Samsung Electronics on Tuesday announced another successful quarter, forecasting a third consecutive record operating profit, as demand for artificial intelligence chips continued to drive growing memory sales.

But investors were not impressed.

Following the forecast, Samsung shares fell as much as 10% before closing down 7%. The heavyweight stock dragged the Kospi to a loss of almost 5%, while its rival SK Hynix finished down 6%.

Samsung estimated second-quarter operating profit of 89.4 trillion South Korean won, or $58.7 billion, nearly 19 times higher than a year earlier and above analysts’ expectations. The company expects revenue to more than double to 171 trillion won.

However, the strong gains failed to satisfy a market that has become accustomed to spectacular results from AI’s biggest winners.

Samsung’s profits have been boosted by strong demand for specialized memory chips that help train and run AI models. Analysts expect demand to continue to outstrip supply.

“The memory cycle is still strong, but the market is starting to wonder if the easy part of trading is behind us,” Charu Chanana, chief investment strategist at Saxo, wrote in a Tuesday note.

While Samsung’s results reinforced the strength of demand for AI-driven memory, investors are now looking for more than just strong sales. They want confident guidance, lasting pricing power and signs that the AI ​​boom is not peaking, he added.

“The question is no longer whether demand for memory is strong. It’s whether today’s shortage could eventually become tomorrow’s overcapacity problem if supply returns too aggressively,” Chanana wrote.

Samsung is not alone. Nvidia, the face of the AI ​​chip boom, has also seen investors react coolly to strong earnings in recent quarters, as Wall Street demanded bigger and bigger surprises.

In November, Nvidia CEO Jensen Huang acknowledged the no-win dynamic after the chipmaker’s third-quarter earnings.

“If we had a bad quarter, it’s evidence that there’s an AI bubble. If we had a great quarter, we’re fueling the AI ​​bubble,” he told employees.

James Thorne, chief market strategist at Wellington-Altus, said AI trading has become a victim of its own popularity.

“That’s what happens when a bottleneck trade fills up: the fundamentals remain strong, but the earnings stop impressing because perfection was already priced in,” he wrote in X on Monday.

He said he believes the next phase of AI development could reward other parts of the market, such as power, grid capacity, cooling and physical infrastructure.

“Artificial intelligence is not over, but easy trading is,” Thorne wrote.