Samsung Posts Record Profit as AI Chip Boom Accelerates, But Investors Question How Long It Lasts

Record Profit, But a Sharp Reality Check for AI Sentiment

Samsung Electronics flagged a nineteen-fold jump in second quarter operating profit on Tuesday, estimating April to June earnings at 89.4 trillion won, around 58.44 billion US dollars, comfortably beating analyst forecasts of 87.3 trillion won. Revenue is expected to rise 129 per cent year on year to 171 trillion won. It is the company's third consecutive record quarter and surpasses its combined earnings over the previous three years.

The market's reaction told a different story. Samsung shares fell as much as 7.9 per cent in early trade, dragging rival SK Hynix down 7.3 per cent and the benchmark KOSPI index down 6 per cent, even as the numbers beat expectations. Analysts pointed to a simple explanation: the results were already priced in after a sharp rally in Samsung shares ahead of the announcement, and investors are increasingly nervous about whether the broader rush by enterprises worldwide to scale their AI infrastructure can keep generating returns that justify the spending.

What Is Driving the Memory Chip Boom

The profit surge is being powered by sustained increases in memory chip prices as AI spending broadens beyond high bandwidth memory into conventional DRAM and NAND products used across smartphones, PCs and enterprise servers. Citi Research data cited in the results showed average selling prices for DRAM and NAND rising 44 per cent and 53 per cent quarter on quarter respectively. Samsung's profit grew even after setting aside sizeable provisions for staff bonuses tied to a wage deal agreed with semiconductor workers in May, a deal that links pay directly to operating profit. Analysts estimate that without those provisions, operating profit would likely have exceeded 100 trillion won.

Jeff Kim, head of research at KB Securities Jefferies, said the memory chip shortage is likely to deepen through this year and next, since capacity growth remains limited while demand stays strong. Building new fabrication capacity takes years, which is why some analysts now argue the current boom is becoming structural rather than the traditional boom and bust cycle memory chips have historically followed.

The Question Every AI Investor Is Now Asking

The bigger risk analysts are watching is whether the AI infrastructure spending underpinning this demand is durable. Goldman Sachs estimates combined AI related capital expenditure by the four largest hyperscalers, Meta, Microsoft, Amazon and Alphabet, will reach 5.3 trillion US dollars between fiscal 2025 and 2030. Investors have raised concerns that these companies will need to borrow heavily to fund that build out with uncertain returns, and that delays to US data centre construction, whether from labour shortages, power constraints or local opposition, could eventually soften demand across the entire AI hardware supply chain.

Samsung itself is committing further to the cycle regardless of the near term jitters, having announced plans last week to invest 2,100 trillion won in South Korea through 2040, though it noted that spending would be adjusted according to market conditions.

What It Means for Enterprise AI Buyers in the Gulf

For enterprise technology leaders in the region, the disconnect between Samsung's record earnings and its falling share price is a useful signal in its own right. Memory prices are unlikely to soften soon given the structural supply constraints analysts describe, which has direct implications for the cost of building or expanding AI infrastructure, including the kind of enterprise deployments already accelerating across the UAE's enterprise AI acceleration partnerships. Organisations planning AI infrastructure investment over the next 12 to 18 months should factor sustained hardware cost pressure into their budgeting, rather than assuming prices will normalise quickly.