A chip factory that once needed more customers is suddenly in a stronger position to choose which orders it takes — and what those orders cost. Samsung Electronics has raised prices on some advanced contract chipmaking services by as much as 15% for new orders, according to Reuters, as artificial-intelligence demand consumes available manufacturing capacity across the industry.
The increases cover several important manufacturing processes, including Samsung’s 4-nanometre, 5-nanometre and older 8-nanometre technologies. Behind the numbers is a much larger shift. AI companies are seeking enormous quantities of computing silicon, automotive manufacturers are putting increasingly sophisticated processors into vehicles, and dominant foundry TSMC is already heavily booked. That combination is giving Samsung something its foundry operation has struggled to command for years: pricing power.
Samsung Is Charging More for Several Important Chip Processes
The biggest reported increases are hitting customers ordering chips manufactured with Samsung’s SF4 and SF5 technologies. Reuters reported that Samsung raised prices in July for new SF4 orders, with customers in China and the United States facing increases of roughly 10% to 15% from the previous month. Taiwanese customers reportedly encountered smaller increases of about 5% to 10%. Prices for wafers produced on Samsung’s 5nm SF5 process increased by 10% to 15%, while pricing for its older 8nm technology rose by nearly 10%.
Those differences matter because semiconductor manufacturing contracts are rarely governed by a single public price list. Process technology, order size, customer relationship and available factory capacity can all influence negotiations. Samsung itself declined to discuss the reported adjustments, saying it does not provide details about operational matters. The result is therefore less like a universal 15% surcharge and more like targeted repricing at production lines where demand has strengthened enough to give the manufacturer considerably more leverage.
TSMC’s Dominance Has Made Alternative Capacity More Valuable
Samsung is benefiting partly from a problem facing its biggest competitor: too many customers want access to TSMC’s most desirable production lines. Counterpoint Research estimated that TSMC controlled about 73% of the pure-play foundry market in the first quarter of 2026, while Samsung held roughly 7%. Counterpoint also calculated that the overall sector expanded 30% from a year earlier, with AI GPUs, custom ASICs and related advanced packaging among the main growth drivers.
That imbalance makes even a distant second-place supplier strategically important. TSMC remains the manufacturing partner of choice for many of the semiconductor industry’s largest designers, but high utilization on advanced processes limits how quickly additional orders can be accommodated. Reuters reported that customers facing tight TSMC capacity have increasingly considered alternatives including Samsung and Intel. For Samsung, scarcity elsewhere changes the negotiating equation. A factory competing mainly on price has limited leverage; a factory holding one of the few available production slots for an urgently needed chip has substantially more.
Samsung’s 4nm Production Is No Longer Waiting for Work
One of the clearest signs of that change can be found at Samsung’s massive Pyeongtaek manufacturing complex in South Korea. Reuters reported that its SF4 production line has been operating at full capacity since late 2025. That line serves outside customers including Qualcomm while also producing base dies used in Samsung’s own high-bandwidth memory products. In other words, external chip designers are not simply competing with one another for manufacturing space; some capacity must also support Samsung’s internal semiconductor businesses.
Samsung had already signalled the tightening conditions months earlier. In its first-quarter 2026 results, the company said its foundry operation intended to reach full utilization of advanced-node production lines during the second quarter, supported partly by increased HBM4 base-die production. By the second-quarter update, Samsung said foundry earnings had improved significantly before certain provisions, citing stronger HBM base-die demand and orders from U.S. customers. A production line that is fully utilized provides far more room to reject low-margin work or renegotiate the economics of new orders.
AI Customers Are Taking an Increasingly Large Share of the Foundry Business
Artificial intelligence is the most immediate force behind Samsung’s new pricing power. Reuters reported that Samsung expects advanced processes to generate more than half of its foundry revenue in 2026. AI and high-performance-computing applications are expected to represent more than 30%, compared with roughly 15% to 20% in late 2025. That is an unusually rapid change in customer mix for a manufacturing business requiring years of investment to add meaningful capacity.
Samsung is also accumulating recognizable AI customers. Nvidia has said Samsung is manufacturing its new Groq-based LP30 inference processor on a 4nm process. Samsung and Broadcom expanded their semiconductor partnership in July, including cooperation on 2nm-and-below foundry technology, advanced packaging and next-generation HBM. Google has separately been reported to be discussing manufacturing arrangements with Samsung for future AI silicon. Each program is different, but collectively they demonstrate why capacity can tighten quickly: AI development is no longer centred on one type of GPU. Hyperscalers, networking companies and accelerator designers are commissioning increasingly specialized chips of their own.
The Automotive Race Is Becoming Part of the Same Advanced-Chip Story
Automotive demand deserves some nuance. Reuters identified AI demand as the immediate driver behind Samsung’s latest foundry price increases, rather than saying automakers alone caused the increases. Yet vehicles are steadily becoming another important destination for advanced manufacturing capacity. Samsung says it already mass-produces in-vehicle infotainment solutions across 14nm, 8nm, 5nm and 4nm technologies. It is also developing advanced 4nm and 2nm processes for higher-performance driver-assistance applications.
Tesla provides the most visible example of how the automotive and AI categories are converging. In 2025, Tesla struck a $16.5-billion semiconductor supply agreement with Samsung running through 2033. Elon Musk said Samsung’s Taylor, Texas, facility would manufacture Tesla’s next-generation AI6 processor, a chip expected to support the company’s future artificial-intelligence systems. Such deals blur the traditional boundary between “automotive chip” and “AI chip.” Modern vehicles increasingly require computing platforms for perception, infotainment and assisted-driving functions, meaning an automaker can effectively compete for the same sophisticated manufacturing ecosystem sought by technology companies.
Chinese Customers Are Facing Particularly Strong Pressure
Demand from China has added another layer of competition for Samsung’s available production. Reuters reported that Chinese customers have been particularly active in seeking foundry capacity and are among those accepting the steepest price increases. Samsung has reportedly been unable to fulfill every requested order because it must balance those customers against U.S. clients and its own internal production requirements. For some Chinese and American customers using SF4, the reported month-to-month increase reached the 10% to 15% range.
Geopolitics is part of that equation. U.S. restrictions on exports of advanced semiconductor manufacturing technology have limited Chinese companies’ ability to obtain some of the equipment needed to expand domestic leading-edge production. That increases the value of accessible overseas manufacturing capacity. The effect is visible beyond Samsung. China’s largest foundry, SMIC, recently reported strong utilization and raised prices for some highly sought-after production after negotiations with customers. Instead of relieving pressure in one market, soaring AI demand is creating interconnected capacity problems across several parts of the semiconductor manufacturing system.
Higher Prices Could Accelerate Samsung Foundry’s Financial Turnaround
For Samsung, the pricing change arrives after a difficult period. Industry estimates cited by Reuters indicate its foundry operation had been loss-making since 2022 as it struggled with utilization, advanced-process yields and the enormous technological gap separating it from TSMC. The economics of semiconductor fabrication make utilization particularly important because expensive facilities and equipment continue generating depreciation and other fixed costs whether production lines are full or partially idle.
Conditions are now moving in Samsung’s favour. The company said in July that improved factory utilization, better production yields and firmer pricing were supporting the foundry business. It also expects stronger sales to major U.S. and Chinese customers and growing HBM base-die production to help produce double-digit year-over-year foundry revenue growth during the second half of 2026. An analyst quoted by Reuters said stronger pricing could potentially bring the operation back to profitability as early as next year. Samsung’s record company-wide semiconductor earnings should not be confused with foundry profitability, however, because booming memory prices remain a major contributor to the wider chip division’s performance.
The Price Increase Signals a Broader Change in the Chip Market
Samsung is not operating in isolation. TSMC is reportedly planning chipmaking price increases of as much as 10% beginning in 2027 as it deals with higher material, equipment and overseas-factory costs. SMIC has already increased prices for sought-after capacity in China. TrendForce has also tracked tightening conditions in parts of the mature-node market, where reductions in 8-inch capacity are colliding with demand for power-management chips used in AI infrastructure and other electronics.
That does not mean a 15% foundry increase automatically becomes a 15% increase in the price of a vehicle, server or consumer device. Wafer manufacturing is only one part of a finished product’s total cost, and suppliers can absorb increases, redesign components or negotiate longer-term agreements. The more important signal is strategic. Advanced manufacturing capacity is becoming something major customers may need to reserve years ahead. Samsung, once trying to fill underused factories and regain lost customers, now has enough demand on key lines to charge more. For AI developers and increasingly sophisticated automotive programs, securing production may matter almost as much as designing the chip itself.
