted that several key customers have recently moved to adjust their HBM specifications. The company interprets this as clear evidence that demand is vastly overwhelming available supply, forcing clients to optimize their hardware configurations to build more AI accelerators out of a strictly limited quantity of physical chips.

"Demand remains so strong that customers’ volume requirement is more than double that of what we can supply today," leadership emphasized during the call.

To lock in revenue and secure long-term stability against market volatility, SK Hynix has moved aggressively to establish multi-year supply pacts. The firm has signed long-term agreements—contracts designed to reserve chip production volumes years in advance—with 14 major global customers, including leading cloud service providers. These binding agreements already cover more than 50% of the company’s projected production capacity for both 2027 and 2028. Typically spanning five-year terms, the contracts incorporate mechanisms allowing prices to fluctuate within an agreed-upon range while incorporating vital risk-mitigation safeguards such as upfront financial deposits.

Capital expenditure is likewise scaling up to meet this generational demand curve. SK Hynix expects its overall investment to climb significantly in 2027 as the initial phase of its massive Yongin semiconductor complex transitions into full-scale commercial operation.

The earnings call also touched upon broader macroeconomic and industry anxieties surrounding the sustainability of the current AI spending cycle. Opening the Q&A session, Citigroup analyst Peter Lee pressed management on what supports their confidence in continuous AI infrastructure investment, particularly at a time when direct monetization by enterprise users appears slow. This exact doubt was raised just eight days prior by Citadel founder Ken Griffin, who publicly asserted that investors are "almost certainly" navigating an artificial intelligence bubble.

Responding to the skepticism, CEO Kwak Noh-Jung pointed to concrete financial indicators from the company’s primary enterprise clients. Major cloud computing providers recently reported year-over-year cloud revenue growth ranging anywhere from 30% to 80%, alongside cloud order backlogs that have expanded between two-fold and five-fold.

Kwak further argued that the structural foundation of AI spending is maturing and diversifying. Whereas initial capital expenditures were heavily concentrated among a handful of massive hyperscale tech giants, spending is now broadening out to include non-tech enterprises, government-backed sovereign AI initiatives, and on-premise setups where individual corporations build out dedicated AI capabilities on their own private servers. This broader base of adoption means that a temporary slowdown by any single major buyer would pose a significantly smaller systemic threat to the broader supply chain.

At the same time, however, broader financial conditions for the wider AI ecosystem are undeniably tightening. Capital access has begun to constrict for smaller infrastructure players and specialized neocloud providers. Highlighting these headwinds, Elon Musk noted on October 25 that debt markets have "almost entirely shut off access to money" for firms that lack substantial non-AI revenue streams to buffer their balance sheets.

For SK Hynix, any potential softening in traditional hardware sectors is expected to be more than neutralized by relentless enterprise demand. While the company anticipates weaker demand from traditional personal computer and smartphone manufacturers—largely driven by higher prevailing memory prices and persistent supply tightness—the runaway momentum of the AI infrastructure market is projected to easily offset any downstream declines.