Samsung Printed Record Profit and the Stock Fell 10%, Because the Market Was Reading the Next Quarter, Not the Last

What Happened

By the numbers, Samsung’s preliminary second-quarter guidance on July 7 was a record. Operating profit ran roughly 19 times higher than a year earlier, a preliminary figure near 89 trillion won (about $58 billion) that topped the market estimate in the 87 trillion won range, its third straight record quarter. AI memory demand led by HBM pushed prices up. And yet the stock dropped more than 10% intraday right after the release. Over $100 billion in market value evaporated in a single day, and SK Hynix slid alongside it in double digits, dragging the KOSPI lower. That disconnect, strong results and a collapsing stock, is the story. The market saw two things. First, revenue of 171 trillion won was up 129% year on year but came in slightly below forecasts, read as a sign that DRAM price increases were more moderate than hoped. Second, a beat this size was already priced in, and what investors actually wanted was not the quarter just past but the next one, and how much longer the AI spending boom would last.

What This Means for Founders

Equity markets price the second derivative, not the level. Not whether profit was large, but whether the rate of profit growth can accelerate from here. Samsung printed a record and still tripped over the question “isn’t this the peak?” The first lesson for a founder is that good news gets priced in. When the upside is already baked into a valuation, into a funding valuation, into how hot the hiring market feels, the moment that news becomes real the market moves on to the next story. That is doubly true for a startup riding the AI boom through a round. What works with investors now is not “we’re AI too” but “here is the demand that survives after the boom cools.”

The second lesson is cycle exposure. The biggest risk analysts flagged for the memory rally was not competition or pricing but a slowdown in AI infrastructure investment. If data-center capex bends down, HBM demand cools first, and the shock flows top to bottom, past memory into GPU clouds, inference APIs, and the AI applications sitting on top. Map where your business hooks into that chain and how much of its revenue depends on the boom continuing. Conversely, if this sell-off is an overreaction, it may open undervalued opportunities near the memory value chain. The gap between market fear and the real demand curve is always where a founder gets to step in.

What You Can Do Now

Split your revenue into the part that depends on the boom and the part that survives without it. If the latter is thin, calculate now how many months of runway you hold in a scenario where AI infrastructure spending slows. If you build near the semiconductor value chain, track the DRAM and HBM price curves and your large customers’ capex guidance as leading indicators; a slowdown in orders always shows up in price signals first. And if you are raising, factor this shift in temperature, the market starting to doubt the durability of the AI boom, into your round timing and your valuation expectations.