SMIC’s Record $3B Quarter Signals New Foundry Pricing Power

SMIC crossed $3 billion in quarterly revenue for the first time and raised wafer prices as utilization reached 93.7%. That combination signals a sharp change in China’s semiconductor market. The country’s largest foundry now has more demand than it can fully serve, despite restrictions intended to limit its access to advanced technology.

A search for samsung tom may surface Samsung foundry coverage from Tom’s Hardware. However, the more consequential Chinese market story now centers on Semiconductor Manufacturing International Corporation, better known as SMIC. Its record quarter shows how export controls can constrain technology while concentrating domestic demand around the strongest available supplier.

The distinction matters. SMIC has not caught TSMC or Samsung at the most advanced manufacturing processes. It has gained something different: pricing power inside a market where many Chinese customers have fewer realistic alternatives. That advantage could improve revenue and margins, but it also places more pressure on SMIC’s capacity, equipment, and manufacturing yields.

SMIC’s Record Quarter Came With Real Pricing Power

SMIC’s results showed that its factories are no longer competing mainly by offering lower prices.

SMIC reported second-quarter revenue of approximately $3.01 billion, up 20% from the previous quarter and 36.1% from a year earlier. Profit attributable to shareholders reached $479.2 million, nearly three times the comparable figure from 2025.

The profit increase needs context. Chief Financial Officer Wu Junfeng said a one-time gain involving a subsidiary also supported the result. Investors should not assume every dollar of the increase came from stronger manufacturing economics.

The operating figures were still notable. According to the company figures summarized in a quarterly earnings report, SMIC shipped 2.9 million 8-inch-equivalent wafers. Shipments rose about 14% from the first quarter, while average wafer selling prices increased 5.7%.

An 8-inch-equivalent wafer is a standardized measure that lets foundries combine production from different wafer sizes. It does not mean SMIC manufactured every chip on an 8-inch production line.

Monthly capacity increased 1.7% from the preceding quarter to roughly 1.1 million 8-inch-equivalent wafers. Utilization reached 93.7%, compared with 93.1% in the first quarter.

Utilization measures how much installed manufacturing capacity a foundry is actively using. A high rate spreads fixed factory costs across more wafers, but it leaves less room for unexpected orders or production problems.

SMIC’s management plans to keep utilization near 95%. The company wants to reserve the remaining capacity for research, development, maintenance, and the normal variation found in semiconductor manufacturing.

This operating position allowed SMIC to negotiate higher prices. Co-CEO Zhao Haijun said discussions with customers concluded during the first quarter. The new terms will apply more fully to wafers processed and shipped during the third quarter.

“We need to negotiate with customers for fairer pricing,” Zhao told analysts, pointing to a remaining gap between SMIC’s prices and those charged by industry leaders.

That statement reveals the central change. SMIC is no longer describing discounts as the only way to compensate for its technology gap. Management believes strong demand and improved manufacturing standards justify higher charges for scarce production capacity.

The increases were targeted rather than universal. Reports indicate that weaker categories, including some smartphone and display-driver products, did not receive the same treatment. SMIC concentrated the increases where orders were strongest and available capacity was tightest.

That approach reduces the risk of pushing price-sensitive customers away. It also suggests the company is managing several semiconductor cycles at once. Data-center power chips can face shortages while consumer electronics remain comparatively soft.

SMIC expects third-quarter revenue to grow another 2% to 4% sequentially. That guidance indicates management expects higher prices and additional shipments to continue supporting sales after the record quarter.

AI Data Centers Need More Than Advanced GPUs

The immediate shortage is not limited to the processors that train or run AI models.

AI data centers require accelerators, but they also depend on power-management chips, networking components, controllers, sensors, and display interfaces. Many of these supporting chips use established production technologies rather than the newest manufacturing process.

This distinction helps explain SMIC’s results. Zhao said the shipment increase came mainly from AI-related products other than central processing units and graphics processors. Chinese customers also placed some orders earlier than SMIC had expected.

Power-management semiconductors offer a useful example. A data center must convert and distribute electricity across servers, accelerator boards, memory, storage, and cooling equipment. Every increase in computing density creates additional power-control requirements.

Some of those components use a BCD process, which combines bipolar, CMOS, and DMOS transistor technologies on one chip. The design supports precise control functions alongside circuits that handle higher voltages.

Demand for these components can benefit foundries with strong mature-node capacity. Mature nodes are established manufacturing processes used for products that do not require the smallest available transistors.

SMIC’s position therefore cannot be measured only by asking whether its best logic process matches TSMC’s newest node. The relevant question is whether Chinese equipment builders can secure enough of every required component.

Industry reporting indicates SMIC expects demand for some data-center power-management products to remain strong into 2027. The company is adjusting existing production and accelerating new lines to address that demand.

This broader product mix also makes the term “captive AI market” more precise. Chinese customers are not necessarily purchasing every wafer from SMIC because it offers the best global technology. They may choose it because geopolitical restrictions, supply-chain policies, qualification requirements, and limited domestic alternatives narrow their options.

China generated about 90% of SMIC’s second-quarter revenue. The United States accounted for about 8%, with other markets contributing the remainder. That domestic concentration connects SMIC’s growth directly to China’s localization campaign.

Localization means replacing foreign suppliers with domestic alternatives across equipment, materials, chip design, fabrication, packaging, and systems. It has become both an industrial goal and a response to trade restrictions.

China’s AI developers still need advanced accelerators. Yet an accelerator cannot operate alone. A restricted supply of high-end processors can increase pressure to extract more value from available hardware, build domestic substitutes, and secure every supporting component.

Developers and enterprise buyers should care because infrastructure limits eventually shape software decisions. Scarce processors encourage smaller models, quantization, inference optimization, and workloads designed around locally available accelerators.

They also create operational uncertainty. A company planning an AI service must consider whether its hardware supplier can obtain enough processors, power components, and networking chips throughout the deployment.

For teams tracking these overlapping technical and business signals, a searchable engineering knowledge base can preserve filings, specifications, and supplier changes in one place. The need is practical when a single export rule can alter several product roadmaps.

Why the Samsung Tom Comparison Misses SMIC’s Captive Market

Samsung and SMIC are both raising foundry prices, but they are benefiting from different competitive structures.

Recent Samsung coverage has focused on reported increases for its 4-nanometer, 5-nanometer, and 8-nanometer processes. The largest quoted increases reportedly affected customers in China and the United States.

Samsung has a more advanced process portfolio than SMIC. It also competes directly with TSMC for major global customers designing smartphone processors, accelerators, and other high-performance chips.

However, Samsung’s foundry business has struggled with low market share and years of losses. Raising prices can help its economics when TSMC’s capacity is full, but customers still evaluate Samsung against a global set of manufacturing alternatives.

The samsung tom search phrase captures that public discussion, particularly reporting about Samsung pricing and Chinese demand. SMIC’s situation has a different mechanism. Its advantage comes from being the leading domestic foundry within a partially separated technology market.

TSMC remains the dominant global contract manufacturer. It produces many of the leading processors used by Apple, AMD, Nvidia, and other major chip designers.

Samsung combines a contract foundry with large memory and consumer-electronics businesses. That structure gives it internal demand but also forces outside customers to compete with Samsung’s own products for selected production lines.

SMIC operates behind a thicker geopolitical boundary. It is China’s largest contract chip manufacturer and the country’s only foundry reported to mass-produce 7-nanometer-class logic components.

That manufacturing capability does not make SMIC technologically equivalent to TSMC. SMIC lacks access to extreme ultraviolet lithography, or EUV, a production method that simplifies patterning at advanced nodes.

SMIC can use deep ultraviolet lithography with additional patterning steps. This path can produce advanced chips, but added steps increase complexity, manufacturing time, defect risk, and cost.

The result is an unusual competitive position. SMIC trails the global leaders in process technology, yet domestic customers can still compete intensely for its capacity.

U.S. restrictions contribute to this structure in two directions. Controls limit Chinese access to selected advanced chips and semiconductor-manufacturing equipment. They also give Chinese companies a stronger incentive to qualify domestic designs and suppliers.

The Commerce Department rules expanded licensing requirements for certain advanced semiconductors and imposed additional due-diligence obligations on foundries. The stated objective was to prevent restricted Chinese entities from obtaining high-performance technology.

Those controls create real constraints. They make it harder for SMIC to acquire, maintain, and expand production using equipment containing controlled American technology.

However, restrictions can also redirect orders. A Chinese designer that cannot reliably use TSMC for an advanced product has a stronger reason to work with SMIC, even when that choice involves lower performance or higher manufacturing difficulty.

That is the reversal behind the quarter. Export controls can slow SMIC’s technical progress while increasing the commercial value of the capacity it already possesses.

The dynamic does not mean sanctions have failed. National-security policy has objectives beyond a company’s quarterly revenue, including limiting access to the most capable computing systems.

It does mean revenue alone is a poor measure of technological containment. SMIC can remain generations behind the leading edge and still gain pricing power in selected domestic categories.

The Samsung comparison therefore has limits. Samsung benefits when global capacity is tight and customers seek a second source. SMIC benefits when Chinese customers have fewer cross-border sources available at all.

The Numbers Do Not Erase SMIC’s Manufacturing Limits

A full factory is evidence of strong demand, not proof that SMIC has solved advanced chip production.

The first uncertainty involves product mix. SMIC does not disclose enough node-level information to show exactly how much revenue came from advanced logic, power-management products, display drivers, or other established processes.

Management connected the quarter to AI demand, but that label covers many different chips. A mature-node power controller does not carry the same manufacturing challenge or economic value as an advanced accelerator.

The second uncertainty concerns yields. Yield is the percentage of usable chips obtained from each processed wafer. Poor yield raises the effective cost of every working processor, even when a fabrication line appears busy.

Outside analysts have repeatedly identified yield as an issue for SMIC’s most advanced processes. The company’s inability to use EUV equipment creates more manufacturing steps and more opportunities for defects.

SMIC did not provide enough public detail in its quarterly results to establish that advanced-node yields now match those of TSMC or Samsung. Claims that the company has reached top industry standards should therefore be read in the specific context of the products receiving higher prices.

A customer can accept a price increase because every competing supplier is full. That does not necessarily mean SMIC has closed its technology gap.

The third uncertainty is equipment access. Semiconductor factories require continuous maintenance, spare parts, process chemicals, measurement systems, and software updates. Export restrictions can affect each layer.

SMIC added 8,000 wafers of monthly 12-inch capacity during the second quarter. It is also considering additional equipment for existing factories that still have usable floor space.

Installing equipment is only one part of a capacity expansion. The company must qualify tools, stabilize processes, train staff, reach acceptable yields, and obtain customer approval for manufactured products.

These requirements limit how quickly SMIC can answer unexpected demand. Zhao acknowledged that incoming wafer volume had exceeded earlier forecasts, and management is adjusting expansion plans.

High utilization creates another risk. A factory running near its practical ceiling has less flexibility when a customer changes a design, a production tool fails, or demand shifts between process technologies.

SMIC wants to maintain a buffer of roughly 5% for development work. That discipline can protect future products, but it also caps the volume available for immediate customer orders.

Financial costs deserve equal attention. First-half capital spending reached approximately $3.4 billion. The company expects full-year amortization of around $5 billion, about 30% above the previous year.

Amortization and depreciation allocate factory and equipment costs across their useful lives. Rising charges can pressure margins even when revenue grows.

SMIC’s stronger average selling price helps offset that burden. However, a sustained margin improvement depends on utilization, product mix, yields, and pricing moving in the right direction together.

The reported profit figure also benefited from a one-time item. Future comparisons should separate recurring manufacturing income from gains that will not repeat.

Consumer demand creates a final complication. Smartphone chips and display drivers remain important parts of SMIC’s business, but some of those markets were too weak to support equivalent price increases.

If data-center demand stays firm while consumer electronics remain soft, SMIC can reallocate selected capacity. Manufacturing processes are not always interchangeable, however. A line qualified for one product cannot instantly produce another.

This means the record quarter should not be interpreted as universal scarcity across SMIC’s portfolio. The shortage is concentrated, and its financial value depends on where constrained production overlaps with customer demand.

U.S. Controls Have Produced a Commercial Reversal

The policy pressure that restricts SMIC’s technology has also strengthened its position with domestic customers.

U.S. export controls seek to limit China’s ability to obtain advanced computing chips and the equipment needed to manufacture them. The restrictions include licensing requirements, entity listings, end-user checks, and controls reaching some foreign-produced items.

SMIC has appeared on the U.S. Entity List since 2020. Suppliers dealing with listed entities face licensing rules that can delay or prevent transactions involving controlled technology.

Additional controls introduced since 2022 have targeted advanced computing products, semiconductor equipment, software, and high-bandwidth memory. HBM is specialized memory that supplies the data throughput required by many AI accelerators.

These measures create substantial technical friction. SMIC cannot simply purchase every system available to TSMC or Samsung, build an identical fabrication line, and compete on equal terms.

The commercial effect is less straightforward. Chinese chip designers once had stronger incentives to seek global foundries when those suppliers offered better processes, yields, or costs.

Export uncertainty changes that calculation. A design made for TSMC becomes commercially risky if new licensing rules can interrupt manufacturing after development has started.

Chip design projects take time and require close coordination with a foundry’s process design kit. A process design kit contains the rules, models, and technical files needed to prepare a chip for manufacturing.

Moving a design to another foundry is not like changing a cloud-service setting. Engineers may need to redesign physical layouts, validate intellectual property blocks, repeat testing, and qualify a new supply chain.

Chinese companies therefore have a reason to establish long-term relationships with SMIC. The domestic foundry may offer a less advanced node, but it can appear more predictable from a policy perspective.

That preference supports utilization and gives SMIC leverage in negotiations. Customers are paying not only for transistor density. They are paying for access to production that fits China’s increasingly localized supply chain.

The process can become self-reinforcing. More customer orders provide SMIC with revenue and manufacturing experience. Higher utilization spreads fixed costs, while greater volume generates additional process data.

More experience does not automatically remove equipment constraints. It can still help engineers improve mature products, stabilize production, and refine process controls.

Policy also encourages Chinese equipment and material suppliers to qualify their products at domestic fabs. Progress at those suppliers can reduce selected dependencies over time, even when the most advanced tools remain unavailable.

This is why the phrase “sanctions handed SMIC a captive market” needs careful handling. U.S. policy did not create Chinese semiconductor demand, and it did not remove every competitor.

Domestic industrial policy, AI investment, supply-chain localization, and expanding data centers were already increasing demand. Export controls narrowed the routes through which that demand could be served.

Hua Hong provides evidence that the effect reaches beyond one company. The smaller Chinese foundry reported quarterly revenue of $717.5 million, up 26.8% year over year. Its net profit rose sharply from a much smaller base.

The Chinese foundry results show domestic fabrication plants operating at high intensity as customers seek locally manufactured chips.

Yet SMIC captures the greatest strategic benefit because it owns China’s most advanced domestic logic capability. That position directs customers toward its production when their designs require more performance than other local foundries can offer.

The reversal remains bounded. SMIC receives more domestic orders, but it also bears the cost of building capacity without unrestricted access to the best equipment.

Customers receive a more secure local supply route, but they may accept lower efficiency or more expensive chips. Chinese AI developers gain domestic accelerators, yet those products still compete against systems built with more advanced processes.

Export controls have therefore created a trade between access and capability. SMIC’s record quarter reflects the value of access inside China, not the disappearance of the capability gap.

Three Signals Will Show Whether SMIC Can Hold the Advantage

The next quarter must show that pricing, capacity, and AI demand can reinforce one another without exposing SMIC’s technical limits.

The first signal is SMIC’s third-quarter average selling price. Management said wafers processed during the quarter would reflect pricing negotiations completed earlier in 2026.

Another sequential increase would confirm that customers accepted the new terms rather than merely accelerating orders before they took effect. A flat or declining figure would weaken the pricing-power argument.

Investors should examine gross margin alongside average selling price. A higher wafer price matters less if rising depreciation, product complexity, or weak yields absorb the improvement.

The second signal is utilization after capacity additions. SMIC expects to keep production near 95% while accelerating new lines and adjusting existing facilities.

Utilization remaining above 90% as capacity grows would support management’s claim that demand exceeds available production. A sharp decline would suggest customers pulled orders forward or that shortages were limited to particular processes.

The composition of capacity also matters. Added mature-node production will not solve a shortage on a more advanced line. SMIC needs to show that equipment installation matches the categories generating the strongest orders.

The third signal is any change in export policy or enforcement. New controls on equipment, spare parts, software, memory, or foreign-produced chips can alter SMIC’s expansion schedule.

The reverse also matters. A licensing change that gives Chinese designers more access to overseas foundries could increase their bargaining power and reduce SMIC’s captive-market advantage.

Policy announcements should be evaluated against actual supplier behavior. A rule can appear narrow while creating broader caution among equipment companies, chip designers, and contract manufacturers.

Samsung and TSMC provide an external benchmark. If their capacity remains tight and their prices continue rising, SMIC has more room to increase charges while remaining cheaper.

If global foundry demand weakens, Chinese customers may still prefer SMIC for supply security. They would nevertheless gain more leverage during contract negotiations.

The samsung tom comparison will therefore remain useful only when it distinguishes global scarcity from China-specific restrictions. Samsung’s pricing reflects competition for advanced capacity across several markets. SMIC’s pricing also reflects the limited routes available to Chinese customers.

For developers and enterprise technology buyers, the practical question is not whether SMIC has defeated TSMC or Samsung. It has not.

The important question is whether China can assemble enough locally manufactured components to keep expanding AI infrastructure despite reduced access to leading global supply chains. SMIC’s quarter says that domestic demand is strong enough to fill the available factories.

It does not yet show that those factories can scale advanced production efficiently. The next results must connect higher prices with recurring profit, stable yields, and usable new capacity.

Watch the average selling price first, utilization second, and export-control changes third. Together, those signals will reveal whether SMIC’s record quarter marked durable pricing power or the temporary peak of a constrained market.