Memory semiconductor companies' profitability has risen to a top global level amid the AI boom. It is a scene that would not have been easy to predict in the past.

Samsung Electronics as well as SK Hynix and U.S. Micron Technology are expected to enter the global top 10 by net profit. As recently as last year, no memory company was in the global top 10 by net profit, but this year has changed by 180 degrees.

According to a recent Wall Street Journal report, Samsung Electronics, if the current trend continues, is expected to rank second by net profit behind Nvidia this year, surpassing so-called big tech companies such as Alphabet, Google's parent, Microsoft and Apple. SK Hynix and Micron are expected to take sixth and ninth place, respectively.

Samsung Electronics, SK Hynix and Micron combined are expected to post net profit of $350 billion this year.

Samsung Electronics' first-quarter net profit topped $30 billion, helped by a memory boom. Some 94 percent of total profit came from the semiconductor division. It is well beyond the prior quarter's record and close to the company's annual peak net profit. With concerns growing over whether AI services can generate profits, providers of infrastructure are raking in profits, the WSJ reported.

For now, there is no sign the earnings uptrend at memory companies will end. Considering Samsung Electronics' order backlog, the phenomenon of supply failing to keep up with demand will worsen further next year, the WSJ reported.

Memory prices keep rising amid the supply shortage. According to market research firm TrendForce, first-quarter memory prices rose by nearly 100 percent from the previous quarter. That is nearly double the expected level. Related shares also appear to be surging amid the memory chip boom. Since early this year, Samsung Electronics shares have risen 72 percent, while SK Hynix and Micron have jumped 90 percent and 65 percent, respectively.

The AI fever also appears to be evolving into a favorable phase for memory companies. Over the past few years, memory companies have focused on producing HBM (high-bandwidth memory), which is used widely to train large language models when combined with Nvidia GPUs. HBM is made by stacking multiple layers of DRAM.

More recently, memory demand has been rising sharply not only for LLM training but also for so-called inference, which uses trained LLMs in real-world settings. The spread of inference has raised demand for general-purpose servers that use conventional memory, and that has led to net profit at Samsung Electronics, SK Hynix and Micron rising to a higher level, the WSJ reported.

Memory is broadly divided into two types: DRAM and NAND flash. DRAM supports fast data processing in servers, PCs and other electronic devices through short-term storage, while NAND flash is used to store data for a long time in smartphones and other devices.

Flash memory companies are also maintaining high growth thanks to AI. SanDisk produced results that exceeded expectations for its fiscal third quarter. The company said demand for its high-performance enterprise SSDs is rising as the spread of AI drives a sharp increase in demand for high-capacity storage. As AI systems require more data storage, demand for flash memory chips is outpacing supply, enabling SanDisk to raise prices, Reuters reported.

SanDisk also appears confident about the fourth quarter. SanDisk forecast fourth-quarter revenue of $7.75 billion to $8.25 billion. That is well above the average analyst estimate of $6.49 billion. SanDisk CEO David Goeckeler (데이비드 괴켈러) said, "This quarter is a fundamental inflection point for SanDisk," adding, "Based on technology leadership, we are shifting our business structure to high value-added markets centered on data centers."