Samsung Raised Foundry Prices, but TSMC Still Controls the Market

Samsung reportedly raised prices for selected foundry orders by up to 15%, reversing the discounts that recently defined its effort to attract chip designers.

The increases reportedly cover new orders using Samsung's 4-nanometer and 5-nanometer processes. Some 8-nanometer automotive products also became more expensive. A nanometer node identifies a manufacturing technology generation, although the label no longer describes one physical transistor dimension.

The underlying report appeared on July 9, 2026, rather than August 20. The story returned to news rankings after additional reporting described July increases across several customer regions. Samsung has not published a price list or confirmed individual customer terms.

That timing matters. This is not a new company announcement issued in August. It is a developing account of a July pricing change, supported by unnamed industry sources and broader evidence of tightening foundry capacity.

The reversal is still important. Samsung spent years competing with lower prices while TSMC accumulated more than 70% of global foundry revenue. Raising rates suggests Samsung finally has leverage in parts of its manufacturing portfolio.

However, pricing power does not erase the competitive gap. Samsung must show that higher rates reflect dependable demand, stronger yields, and repeatable customer wins. Otherwise, the increase remains a temporary benefit from capacity spilling out of TSMC.

Samsung Foundry Prices Reportedly Rose by Up to 15%

The reported increase is selective, customer-specific, and concentrated in manufacturing nodes where available capacity has tightened.

The original reported price increase concerned new customers seeking production on 4-nanometer and 5-nanometer lines. Selected 8-nanometer automotive orders were also included.

Subsequent reports added more detail. Samsung reportedly increased SF4 prices for some United States and Chinese customers by 10% to 15% during July. SF4 is Samsung's commercial name for its 4-nanometer process family.

Reported increases for Taiwanese customers were smaller, ranging from 5% to 10%. Orders using Samsung's SF5 process reportedly rose by 10% to 15%. Certain 8-nanometer products increased by nearly 10%.

These figures have not appeared in a public Samsung rate card. Foundry contracts are negotiated privately and can reflect volume, design support, packaging, scheduling, and long-term capacity commitments.

Samsung declined to discuss specific operating terms, according to reports carrying the new details. That response leaves the central claim credible but not independently confirmed through company documents.

The July 9 account was narrower than some later headlines. It focused on prices for new clients, not a universal increase covering every existing contract. That distinction changes how buyers should interpret the news.

A broad increase would demonstrate pricing power across Samsung's customer base. A selective increase can instead capture the highest bids for scarce production slots without disturbing established agreements.

The affected nodes also deserve careful interpretation. Four-nanometer and 5-nanometer processes remain useful for AI accelerators, networking chips, mobile processors, and high-performance computing designs.

They are not Samsung's newest technologies. Samsung is also developing and expanding 2-nanometer manufacturing, which uses gate-all-around transistors to improve electrical control as transistors shrink.

The price increase therefore reflects demand for proven capacity, not simply excitement surrounding the newest node. Customers often choose an established process when schedules, yields, and design costs matter more than maximum density.

Automotive 8-nanometer capacity presents a similar dynamic. Vehicle components face long qualification cycles because manufacturers must verify reliability across demanding temperatures and operating conditions.

Once a design completes that process, changing foundries becomes difficult. A customer might accept a higher wafer rate instead of repeating qualification with another manufacturer.

This combination gives Samsung room to price individual orders according to urgency. Customers seeking immediate capacity have fewer alternatives than customers planning a product several years ahead.

The story also represents a notable reversal. Reports earlier in 2026 described Samsung discounting some advanced manufacturing offers to compete for orders against TSMC.

Discounting can help fill factories, but it can also weaken margins. It becomes especially painful when low production yields reduce the number of usable chips obtained from each wafer.

Yield is the share of manufactured dies that pass testing. A foundry with lower yields must process more wafers to deliver the same number of functional chips.

Higher utilization can improve the economics because expensive equipment operates more consistently. Higher prices add another benefit, but only when customers believe the production slot is worth securing.

Samsung appears to have reached that position on selected lines. The company has not established that the same leverage extends throughout its foundry portfolio.

That qualification separates the event from the most optimistic interpretation. Samsung gained negotiating room where demand is strong. It did not suddenly match TSMC's scale, technology execution, or customer concentration.

AI Demand Changed the Capacity Equation

AI demand is pushing foundry competition away from simple price discounts and toward access to usable manufacturing capacity.

Designers do not buy an abstract process node. They need qualified production, packaging, intellectual property, engineering support, and delivery schedules that align with a finished system.

TSMC remains the preferred manufacturer for many leading AI processors. Its advanced lines serve products from major accelerator designers, cloud companies, smartphone vendors, and other high-volume customers.

That concentration creates a bottleneck. When TSMC's most desirable capacity fills, chip designers must either wait, pay more, redesign for another process, or seek an alternative foundry.

Samsung benefits from that overflow even when it is not the customer's first choice. A credible second source becomes more valuable when the leading supplier cannot provide capacity on the required schedule.

This is the mechanism behind the reported increase. Samsung does not need to surpass TSMC across the entire market before raising prices. It only needs more demand than available capacity on particular lines.

AI infrastructure broadens that opportunity. Demand no longer comes only from merchant processor companies selling standard chips to many customers.

Cloud providers increasingly commission custom accelerators and networking silicon. Memory systems require logic dies, controllers, and advanced packaging. Each layer can consume foundry capacity.

Samsung also has an unusual internal advantage. It operates major memory, logic, packaging, and foundry businesses within one company.

High-bandwidth memory, or HBM, stacks multiple memory dies to feed AI processors with data. These products require supporting logic and packaging steps alongside the memory itself.

Samsung can allocate some foundry output to internal products, including base dies used in advanced memory systems. That internal demand reduces the capacity available to outside customers.

The effect can support utilization even before Samsung wins enough external orders to challenge TSMC directly. It also complicates comparisons because internal and external demand do not carry identical commercial value.

Samsung highlighted this strategy in its second-quarter outlook. The company said it planned to expand 4-nanometer LPU and base-die sales during the second half of 2026.

An LPU is a processor designed for language-model inference workloads. A base die manages communication and control functions beneath stacked HBM memory.

Those products help connect Samsung's memory strength with its foundry capacity. They also show why 4-nanometer lines can tighten even as the industry discusses newer 2-nanometer technology.

Samsung said it would also increase production of mobile products using its second-generation 2-nanometer process. That plan places two very different challenges on the foundry business.

The first is extracting better economics from established nodes. The second is proving that newer manufacturing can reach sufficient yield and volume for demanding external customers.

The reported price increase addresses the first challenge. It turns fuller factories into higher revenue per order and reduces the need to compete primarily through discounts.

It does not resolve the second challenge. Customers considering an expensive new design care about performance, yield, software support, packaging, and delivery confidence.

Moving a design between foundries is rarely simple. Each manufacturer supplies different design rules, libraries, transistor characteristics, and packaging options.

A customer must adapt its chip, verify performance, and complete physical design again. That process consumes engineering resources and can delay a product.

This friction gives TSMC considerable protection. Customers cannot instantly redirect every design because Samsung offers a lower rate or an earlier production slot.

However, prolonged capacity constraints change that calculation. A delayed AI product can lose more commercial value than a redesign costs, especially during a rapid infrastructure expansion.

That pressure explains why Samsung can ask more for selected nodes. The alternative for some buyers is not a cheaper TSMC order. It is a later product launch.

The increase also signals a wider supplier advantage. Foundries have spent heavily on fabrication plants, lithography systems, process development, and advanced packaging.

When demand outruns qualified supply, manufacturers gain leverage over delivery terms. The cost reaches chip designers first, then moves through servers, networking equipment, vehicles, and consumer devices.

Individual wafer increases do not translate directly into identical product increases. Chip size, yield, packaging, memory, and system design all affect the final bill.

Still, the direction is clear. Buyers must budget for capacity security rather than assume that a maturing node will automatically become cheaper.

TSMC Remains Samsung's Real Opponent

Samsung's pricing gain matters because it emerged under TSMC's shadow, not because Samsung has displaced the market leader.

TrendForce estimated that TSMC held about 72% of global foundry revenue during the first quarter of 2026. Samsung remained second with approximately 6.5%.

The foundry market data also showed different quarterly directions. TSMC's revenue increased by 6.3%, while Samsung's declined by 5.8%.

That gap frames the entire story. Samsung's ability to raise some prices is commercially useful, but it starts from a much smaller market position.

TSMC's advantage is not limited to production volume. Its manufacturing processes sit inside a large network of design tools, validated intellectual property, packaging technologies, and customer engineering experience.

Customers know how to move designs through that system. Suppliers optimize tools around it, while designers build road maps that assume access to future TSMC nodes.

Samsung has its own design ecosystem and manufacturing partners. Its 2026 foundry forum covered technologies spanning 8-nanometer through 2-nanometer processes.

The company emphasized design technology co-optimization, which adjusts chip architecture and manufacturing rules together. It also presented intellectual property for AI, automotive, and high-performance systems.

Those investments are necessary because a process cannot succeed through transistor specifications alone. Customers need a complete path from early design through packaged, tested chips.

Samsung therefore competes with TSMC on trust as much as technology. A chip designer must believe production will arrive at the required volume and performance.

That trust develops through completed products. Every successful customer program gives engineers data, improves manufacturing learning, and makes the platform safer for the next buyer.

Pricing can accelerate or obstruct that cycle. Discounts attract experimental projects but can damage profitability. Higher rates improve economics but demand stronger confidence from customers.

Samsung's reported increase suggests its position has improved enough to charge more on constrained nodes. It does not show that customers now view Samsung and TSMC as interchangeable.

The distinction becomes sharper at leading nodes. TSMC has repeatedly translated new manufacturing technologies into large production volumes for major customers.

Samsung has introduced important transistor technologies, including gate-all-around manufacturing. Yet technical leadership on a road map does not guarantee commercial leadership in volume production.

Production yield remains the central test. A high quoted wafer rate can still deliver competitive chip economics when many dies work correctly.

Conversely, a discounted wafer can become expensive when too few dies pass testing. Buyers compare the cost of usable chips, not only the contractual rate per wafer.

Schedule reliability carries similar weight. An AI processor might anchor an entire server launch, model deployment, or cloud capacity plan.

Missing a manufacturing window can delay revenue across that system. Buyers will pay more to reduce that risk, which strengthens the incumbent with the best execution record.

Samsung must therefore use its newly reported pricing power carefully. Raising rates too quickly could discourage customers that originally approached the company for a lower-cost alternative.

Keeping rates too low would sacrifice the benefit of tight capacity. It could also make expansion harder to justify when advanced factories demand sustained investment.

A selective policy offers a middle path. Samsung can preserve negotiated terms for strategic clients while charging more for urgent or capacity-intensive new orders.

Regional differences support that interpretation. Reported increases vary by customer location, suggesting negotiations reflect demand, timing, and available alternatives.

This is not unusual in contract manufacturing. Two customers using the same named process can receive different terms because their designs, volumes, support needs, and commitments differ.

The reported increases should not be treated as a standardized market index. They are evidence of stronger bargaining power inside a private contracting system.

TSMC is also reportedly raising rates across advanced technologies. That reduces the risk that Samsung becomes uncompetitive merely by charging more.

The two companies still occupy different positions. TSMC can raise prices from a position of dominant share and broad customer dependence.

Samsung is raising selected prices while trying to convert overflow demand into lasting relationships. That makes execution after the contract more important than the increase itself.

What the 15% Figure Does Not Prove

A higher quoted rate does not prove that Samsung has fixed its yields, secured durable demand, or restored foundry profitability.

The first uncertainty is verification. Samsung has not publicly confirmed the reported percentages, customer categories, or affected order volumes.

The source reports rely partly on unnamed people familiar with negotiations. Their accounts can accurately describe individual deals without representing every contract.

Readers should therefore avoid treating 15% as a company-wide increase. "Up to" identifies the high end of a reported range, not the average change.

The second uncertainty concerns utilization. A line can become full because external customers are competing for capacity, because internal products consume it, or because available equipment is limited.

Each scenario produces different long-term implications. Strong third-party demand validates Samsung as an independent foundry. Internal allocation improves utilization but provides weaker evidence of outside confidence.

The third uncertainty is yield. Samsung does not publish customer-specific production yields, and public estimates often use inconsistent definitions.

Improvement is plausible when a process remains in production for several years. Engineers identify defects, refine equipment settings, and redesign manufacturing steps as volume accumulates.

However, the price report does not disclose usable die counts. It cannot establish whether customers receive better economics after the increase.

The fourth uncertainty is customer concentration. A few large programs can fill a factory, especially when chips are physically large or require substantial production volume.

That concentration can lift near-term utilization while exposing the foundry to sharp changes when one product cycle ends.

Diversified demand would offer stronger evidence. Orders across AI processors, networking silicon, automotive chips, mobile products, and memory logic would reduce dependence on one buyer.

The fifth uncertainty is how customers respond. Some may accept the increase because alternatives are unavailable during the current capacity squeeze.

Acceptance under pressure does not guarantee loyalty. Buyers can begin qualifying TSMC, Intel, or another process for later product generations.

Intel Foundry remains a less established external supplier, but it represents another potential source of advanced capacity. Its progress matters when customers seek geographic and supplier diversity.

Mature-node foundries also compete for less advanced products, though they cannot replace Samsung for every high-performance design. The relevant alternative depends on the chip's requirements.

Export controls add another layer. They affect which technologies, equipment, customers, and destinations can participate in advanced semiconductor trade.

A foundry might receive more inquiries from one region while remaining unable to serve particular designs. Reported customer interest does not always become a manufacturable order.

These constraints make regional price comparisons difficult. A larger increase can reflect intense demand, limited legal alternatives, or more complicated compliance requirements.

Samsung also faces an internal capital allocation question. Its memory business can absorb investment during an AI boom, while foundry expansion requires long-term commitments.

Factories take years to plan and equip. A temporary utilization spike does not automatically justify capacity that must operate through future semiconductor cycles.

The company's updated road map illustrates that caution. Recent reporting indicates Samsung moved its planned 1.4-nanometer introduction to 2029 while concentrating on its 2-nanometer family.

That adjustment can strengthen execution by focusing resources. It can also signal that commercial readiness matters more than maintaining an aggressive sequence of node labels.

The relevant question is not whether Samsung announces the smallest number first. It is whether customers can ship competitive products at dependable volume.

Higher prices can fund that effort, but only if they accompany improving manufacturing performance. They cannot substitute for it.

This skepticism does not make the report unimportant. It defines what the report actually demonstrates.

Samsung appears to have gained near-term leverage on several valuable production lines. The evidence stops short of proving a permanent change in the foundry hierarchy.

Three Signals Will Show Whether Samsung's Leverage Lasts

Customer conversion, sustained manufacturing performance, and competitive pricing will determine whether this increase marks a durable Samsung recovery.

The first signal is Samsung's foundry performance through the next two quarterly updates. Investors should watch management's comments about utilization, advanced-node sales, and external customer demand.

Samsung reports its foundry business alongside System LSI within the Device Solutions segment. That structure limits financial visibility, so operational language will carry unusual importance.

A continued expansion in 4-nanometer LPU and HBM base-die output would support the capacity argument. Broader external orders would provide stronger evidence that demand extends beyond internal products.

Improved earnings would strengthen the claim that higher utilization and rates are repairing the business model. Weak results would suggest the pricing benefit remains too narrow.

The second signal is commercial production on Samsung's second-generation 2-nanometer process. Mobile output will test whether the company can move advanced manufacturing beyond engineering milestones.

Customer announcements matter, but shipped products matter more. A named design entering dependable volume production would validate yield, design support, and delivery execution together.

Samsung must then translate that performance into repeat orders. One successful launch establishes credibility, while multiple product generations build a foundry franchise.

The third signal is the response from TSMC and other manufacturers. Further TSMC price increases would preserve room for Samsung to charge more without losing its relative position.

More available TSMC capacity would weaken that leverage. Customers could return to their preferred supplier or negotiate harder with Samsung.

Intel's progress could also change the balance. Qualified advanced capacity from another supplier would give buyers more negotiating options, particularly for future United States production.

These signals should appear in a specific order. First, Samsung must show that constrained lines improve business performance. Second, it must prove its newer process in volume.

Third, the market must remain tight enough for those gains to persist. Failure at any stage would weaken the idea that July's increase represents a lasting turn.

For chip buyers, the immediate lesson is practical. Capacity planning now belongs near the start of product strategy, not after a design approaches completion.

Teams should compare total usable-chip economics, qualification schedules, packaging access, and delivery commitments. The quoted wafer rate captures only one part of that decision.

They should also preserve the reasoning behind supplier choices, capacity assumptions, and manufacturing risks. A searchable engineering knowledge base can keep those decisions connected to changing technical evidence.

Samsung's reported increase is meaningful because it reverses the company's recent role as the discounted alternative. It shows that AI demand can create leverage even for a distant second-place supplier.

Yet the market hierarchy has not reversed. TSMC still commands the scale, ecosystem, and execution record that shape advanced chip manufacturing.

Watch what customers ship, not only what they reportedly pay. If Samsung converts tight 4-nanometer capacity into repeat external orders and reliable 2-nanometer production, the increase will look structural.

If demand loosens before that happens, July's higher rates will look more like a favorable moment in the cycle. The next few quarters will distinguish those outcomes.