SK Hynix, a critical player in the global memory chip market and a primary supplier to NVIDIA, has shattered previous earnings records, propelled by an insatiable demand for advanced AI chips. The company is now offering a remarkably bullish outlook, predicting a prolonged “super cycle” in the memory market that could extend well into 2027 and 2028. This unprecedented surge, fueled by massive global investments in artificial intelligence infrastructure, signals a transformative period for the tech industry and sets the stage for significant shifts in chip availability and pricing for developers and data centers worldwide.
The semiconductor world is abuzz following SK Hynix’s announcement of record-breaking quarterly earnings. The South Korean memory-chip giant, a vital component supplier for AI leader NVIDIA, reported stellar third-quarter performance, attributing its success to robust demand for its high-bandwidth memory (HBM) products and rising prices for DRAM chips. This financial triumph underscores the relentless pace of AI development and the critical role advanced memory plays in powering it.
For the three months ended September, SK Hynix reported a net profit that more than doubled year-over-year, reaching an impressive 12.598 trillion won ($8.80 billion). This figure not only surpassed a previous high set earlier in the year but also significantly beat market expectations, which had estimated a bottom line of 8.491 trillion won, according to a FactSet consensus estimate. The company also achieved record revenue, climbing 39% to 24.449 trillion won, and saw operating profit exceed 10 trillion won for the first time, jumping 62% to 11.383 trillion won.
The AI Fuel: HBM and DDR5 at the Core
The bedrock of SK Hynix’s success lies in its advanced AI chip portfolio, particularly its High Bandwidth Memory (HBM) products. The company is a main supplier of these specialized memory modules to NVIDIA, the leading force in AI accelerators. Executives highlighted brisk shipments of higher-end HBM3e products, including 12-layer variants that currently dominate the premium segment of the AI-chip market, as a key driver of earnings.
Demand isn’t limited to HBM alone. Higher prices for DRAM chips, especially high-capacity Double Data Rate 5 (DDR5) products used in data servers for AI training and mobile devices, also contributed significantly. The surging demand for enterprise Solid State Drives (SSDs) further boosted the company’s NAND chip business.
The future looks equally bright, or perhaps even tighter, for supply. SK Hynix executives revealed that its HBM and other memory-chip products are almost entirely sold out for 2026 due to the overwhelming AI-driven demand. Ki Tae Kim, head of the HBM sales and marketing department, explicitly stated that “the company’s HBM supply will remain tight in 2027.”
Looking ahead, SK Hynix has already completed discussions with key customers regarding 2026 HBM supply and expects to begin shipments of next-generation HBM4 chips in the fourth quarter of 2025, with a full-scale sales expansion planned for 2026. This aggressive roadmap aims to maintain its competitive edge against rivals like Samsung Electronics and Micron.
Decoding the Memory Chip ‘Super Cycle’
The term “super cycle” is gaining traction, and SK Hynix is confident that the memory chip market is entering an extended one. A super cycle refers to a prolonged period where demand significantly outstrips supply, leading to sustained higher prices and robust earnings. This particular cycle is directly linked to the burgeoning global investments in artificial intelligence.
Analysts are largely in agreement with this optimistic outlook. HSBC analysts Ricky Seo and Hank Il Chang predict that growing use of AI applications will prompt cloud computing service providers to increase investments in data centers, driving both DRAM and NAND chip prices higher over the next one to two years. They forecast SK Hynix’s operating profit could reach 15 trillion won in the final quarter of this year and hit 67 trillion won for 2026, as reported by Dow Jones Newswires.
Nomura analysts, led by C.W. Chung, have also given a similarly upbeat assessment, speaking of an “unprecedented supercycle” that is approaching. They believe substantial AI-fueled investments by big U.S. tech companies are likely to continue into 2027 and even 2028. This long-term projection suggests that the current boom is not a fleeting trend but a fundamental shift in the market.
M.S. Hwang, a research director at Counterpoint Research, anticipates the HBM market alone will grow sharply, reaching an estimated $43 billion by 2027. This immense growth is expected to provide strong earnings leverage for memory manufacturers like SK Hynix, with DRAM and NAND chip prices potentially rising 15%-20% in the fourth quarter, signaling strong earnings guidance from the company.
Structural Constraints and Industry Implications
A crucial factor underpinning this super cycle is a structural constraint on memory supply growth. While demand is accelerating, SK Hynix noted that the shift in production towards complex HBM chips inevitably strains the output of conventional memory products like DRAM. Kim Kyu-Hyun, head of DRAM marketing at SK Hynix, explained that “total output will inevitably remain limited.”
This limitation is leading to an intense scramble among customers to secure supplies. The company observed that a growing number of clients are placing advance orders for next year and pushing for long-term contracts. This reflects market anxiety and a proactive effort to lock in essential components for future AI deployments.
The company further detailed that DRAM inventories have fallen to “extremely low levels,” and for DDR5 products, newly produced chips are shipped immediately to meet urgent demand. This immediate consumption of new supply highlights the severe imbalance between supply and demand that characterizes a super cycle.
Strategic Investments and the AI Community
With confidence growing in the long-term earnings potential of the AI market, global AI companies are significantly ramping up their investment spending. This drives rapid growth in demand across a wide spectrum of memory products, including HBM, DDR5, and enterprise SSDs. A recent letter of intent with ChatGPT maker OpenAI for HBM supply underscores the critical importance of securing these advanced products for leading AI innovators.
To meet this escalating demand, SK Hynix expects its own capital expenditure to grow substantially from this year’s levels. This commitment to increased investment signals the company’s intent to capitalize on its advantage in advanced chips and to keep pace with the industry’s rapid expansion.
For the AI community, developers, and data center operators, a prolonged memory super cycle has significant implications. Expect continued high prices for cutting-edge memory, potentially impacting the cost of scaling AI infrastructure and developing next-generation AI models. However, it also signifies a vibrant, innovative ecosystem where investment pours into groundbreaking technologies, pushing the boundaries of what AI can achieve. The race for advanced memory is more intense than ever, and SK Hynix is firmly in the lead, shaping the future of AI. This outlook was echoed by Reuters, which noted SK Hynix’s expectation of a prolonged “super cycle” due to limited supply growth and expanding demand for AI applications.
