Samsung's foundry can finally raise prices. Read whose shortage is doing the talking.

Samsung has raised the price of making advanced chips for other companies by as much as 15 percent, and the framing writes itself: the foundry is back. After years in the red, the story goes, Samsung finally has the pricing power that comes with being wanted. It is a clean narrative, and it is mostly true, and the part it leaves out is the part that tells you how long it lasts. A price increase is one of the most honest documents a company produces — it is a number the company charges rather than a sentence it says — so it is worth reading closely. Read closely, this one is less a story about Samsung getting stronger than about someone else running out of room.

What the increase actually says

Start with the numbers, because they are specific in a way that matters. According to the reports, Samsung lifted prices on its 4-nanometer process (the one it calls SF4) by 10 to 15 percent for customers in China and the United States, and by a smaller 5 to 10 percent for customers in Taiwan. Its 5-nanometer line went up 10 to 15 percent. Even the older 8-nanometer technology rose by nearly 10 percent. And the largest increases, per the reporting, are the ones Chinese customers are accepting.

Hold onto that last detail, because it is the tell. If Samsung were raising prices because its process had suddenly become more desirable on its merits, you would expect the increases to fall most heavily where Samsung competes hardest for prestige work — and you would expect the sophisticated buyers with the most alternatives to push back hardest. Instead the opposite is happening. The customers with the fewest alternatives are paying the most. That is not the signature of a company that has won an argument about quality. It is the signature of a company that is the last seat in a full room.

The room is full because of TSMC, not Samsung

Here is the context the price hike is quietly conceding. TSMC, which makes the majority of the world's advanced logic chips, has had its leading-edge capacity bought out by the AI build-out. Its next-generation 2-nanometer slots are reported sold out through the first half of 2027. When the dominant supplier is full, demand does not evaporate; it spills. And the largest vessel positioned to catch the overflow is Samsung. So the price increase is real, and Samsung is genuinely collecting it, but the force generating it originates in another company's order book. Samsung is charging more because TSMC cannot take the work, and the customers most exposed to that — including Chinese buyers with the narrowest set of options — are absorbing the largest markups because they have the least leverage to refuse.

You can see the imbalance in one pair of figures that no press release will put side by side. In the first quarter of this year, Samsung accounted for roughly 7 percent of global foundry revenue. TSMC accounted for more than 70. This is not a duel between near-equals in which Samsung has landed a blow. It is a market with one dominant maker running at capacity and a distant second collecting the demand that has nowhere else to go. The 15 percent is not a measure of how much Samsung has closed that gap. It is a measure of how tight the leader's constraint has become.

Samsung is charging more because TSMC is full. That is real money. It is also borrowed pricing power — rented from someone else's shortage.

The turnaround is real. It is also borrowed.

I want to be fair to the good news, because there is some, and Samsung has earned part of it. The foundry division has lost money since 2022 — a long, expensive stretch of building capacity customers did not fill. That is changing. Utilization has climbed; reported figures put it around 80 percent earlier this year, which is the difference between a fab that bleeds and a fab that pays. Yields on the advanced nodes have improved. The company said in the summer that it expects the foundry unit to return to profit in the near future, and some analysts now see that rebound arriving as early as the third quarter of this year, ahead of Samsung's own earlier guidance, helped by 2-nanometer orders reported up sharply. Higher utilization, better yields, firmer prices: that is a genuine operational recovery, and it is the honest core of the story.

But recovery and durability are different claims, and the price hike lets you tell them apart. An operational recovery — better yields, fuller lines — is something Samsung did and can keep. Pricing power sourced from a rival's capacity crunch is something the market handed Samsung and can take back. The two are arriving at the same time, which makes them easy to confuse, and the company has every reason to let you confuse them. The recovery is Samsung's. The pricing power is on loan.

The date on the loan

Loans have terms, and this one has a date on it. TSMC's leading-edge crunch is a function of how fast the industry's largest maker can bring new 2-nanometer capacity online — and it is bringing it. When those slots open through 2027, the overflow that is currently paying Samsung a premium acquires somewhere else to go. At that point the 15 percent is exactly as defensible as Samsung's process is on its own merits, against a supplier with more than ten times its foundry share. The question that decides Samsung's next few years is not whether it can collect a scarcity premium now. It plainly can. The question is what it builds with the money before the scarcity eases.

There is a version of this that ends well for Samsung, and it is worth stating plainly because I am not predicting failure. If the company uses this window — the fuller lines, the firmer prices, the customers who currently have nowhere else to go — to close the yield and trust gap that sent them elsewhere in the first place, then some of the borrowed pricing power converts into the earned kind. A customer who came to Samsung because TSMC was full, and stayed because the chips were good and the relationship worked, is a customer Samsung keeps at market prices. That is the actual prize, and it is won on the factory floor over the next several quarters, not in this month's price sheet.

What to watch instead of the headline

So when the coverage tells you Samsung's foundry is back, the accurate translation is narrower and more useful: Samsung's foundry is finally full, at a moment when the only supplier customers prefer cannot take their orders, and Samsung is charging the people with the fewest options the most to fit them in. Every clause of that is good for Samsung right now. Not one of them is guaranteed to survive TSMC adding capacity.

The things I would watch are not the price increase, which the company was always going to take, but the things that reveal whether the recovery is the durable kind:

• Retention, not wins. Which of the customers who arrived during TSMC's crunch place a second and third order once they have somewhere else to go.
• The share number, not the price number. Whether Samsung's slice of global foundry revenue actually moves off 7 percent, or whether it collects premiums without gaining ground.
• Yields on the leading edge. The 2-nanometer ramp is where trust is rebuilt or lost, and it is the only thing that converts borrowed pricing power into the earned kind.
• What happens to the 15 percent when TSMC's 2-nanometer slots open in 2027 — whether it holds, or quietly comes back off.

A price increase is the truest thing a foundry publishes, because a customer has to actually pay it. This one says demand is real and Samsung is well placed to catch it. It does not say Samsung has closed the gap with the company whose shortage created the opening. Those are two different sentences, and only the first one is on the invoice. The company would rather you read them as the same. The date on the loan is why you shouldn't.