Samsung Beats Q2 Profit Forecast, Stock Falls on AI Capex Doubts

Published by James Harris on July 7, 2026July 7, 2026

Samsung just reported a near-19-fold jump in operating profit for Q2 2026, posting roughly 89.4 trillion won against 4.68 trillion won a year earlier, driven almost entirely by AI memory demand. The shares fell anyway.

The disconnect is the story. Samsung beat the LSEG SmartEstimate of 87.3 trillion won, logged what the Financial Times called its third consecutive quarter of record operating results, and still could not hold its stock price. This is a pattern familiar from prior tech cycles: once expectations become calibrated to near-perfect conditions, a beat is just confirmation of what was already priced in. Ryu Young-ho at NH Investment and Securities put it plainly, saying investors are questioning whether massive AI infrastructure spending by big tech will continue at its current pace. That concern is not abstract. The hyperscale buildout driving demand for high-bandwidth memory, server DRAM, and NAND flash is ultimately a function of capital expenditure decisions made in Menlo Park and Redmond, and those decisions are increasingly subject to scrutiny over whether AI returns justify the outlay. The growing entanglement of AI infrastructure with government access frameworks adds another variable to how freely that capex flows.

Tom Kang at Counterpoint Research said the market was expecting even stronger results, which tells you more about where sentiment sits than the numbers themselves do.

The regional spillover matters here. South Korea accounts for over a quarter of the MSCI Emerging Markets Asia Index, so a selloff in Korean tech is not contained. SK Hynix, arguably the clearest beneficiary of HBM demand given its supply position with Nvidia, fell alongside Samsung despite having less to answer for operationally. Micron came under pressure too, confirming that investors are repricing the entire memory complex rather than reacting to any single company’s guidance. The question now is whether HBM pricing holds if hyperscaler capex guidance softens in upcoming earnings calls, particularly given that Beijing’s restrictions on foreign AI chips and its pressure on domestic firms to limit American partnerships are quietly reshaping who builds AI infrastructure and where.

According to Reuters, analysts still expect the favorable pricing environment to continue on the basis that the AI supply chain remains constrained. Samsung’s full second-quarter results, which will include a segment breakdown and management commentary on HBM allocation, are expected later this month and will likely clarify whether the margin expansion is holding across product lines or concentrated in a narrower slice of the memory stack.