Samsung Hikes Foundry Prices Up to 15% as AI Demand Shifts Orders to Korea

For most of the past four years, Samsung Electronics’ foundry business has been the sad chapter in an otherwise triumphant corporate story. While the company’s memory division rode the AI boom to record profits, the contract chipmaking unit bled red ink every year since 2022, unable to close the gap with Taiwan Semiconductor Manufacturing Co. (TSMC), which controls more than 70% of global foundry revenue against Samsung’s 7%.

That narrative just flipped. According to a Reuters report published August 19, 2026, Samsung has raised prices for some of its advanced contract chipmaking services by up to 15% for new orders — and customers are paying, not walking away.

What happened

The price increases took effect in July for chips made on Samsung’s 4-nanometer process, known as SF4, according to two people familiar with the matter who spoke on condition of anonymity. The details reveal a carefully tiered strategy:

• SF4 (4nm): Chinese and U.S. customers saw increases of 10–15% month-over-month, while customers in Taiwan — TSMC’s home market — saw milder increases of 5–10%.
• SF5 (5nm): Prices rose 10–15%.
• 8nm (legacy): Prices rose by nearly 10%.

Samsung declined to comment, citing a policy of not disclosing operational matters.

The most striking detail: Chinese customers, whose demand has been particularly strong, are among those accepting the steepest increases. U.S. export controls on advanced chipmaking equipment have made it harder for Chinese firms to build leading-edge capacity domestically, increasing their reliance on overseas foundries — and their tolerance for price hikes.

Why Samsung suddenly has leverage

The story underneath the story is capacity. TSMC’s leading-edge nodes are effectively sold out, booked solid by AI chip demand. When the dominant supplier can’t take more orders, demand spills to rivals — and rivals gain the power to charge for it.

Samsung’s SF4 production line at its Pyeongtaek campus in South Korea has reportedly been running at full capacity since late last year. That line produces logic chips for customers including Qualcomm, as well as the base dies used in Samsung’s own multi-layer high-bandwidth memory (HBM) chips — the memory that sits alongside AI accelerators like Nvidia’s.

“As TSMC faces tight capacity and raises prices, customers are shifting to rivals such as Samsung and Intel, prompting Samsung to raise its prices as well,” said Lee Min-hee, a Seoul-based analyst at BNK Investment & Securities.

The customer pipeline explains why the lines are full. Samsung secured foundry deals from Tesla and Apple last year, announced an AI chip production contract with Broadcom in July, and Nvidia CEO Jensen Huang said in March that Samsung would manufacture its new AI inference processor. Reuters’ sources add that Google is in talks with Samsung to manufacture chips using the SF4 process.

The profitability inflection

Samsung has been signaling this moment for months. In July, the company said it expects the foundry unit to return to profit “in the near future,” helped by higher factory utilization, improved production yields, and firmer pricing. It projected that advanced processes would account for more than half of foundry revenue this year, with AI and high-performance computing applications making up more than 30% — up from 15–20% in late 2025.

The company also guided that rising sales to major U.S. and Chinese customers, plus demand for HBM base dies, should lift foundry revenue by double-digit percentage points in the second half of the year versus a year earlier.

If the price hikes stick, the financial implications are significant. “If Samsung raises prices from here, its foundry business could potentially become profitable as early as next year, earlier than previously expected,” BNK’s Lee told Reuters.

Why it matters

For the AI supply chain: Foundry pricing is a leading indicator of AI infrastructure costs. When TSMC is full and Samsung raises prices by double digits, the cost pressure eventually propagates through the entire stack — from chip designers to cloud providers to end users of AI services. The era of cheap, abundant leading-edge logic capacity is over, at least for now.

For geopolitics: The fact that Chinese customers are accepting the largest increases is a direct consequence of U.S. export controls. Restrictions intended to slow China’s domestic chip capabilities have instead created a captive class of customers for overseas foundries — a dynamic that benefits Samsung’s bottom line while complicating the broader de-risking narrative.

For the Samsung-TSMC rivalry: A 7% market share player raising prices into strength is not a challenger overtaking the leader — but it is a challenger that no longer has to buy every deal with discounts. Pricing power is the first tangible evidence that Samsung’s foundry turnaround, long promised, may finally be arriving. Yield improvements, a full 4nm line, and marquee customers (Tesla, Apple, Broadcom, Nvidia, potentially Google) add up to real momentum.

For the memory duopoly: The HBM base die demand tying foundry to memory underscores how intertwined Samsung’s businesses have become. Every AI accelerator sold pulls HBM — and now foundry capacity — through Samsung’s ecosystem.

The caveat

These figures come from anonymous sources discussing sensitive commercial matters, and Samsung itself won’t confirm them. “Up to 15%” describes the top of a range on some processes for new orders, not a uniform increase across the business. And foundry profitability projections depend on the hikes holding as TSMC brings new capacity online through 2027.

But the direction is unambiguous: AI demand has tightened the global supply of advanced logic manufacturing to the point where even the industry’s second-place foundry can name its price — and where the customers with the fewest alternatives, in China, are paying the most.