SK hynix shares topped 600,000 won for the first time, setting a record high. The share price is not only high, but the rally is so steep that the stock has been designated as an investment warning issue. After starting in the 170,000-won range at the beginning of the year, SK hynix shares jumped 262% in 306 days.
In the past, such a surge would have triggered bubble warnings, but the securities industry is instead raising target prices. The average target price from 25 brokerages that issued targets for SK hynix is 698,462 won. The recent rise in SK hynix shares seems hard to explain by earnings improvement alone.
Experts say SK hynix's position has changed from before due to growth in the artificial intelligence (AI) market. In an absolute supply shortage where production cannot keep up with demand, SK hynix is said to be holding the initiative in the market. Thanks to this, securities firms' perspectives on estimating the corporate value of SK hynix are also changing.
1) The rise of a new industry called "AI"
On the 3rd, when SK hynix shares exceeded 600,000 won for the first time ever, SK Group Chairman Chey Tae-won said, "Many corporations are requesting supplies of memory semiconductors, and we are struggling with how to respond to them all," adding, "We are in an era where bottlenecks appear in supply."
What caused the severe shortage of memory semiconductors is the emergence of AI, which is changing the paradigm of global industry, and global big tech's AI infrastructure investments. Yet, the supply of memory semiconductors—an essential for AI and the components for data storage and processing—has not kept pace.
Since the advent of Generative AI, the size of the global AI industry has grown rapidly. According to MarketsandMarkets, the global AI industry grew from $37.8 billion in 2020 to $257.6 billion in 2024. The compound annual growth rate reaches 61.6%. To preempt the Generative AI market, global big tech is pouring in massive funds. In particular, they have significantly expanded investments related to computing infrastructure such as AI Semiconductors and data centers, which are needed for high-performance AI development.
Here, SK hynix's high bandwidth memory (HBM) plays a key role. HBM is a high-performance memory semiconductor made by stacking multiple layers of DRAM, and compared with conventional DRAM, it has a much faster data transfer speed.
Generative AI like ChatGPT requires processing vast amounts of data, and HBM is being adopted at scale in graphics processing units (GPUs), which are essential for AI training and inference.
SK hynix began mass production of HBM3 in June 2022 for the first time in the world, and has continued to raise performance, establishing a mass production system for 12-layer HBM4 this year. Heungkuk Securities' research center projected that SK hynix's sales next year for sixth-generation HBM (HBM4) will reach about $19.4 billion (about 28 trillion won). That far exceeds the expected sales of rival Samsung Electronics or U.S. Micron.
2) Market leadership shifts to chipmakers
The shortage has not only increased workloads for chipmakers. It has also given chipmakers a chance to seize market leadership.
Until now, chipmakers expanded facilities ahead of time by forecasting orders from clients. They could not respond nimbly to changes in demand, and if they failed to forecast demand and oversupply occurred, it inevitably led to deteriorating results along with a downturn cycle.
But with the emergence of the AI industry, semiconductor demand has increased sharply, and the market structure has begun to change. Explosive demand growth and supply constraints are turning the semiconductor industry into a "order first, expand later" structure. Also, while the chip industry used to see two to three years of boom followed by a downturn, the boom period is expected to last longer with the advent of AI.
Lee Min-hee, an analyst at BNK Investment & Securities, explained, "Unlike the past cycle centered on IT products, global AI infrastructure investments are changing chip order contract methods to long term, causing a structural shortage of products."
Ko Young-min, an analyst at DAOL Investment & Securities, explained, "Over the past two years, only products directly linked to AI investment were booming, but as this AI trend expands, we are seeing broader demand growth for memory semiconductors."
3) Changing the 'valuation yardstick' for fair price
Given the situation, some are even changing the very standards used to calculate fair prices for semiconductor corporations. SK Securities' setting of a 1 million won target price for SK hynix is a representative case.
SK Securities applied the price-earnings ratio (PER) instead of the price-book value ratio (PBR) that had been used to value semiconductor corporations. It set the 1 million won target by applying a PER of 11 times to next year's expected earnings per share (EPS).
The memory semiconductor industry has typically seen stark booms and busts depending on the cycle, leading to wide swings in results. Accordingly, the PBR method, which values companies by stable net worth rather than net income, was used. However, in the AI-led memory semiconductor cycle, it is hard to explain share price trends under the old method, so the valuation yardstick has been changed.
Han Dong-hee, an analyst at SK Securities, said, "The structure of the memory industry is changing within the AI cycle," adding, "Unlike the past, SK hynix's results over the past three years have not tracked macroeconomic trends, and the intensity of the memory cycle is getting stronger."
There is also analysis that, because SK hynix holds an advantage in the HBM market, profit volatility by cycle may not be as strong as before.
An analyst at a securities firm said, "In the days before HBM, the memory semiconductor cycle moved only with front-end demand for general-purpose memory, so there were ups and downs," adding, "Since SK hynix came to hold HBM, profit volatility due to cycle fluctuations has decreased."
However, the AI bubble theory that has been raised since the AI boom is an issue to watch closely. When global big tech first embarked on large-scale investments in AI, they used surplus cash, but recently, more cases have emerged of raising funds through corporate bonds for AI investment.
Song Myeong-seop, an analyst at iM Securities, said, "Although companies like OpenAI are currently leading the AI boom, if large net losses continue, there is a possibility that investment will suddenly shrink," adding, "We cannot be blindly optimistic that enormous investment like now will continue going forward."