Samsung could multiply its profit 18-fold thanks to the AI memory boom

Samsung Electronics is heading for another record quarter. According to LSEG data cited in the article, analysts project operating profit of around 86 trillion won (approximately $56 billion) for the second quarter, nearly eighteen times more than a year earlier. Reuters expects the company to string together its third consecutive quarter of record profits, and Samsung will release its preliminary earnings guidance on Tuesday.

The main driver of this jump is global demand for AI infrastructure. Not only are specialized HBM memory chips in short supply, but demand for standard DRAM and NAND memory is also growing strongly, as organizations adopt AI on a larger scale and traditional servers and storage systems require ever more memory.

The article notes that the rise of agentic AI plays a key role in this process. While earlier generations of AI focused mainly on training large language models, AI agents carry out complete tasks and workflows autonomously. During the inference phase, this demands more working memory and additional storage capacity to process and retrieve data, which translates directly into higher demand for memory chips.

According to Citigroup analysts mentioned in the text, average selling prices for DRAM memory rose 44% in the second quarter compared with the previous three months, while for NAND the increase reached 53%. Analysts expect the current shortage in the memory market to extend into 2027.

Despite this favorable backdrop, there is uncertainty over the final quarterly profit. In late May, Samsung signed a new collective bargaining agreement with employees of its chip division under which 10.5% of operating profit will go toward performance bonuses. Depending on when Samsung recognizes these obligations in its accounts, the effect on reported results could be significant: some analysts anticipate bonus provisions that would exceed 40 trillion won. Full quarterly results will be published later this month.

Looking further ahead, attention shifts to the pace of investment in AI infrastructure. As long as hyperscalers keep building new AI capacity, demand for memory chips will remain high. However, several analysts warn that this investment could level off over time. A recent JPMorgan Chase analysis cited in the article raises doubts about whether cloud providers will keep devoting an ever-larger share of their investment budgets to AI hardware in the coming years; this year more than half of their capital spending is expected to go toward AI, a share that could rise even further next year.

That risk is relevant for Samsung and SK Hynix, which last week jointly announced investments of approximately 3,200 trillion won to expand chip production in South Korea, a program Samsung will spread out through 2040.

The article also points to a less favorable side of this boom: Samsung's own mobile division now pays much more for memory components, which is squeezing its margins. Although Samsung has already raised the prices of several smartphones, analysts consider further increases likely later this year; it is noted that Apple has also recently raised the prices of several hardware products. Overall, the memory chip market should remain tight as long as AI workloads keep growing and new production capacity comes online only gradually.