Semiconductor Manufacturing International Corporation (0981.HK) - Deep Dive

Research report. Prepared 2026-06-14. All figures in USD unless noted; share-trade values in HKD. No valuation, no price targets.

Reporting calendar used. SMIC reports quarterly. The five most recent reporting periods as of 2026-06-14 are: Q1 2026 (call May 15, 2026), Q4 2025 (Feb 10, 2026), Q3 2025 (Nov 13, 2025), Q2 2025 (Aug 7, 2025), Q1 2025 (May 9, 2025). The most recent call is within 90 days of today, as required.

Section 1: What the Company Does

SMIC is a contract chip manufacturer, a "foundry." It does not design chips and it does not sell chips under its own brand. Other companies design a chip, hand SMIC the design files, and SMIC turns those files into physical silicon wafers covered in finished circuits. The customer takes those wafers, cuts them into individual chips, and sells them. SMIC charges per wafer. That is the entire business: renting out some of the most expensive and difficult-to-operate factories on earth to companies that cannot or do not want to build their own.

What makes SMIC distinct from the other large foundries is geography and politics. SMIC is China's largest foundry and, in 2025, the third-largest foundry in the world by revenue, behind TSMC and Samsung, with roughly $9.33 billion in revenue and about a 5.3% share of the global foundry market (Taipei Times, 2026-03-14). It is the national champion of a country that is trying, under heavy US export controls, to build a domestic semiconductor supply chain that does not depend on the West. That single fact shapes almost everything about the company: who owns it, who its customers are, what equipment it can buy, what it can make, and why it is valued the way it is.

The technical core of the business is hard to overstate. A modern fab is a building where air is thousands of times cleaner than a hospital operating room, where a wafer of silicon passes through hundreds of process steps over weeks - deposition, lithography, etching, ion implantation, polishing - each repeated dozens of times to build up a chip layer by layer. A single misaligned step ruins the wafer. The machines that print the circuit patterns (lithography scanners from ASML, Applied Materials, Lam, Tokyo Electron) cost tens of millions of dollars each, and the most advanced ones - EUV lithography - are banned from sale to SMIC under US-led export controls. The deepest part of SMIC's story is that it has been forced to manufacture relatively advanced chips (down to a 7nm-class node) using only the older "DUV" lithography that it is allowed to buy, by running each wafer through the machine multiple times ("multi-patterning"). This is slower, more expensive, and lower-yielding than EUV, but it works.

A concrete example of what SMIC actually does: Huawei wanted a domestic 5G smartphone processor after US sanctions cut it off from TSMC. Huawei's chip-design arm, HiSilicon, designed the Kirin 9000s. SMIC manufactured it on its second-generation 7nm process, internally called "N+2," without EUV. TechInsights tore down the resulting Huawei Mate 60 Pro in 2023 and confirmed the chip was made on SMIC's 7nm node - the first commercial 7nm-class logic chip ever produced without EUV (TechInsights). By 2025-2026 the same partnership had moved to producing Huawei's Ascend AI accelerators and newer Kirin parts. That is SMIC's role in one sentence: it is the factory that lets China's chip designers ship products the rest of the world assumed they could not build.

SMIC is not formally a state-owned enterprise, but its largest shareholders are tied to the Chinese state, party, and military apparatus: Datang Holdings (a subsidiary of state-owned Datang Telecom), Xinxin (Hong Kong), and the National Integrated Circuit Industry Investment Fund, known as the "Big Fund" (The Wire China). Understanding SMIC means understanding that its mission - chip self-sufficiency for China - is partly a national-policy objective, not just a profit objective.

Section 2: Business Segments

SMIC is, at the reporting level, a single-business company: it is a pure-play wafer foundry. In Q1 2026 wafer revenue was 93.9% of total revenue, with the small remainder coming from mask-making, testing, and other services (GuruFocus / Q1 2026 call). There is no separate "products" division, no design business, no consumer arm. So rather than reportable segments, the meaningful way to understand SMIC's mix is along three axes that management actually discusses: by application end-market, by process node, and by geography. Each functions like a de facto segment with its own competitive dynamic.

By application end-market

SMIC's wafers go into five broad application buckets that management tracks each quarter: smartphones, computers and tablets, consumer electronics, connectivity and IoT (often called "wearables/connectivity"), and industrial and automotive.

Smartphones are historically SMIC's single largest application bucket. This is where the strategically important Huawei work sits (Kirin SoCs, RF, power management, display driver ICs). It is also the most cyclical and price-competitive bucket.
Consumer electronics and computers/tablets absorb large volumes of mature-node parts: display drivers, image sensors, power management, microcontrollers. These are commodity-like and exposed to inventory cycles.
Industrial and automotive is the bucket management has flagged as the growth and margin priority. On the Q1 2025 call (May 9, 2025), management noted industrial and automotive revenue grew more than 20% sequentially, lifting its share of total revenue from 8% to 10%. By Q1 2026 the company was emphasizing demand for its automotive-grade BCD (Bipolar-CMOS-DMOS) power platform and for power-management and data-transmission chips tied to the AI build-out (Q1 2026 call, May 15, 2026).

This bucket exists as a strategic priority because it carries higher switching costs (automotive parts require years of qualification and have long lifecycles) and is less exposed to consumer inventory swings. Management talks about it as the quality-of-earnings improver.

By process node (the de facto "two businesses")

The cleanest way to see SMIC as two businesses is by node.

Mature and mainstream nodes (28nm and above). This is the overwhelming majority of SMIC's commercial volume - 350nm down through 28nm, covering analog, power management, RF, image sensors, microcontrollers, and display drivers (SemiWiki / SEMI roadmap discussion). This is where SMIC's massive capacity expansion is aimed (the Lingang, Shenzhen, Tianjin, and Beijing 12-inch fabs targeting 28nm-and-above). The core capability here is scale and cost, and the competition is fierce: UMC, GlobalFoundries, Hua Hong, Nexchip, and others. This is a commodity-leaning business where China's capacity build risks oversupply and price erosion.

Advanced nodes (14nm and below, including 7nm "N+2/N+3"). This is small in revenue but enormous in strategic value. It is the only place outside Taiwan, Korea, and the US where chips at this class are made, and SMIC has a near-monopoly on it inside China. SemiAnalysis estimated SMIC's advanced-node (7nm-and-below) capacity at roughly 45,000 wafers per month at end-2025, rising toward 60,000 in 2026 and 80,000 in 2027 (SemiAnalysis, 2025-09-08). This node serves Huawei's Kirin and Ascend chips and other domestic AI/HPC designs. The capability - running a 7nm-class flow on DUV with multi-patterning, at usable yields - took years to build and cannot easily be replicated by any other Chinese foundry. Beijing reportedly rations this scarce advanced capacity toward priority customers like Huawei.

By geography

SMIC's revenue is overwhelmingly Chinese. In Q1 2026 the split was China 89%, the Americas 9%, and the rest of Asia ("New Asia") 2% (Q1 2026 call). The China share has risen steadily as US customers de-risked away from SMIC after sanctions and as domestic Chinese fabless designers grew. This concentration is both a strength (insulated demand, policy support) and a risk (single-economy exposure).

"Segment" axis	What it is	Key end markets	Competitive edge	Strategic priority
Mature nodes (≥28nm)	Bulk of wafer volume	Consumer, smartphone, IoT, industrial	Scale, cost, domestic supply security	Cash/volume engine; capacity expansion
Advanced nodes (≤14nm, 7nm)	Small revenue, high strategic value	AI accelerators, flagship smartphone SoCs	Only China-domestic 7nm-class capability	Strategic option / national priority
Industrial & automotive	Cross-node, growth bucket	Auto power (BCD), MCUs, sensors	Long qualification, sticky lifecycles	Margin/quality-of-earnings driver

Section 3: Products and Business Detail

SMIC's "products" are process technologies, sold as wafer-fabrication services. The catalogue spans roughly 350nm to 7nm.

The mature-node catalogue (the volume business).

180nm-350nm: analog, power, high-voltage, some MCUs and specialty.
130nm-90nm-65nm/55nm: microcontrollers, RFID, power management, embedded flash, mixed-signal.
40nm: embedded NOR flash MCUs, RF, some image sensors and DDIC.
28nm (PolySiON and HKMG variants): the workhorse advanced-mature node - display driver ICs (DDIC), application processors, networking, image signal processors. This is the single most important node for the capacity build-out, with the Lingang/Shanghai (about $8.87B), Tianjin (about $7.5B), Beijing, and Shenzhen 12-inch fabs all centered on 28nm-and-above (Sourcengine). Lingang alone targets 100,000 300mm wafers per month.

The advanced-node catalogue (the strategic business).

14nm FinFET: SMIC's first FinFET node, in production since 2019.
"N+1" (roughly 8nm-class) and "N+2" (7nm-class): the EUV-free advanced nodes. N+2 powered the Kirin 9000s in the Huawei Mate 60 Pro (2023). Later Kirin parts (e.g., the Kirin X90 in the Huawei MateBook Fold) and Ascend AI accelerators run on these flows (TechInsights).
5nm in development: SMIC has reportedly entered pilot runs on a 5nm-class process targeting Huawei and other domestic AI customers, with early yields modest (commentary around 20%) (SemiAnalysis / industry reporting).

Manufacturing and constraints. SMIC runs both 8-inch (200mm) and 12-inch (300mm) fabs across Shanghai, Beijing, Tianjin, Shenzhen, and Lingang. The defining constraint is equipment access: US, Dutch, and Japanese export controls block EUV lithography and the most advanced DUV tools and deposition/etch equipment. SMIC therefore manufactures its most advanced chips using DUV immersion scanners with multi-patterning - printing a pattern, etching, and re-printing offset patterns to achieve features finer than a single exposure allows. This raises cost, lengthens cycle time, and caps yield and throughput at the leading edge. Capacity, not design, is the binding limit on China's most advanced chips.

Capacity milestones. By the end of Q3 2025 SMIC reported monthly capacity had reached the 1 million wafers level (8-inch equivalent) with utilization of 95.8% (Q3 2025 call, Nov 13, 2025). A landmark corporate event came in 2025-2026: SMIC agreed to buy out the 49% minority stake in its key Beijing 12-inch subsidiary, Semiconductor Manufacturing North China (SMNC), for about RMB 40.6 billion (roughly $5.9 billion), issuing 547.2 million A-shares to five sellers including the Big Fund. The deal cleared its final Shanghai Stock Exchange hurdle in May 2026 and is the largest wafer-foundry M&A in Chinese history; SMNC (65nm-24nm, ~95.8% utilization) becomes wholly owned (TechNode, 2026-05-12; Caixin).

Section 4: Customers

SMIC's customers are fabless chip designers and integrated device makers, overwhelmingly Chinese. The buyer is a chip company's operations/supply-chain leadership together with its process-integration engineers, who decide which foundry and which node to use based on a multi-criteria evaluation: does the foundry have a qualified process for this chip, what is the yield and therefore cost per good die, what is the capacity allocation and lead time, and - uniquely for SMIC - is the foundry safe from a geopolitical supply-chain standpoint.

Who buys and why. Three customer types dominate:

Strategically critical domestic champions - above all Huawei/HiSilicon. They buy from SMIC because, post-sanctions, they have no alternative: TSMC and Samsung are off-limits to them. This is the stickiest relationship in the company, reinforced by state capacity allocation.
Chinese fabless designers across consumer, IoT, power, and display (display-driver makers, MCU vendors, power-IC houses, CIS designers). They choose SMIC for domestic supply security, local support, competitive mature-node pricing, and increasingly because their own customers (Chinese OEMs) prefer a China-made supply chain.
A shrinking set of international customers (the ~9% Americas, ~2% rest-of-Asia revenue) buying mature-node commodity capacity on price.

Switching costs. These are high once a chip is designed into a process. A chip is co-developed with a specific foundry's design rules and process design kit; porting it to another foundry requires a re-spin, re-qualification, and often months of work and millions of dollars. In automotive the qualification burden (AEC-Q100, long reliability testing) makes switching even harder and is a key reason SMIC prioritizes its automotive BCD platform. At the advanced node the switching cost is effectively infinite for Chinese designers: there is no other domestic 7nm-class option.

Concentration. Customer concentration is real but partly masked. The Huawei relationship is the most important single dependency, and China is 89% of revenue, so SMIC is effectively a leveraged bet on the health of China's domestic electronics and AI ecosystem. Management frames this concentration as the deliberate result of becoming the national supply-security backbone, not as a vulnerability - but it is both.

Contract structure. Foundry revenue is a mix of longer-term capacity commitments (especially for strategic customers and automotive, where allocation is planned well ahead) and shorter-cycle volume orders that flex with the consumer cycle. The blended ASP moves quarter to quarter with mix and pricing - it fell 6.4% sequentially in Q2 2025, then management pointed to AI-driven price increases lifting Q2 2026 (Q2 2025 call; Q1 2026 call). Revenue is therefore moderately cyclical rather than contractually locked.

Section 5: Competitive Landscape

The foundry industry is extraordinarily concentrated at the top and fragmented below it. TSMC took roughly 70% of all foundry revenue in 2025; everyone else fights over the remaining ~30% (Taipei Times, 2026-03-14). SMIC competes in two different arenas with two different competitive realities.

At the leading edge (7nm and below): SMIC's only true global peers are TSMC, Samsung, and Intel Foundry - and SMIC cannot compete with them on a level field because it is barred from EUV. TSMC is at 2nm-class production while SMIC's most advanced is 7nm-class with 5nm in pilot. So at the frontier SMIC does not win on technology; it wins on the fact that it is the only option for Chinese customers who are locked out of the leaders. Its "moat" at the advanced node is a regulatory wall, not a technology lead - a moat that exists precisely because the technology gap is large.

At mature nodes (28nm and above): here SMIC competes head-to-head with UMC, GlobalFoundries, Hua Hong, Nexchip, and Tower. This arena is becoming structurally tougher because China (SMIC and Hua Hong included) is adding enormous mature-node capacity, which analysts widely expect to drive oversupply and price competition (EE Times; SemiWiki). SMIC's edge here is cost, scale, domestic demand, and the "China-for-China" supply-security premium; its exposure is that mature nodes are commoditizing and its own country's capacity build is part of what is compressing prices.

Barriers to entry are among the highest of any industry: tens of billions in capex per fab, decades of accumulated process know-how, scarce equipment, and a workforce of specialized engineers. For SMIC specifically there is an additional asymmetry - it is shielded from international competition inside China by the same geopolitics that shut it out of leading-edge tools abroad.

Competitor	Country	Listing	Approx Market Cap	Product Overlap	Relative Strength vs SMIC
TSMC	Taiwan	NYSE: TSM / TWSE: 2330	~$2.2T (Jun 2026)	Full overlap, leads at every node	Vastly ahead on technology, yield, scale; SMIC wins only on China access
Samsung Foundry	South Korea	KRX: 005930 (division of Samsung Electronics, ~$400B+ group)	Group >$400B (2026)	Advanced + mature	Ahead at leading edge; weaker yields than TSMC
GlobalFoundries	USA	Nasdaq: GFS	~$42B (Jun 2026)	Mature/specialty only (no <12nm)	Specialty/US-EU focus; SMIC has scale + cost edge in China
UMC	Taiwan	NYSE: UMC / TWSE: 2303	~$22B (Apr 2026)	Mature nodes, moving to 12nm via Intel	Comparable mature-node peer; broader non-China customer base
Hua Hong Semiconductor	China	HKEX: 1347 / SSE: 688347	~$20B / HK$156B (late 2025)	Mature/specialty in China	SMIC larger and more advanced; both add China capacity
Nexchip / others	China	SSE-listed / private	—	Mature consumer (DDIC, CIS)	Aggressive on price; intensify domestic oversupply

The honest read: SMIC has a powerful structural moat in China driven by geopolitics, and almost no moat against the global leaders on the open market. Outside the protected China-for-China demand, its mature-node business is exposed to genuine commoditization and oversupply.

Section 6: Industry

Foundry demand is driven by the spread of chips into everything - smartphones, PCs, cars, industrial equipment, and now, decisively, AI infrastructure. The global foundry market was roughly $113 billion in 2025 and is growing at a high-single to low-double-digit pace over the cycle, though the growth is wildly uneven: leading-edge AI logic is booming while parts of the mature-node consumer market remain soft (Taipei Times; industry market reports).

SMIC sits in a specific and politically charged spot in the global supply chain: it is the most advanced foundry in China and the linchpin of China's import-substitution drive. China imports hundreds of billions of dollars of chips annually and is under national policy to replace as much of that as possible with domestic production. SMIC is the primary vehicle for that substitution at logic. The import-substitution dynamic is the single biggest secular tailwind behind SMIC's volume growth: every Chinese OEM that switches from a foreign chip to a China-made one is, increasingly, a chip SMIC or Hua Hong made.

The regulatory environment is the defining industry feature for SMIC. US export controls (Entity List placement in 2020, tightened repeatedly since) restrict its access to EUV and advanced DUV tools and to some US customers. On the other side, Chinese industrial policy (the Big Fund, provincial subsidies, state capacity allocation) pours capital in. SMIC lives at the intersection of these two forces.

Cyclicality is intense. Foundry demand swings with consumer-electronics inventory cycles and capex cycles; ASPs and utilization can move sharply within a year, as SMIC's own 2025 trajectory showed (utilization in the mid-90s, ASP down 6.4% in one quarter, then AI-driven price increases the next year). The current cross-currents: AI demand and pricing power at the leading edge and in power/data-transmission chips (tailwind) versus mature-node overcapacity from China's own build-out (headwind).

Section 7: Growth Triggers

Extracted only from the five most recent concalls, each cited.

AI-driven Q2 2026 surge. Management guided Q2 2026 revenue up 14-16% sequentially with gross margin expanding to 20-22%, attributing it to strong AI-driven demand and price increases (Q1 2026 call, May 15, 2026).

"Strong AI-driven demand and price increases are expected to boost Q2 revenue by 14%-16%." (Q1 2026 call, May 15, 2026)
Power-management and data-transmission chips tied to AI. Management flagged that the AI boom has increased demand for power-management and data-transmission chips, alongside strong demand for its automotive-grade BCD platform (Q1 2026 call, May 15, 2026).
Industrial and automotive ramp. Industrial and automotive revenue grew more than 20% sequentially in Q1 2025, lifting its share from 8% to 10% of revenue, with management positioning the automotive-grade BCD platform as a continuing driver (Q1 2025 call, May 9, 2025; repeated Q1 2026 call). Repeated theme across calls.
Capacity ramp to and beyond 1 million wafers/month. SMIC reached the 1 million wafers/month level (8-inch equivalent) by end-Q3 2025 at 95.8% utilization, the new 12-inch fabs (Lingang, Tianjin, Beijing, Shenzhen) continuing to fill (Q3 2025 call, Nov 13, 2025). Repeated theme - new capacity coming online underpins multi-year volume growth.
Full-year 2026 revenue growth above industry average. Management guided 2026 revenue growth to exceed the industry average, with capex held roughly flat to 2025 - i.e., growth without a fresh capex step-up (Q4 2025 call, Feb 10, 2026).
SMNC full ownership. The $5.9B buyout of the 49% SMNC minority (cleared May 2026) consolidates 100% of a profitable, high-utilization Beijing 12-inch subsidiary into SMIC, simplifying capacity control (TechNode, 2026-05-12). Disclosed via filings around the Q4 2025 / Q1 2026 period.

Trigger	Timeline	Concall source	Status
AI-driven 14-16% Q2 revenue jump	Q2 2026	Q1 2026 (May 15, 2026)	New
AI power-mgmt / data-transmission demand	2026	Q1 2026 (May 15, 2026)	New
Industrial & automotive (BCD) ramp	Ongoing	Q1 2025, Q1 2026	Repeated
Capacity >1M wafers/month, new 12" fabs	2025-2027	Q3 2025 + prior	Repeated
FY2026 revenue > industry average	FY2026	Q4 2025 (Feb 10, 2026)	New
SMNC 100% consolidation	Closing 2026	Q4 2025 / Q1 2026 filings	New

Section 8: Key Risks

Export-control escalation (high impact, ongoing). SMIC's access to lithography, deposition, and etch equipment, spare parts, and some software is restricted by US, Dutch, and Japanese controls. A tightening - for example, a cutoff of DUV tool servicing or consumables - could directly cap its ability to expand or even maintain advanced-node output. This is the single most existential risk because it attacks the means of production, not just demand. It is the reason SMIC must use multi-patterning instead of EUV at all.

Mature-node oversupply and price erosion (high probability, moderate-to-high impact). China's own foundries, SMIC included, are adding huge mature-node capacity. Analysts broadly expect this to drive overcapacity and price competition at 28nm and above (EE Times). Because mature nodes are the bulk of SMIC's revenue, sustained ASP pressure would squeeze the margins that depreciation from the new fabs is already weighing on. SMIC's own ASP fell 6.4% sequentially in Q2 2025, a live illustration of the mechanism (Q2 2025 call).

Depreciation drag from the capex super-cycle (high probability, moderate impact). SMIC is spending tens of billions on fabs whose depreciation hits the P&L for years whether or not the capacity is fully utilized at good prices. Management repeatedly cited depreciation as a margin headwind in the 2025 guidance (gross margin guided to 18-20% through most of 2025). If utilization or pricing disappoints while depreciation steps up, margins compress.

Customer and geographic concentration (structural). With China at ~89% of revenue and the Huawei relationship strategically central, SMIC is a leveraged bet on the domestic Chinese economy and electronics ecosystem. A sharp China consumer-electronics downturn, or any disruption to the Huawei relationship, would hit hard with little geographic offset.

Demand visibility and tariff/geopolitical shocks (episodic). On the Q1 2025 call (May 9, 2025) management itself flagged limited second-half visibility due to tariff and end-market uncertainty. Trade-policy shocks can pull in or push out orders abruptly, making near-term planning hard.

State priorities over shareholder returns (structural). Because SMIC's controlling owners are state-linked and its mission is national supply security, capital allocation favors capacity and self-sufficiency over dividends or buybacks. The 2025 dividend was skipped to fund 2026 capex (BigGo Finance). Minority shareholders are along for a national-policy ride, not a capital-return story.

Section 9: Walk the Talk

The five concalls used: Q1 2025 (May 9, 2025), Q2 2025 (Aug 7, 2025), Q3 2025 (Nov 13, 2025), Q4 2025 (Feb 10, 2026), Q1 2026 (May 15, 2026). The most recent is within 90 days of today.

Start with Q1 2025 (May 9, 2025). Revenue came in at $2,247M, up 1.8% sequentially, gross margin flat at 22.5%, and management guided Q2 2025 down 4-6% with gross margin of 18-20%, explicitly warning that ASP pressure and depreciation would weigh and that second-half visibility was limited by tariffs and end-market uncertainty. This was a notably cautious guide.

What actually happened in Q2 2025 (Aug 7, 2025): revenue was $2,209M, down 1.7% sequentially - better than the guided 4-6% decline - and gross margin was 20.4%, comfortably above the 18-20% guide. Blended ASP did fall 6.4% as warned, but shipments rose 4.3%, cushioning revenue. So the cautious Q1 guide proved conservative: SMIC beat its own revenue and margin guidance. Management then guided Q3 up 5-7% with margin again at 18-20%.

Q3 2025 (Nov 13, 2025) beat again: revenue of $2,382M was up 7.8% sequentially (above the 5-7% guide), and gross margin of 22% was well above the 18-20% guide. Capacity hit the 1 million wafers/month milestone at 95.8% utilization. Management guided Q4 flat to +2%, full-year revenue above $9B, margin back to 18-20%.

"Q4 revenue is guided flat to +2%, with full-year revenue expected to surpass $9B." (Q3 2025 call, Nov 13, 2025)

Q4 2025 (Feb 10, 2026) delivered: revenue of $2,489M was up 4.5% sequentially, beating the flat-to-+2% guide, and gross margin of 19.2% landed within the 18-20% guide. Full-year 2025 revenue was a record $9.33B, up 16.2%, with gross margin up 3 points to 21% - the full-year revenue cleared the ">$9B" promise. Management then guided Q1 2026 to roughly flat revenue with margin of 18-20%, and full-year 2026 revenue growth above the industry average with flat capex.

Q1 2026 (May 15, 2026) was in line to better: revenue of $2,505M was up 0.7% (the "roughly flat" guide), and gross margin of 20.1% beat to the top of the 18-20% guide. Management then issued an unusually strong Q2 2026 guide of +14-16% revenue and 20-22% margin on AI demand and price increases.

The pattern across five quarters is consistent and clear: management guides gross margin conservatively (almost always 18-20%) and then prints at or above the top of the range (20.4%, 22%, 19.2%, 20.1%), and guides revenue cautiously and then meets or beats it. Across the entire stretch there is no example of SMIC missing its own guidance to the downside. This is a management team that under-promises and over-delivers on the near-term numbers, which is the credibility profile you want. The one caveat: the conservatism is partly a function of genuine uncertainty (tariffs, ASP, depreciation timing) that they flag honestly rather than spin, and the bigger strategic promises (capacity ramps, self-sufficiency) play out over years and are harder to score quarter to quarter. On the trackable commitments, though, they walk the talk.

Guided	When	Outcome
Q2 2025: rev -4% to -6%, GM 18-20%	Q1 2025	Rev -1.7% (beat), GM 20.4% (beat)
Q3 2025: rev +5% to +7%, GM 18-20%	Q2 2025	Rev +7.8% (beat), GM 22% (beat)
Q4 2025: rev flat to +2%, FY >$9B, GM 18-20%	Q3 2025	Rev +4.5% (beat), FY $9.33B (met), GM 19.2% (in range)
Q1 2026: rev ~flat, GM 18-20%	Q4 2025	Rev +0.7% (met), GM 20.1% (beat to top)
Q2 2026: rev +14-16%, GM 20-22%	Q1 2026	Pending

Section 10: Shareholder Friendliness Index

Dividends. SMIC (the foundry; not to be confused with the Philippine conglomerate of the same ticker abbreviation) has historically paid no ordinary dividend, and it explicitly skipped any 2025 distribution despite over 30% net-profit growth, citing the need to fund heavy 2026 capacity expansion (BigGo Finance, on 2025 results; TipRanks AGM note). There is no DPS trend to report because there is effectively no dividend; the company reinvests substantially all earnings into fabs and R&D. Payout ratio is therefore near zero by design.

Buybacks and dilution. MoatMap's buyback feed recorded zero SMIC buybacks in the trailing ~90 days (since 2026-03-16), and external searches across annual-report capital-management notes, HKEX announcements, and financial news surface no buyback program over the last three years either; SMIC is not a repurchaser of its own stock. The share count is moving the other way: rather than retiring shares, SMIC is issuing them. The SMNC acquisition alone involves issuing 547.2 million new A-shares to the five selling minority holders (including the Big Fund) to fund the buyout (TechNode, 2026-05-12). So over the last three years the trajectory is share creation and capital retention, not capital return.

Verdict: Hoards Capital - SMIC pays no dividend, runs no buyback, and is net-issuing shares to fund expansion, because its controlling state-linked owners prioritize capacity and national self-sufficiency over returns to minority shareholders.

Section 11: Insider Activities

Source rule: Hong Kong's HKEX Disclosure of Interests portal is gated, so per the instructions the MoatMap cross-market disclosure database is the canonical source for recent insider transactions here. One material transaction was recorded in the trailing 12 months, cross-checked against HKEX-reported news.

Recent transactions (most recent first):

Date	Insider (Name & Role)	Type	Shares	Approx Value	Notes
2026-05-26	China Integrated Circuit Industry Investment Fund ("Big Fund"), Substantial Shareholder	Open-market sell	7,000,000	HK$606.83M (HK$86.69/sh)	~0.12% of O/S; phased trim under a previously disclosed plan during a share-price rally (HKEX DI, 2026-05-26)

Reading the sell. This is the state-backed Big Fund (国家集成电路产业投资基金) trimming a sliver of its large position, not an operating insider (director or officer) cashing out. The reason is reasonably clear from disclosure and news: the Big Fund reduced holdings across several semiconductor names in late May 2026 to capitalize on a strong share-price rally, described as phased adjustments under previously disclosed plans, and its SMIC long position fell to 7.99% (BigGo Finance, 2026-05). At 0.12% of shares outstanding this is a routine partial monetization by a government investment vehicle that has held SMIC for years and is also simultaneously receiving 547.2M new SMIC shares through the SMNC deal. It is a portfolio-management action by a state fund, not a judgment on near-term fundamentals by management.

Buys. There were no open-market insider purchases by directors, officers, or substantial shareholders in the window. The absence of insider buying is unremarkable for a state-controlled Chinese issuer where executives are not typically large open-market buyers.

Net assessment. Insider activity over the last 12 months is minimal: a single, small, plan-driven sell by a state investment fund harvesting a rally, with no operating-insider buys or sells. There is no cluster, no executive conviction signal in either direction, and the sole transaction is explained by the seller's portfolio strategy rather than the business outlook. Net read: neutral. The Big Fund trim is mild noise, not a red flag, and certainly not the kind of executive selling that would warrant concern; equally, there is no insider buying to lean on as a bullish signal.

Section 12: Scenarios

Bull case. China's AI build-out keeps accelerating and Beijing keeps routing scarce domestic advanced capacity through SMIC. The 5nm pilot matures into usable volume, yields on 7nm/N+2 and N+3 keep climbing, and SMIC becomes the indispensable manufacturer for Huawei's Ascend AI accelerators and the next generation of Chinese AI silicon just as HBM and packaging bottlenecks ease. The Q2 2026 AI-driven surge proves to be the start of a multi-year up-cycle rather than a one-quarter pop, and AI-linked power-management and data-transmission chips lift both volume and ASP. On mature nodes, import substitution runs faster than the feared oversupply, so SMIC fills its new Lingang, Tianjin, Beijing, and (now wholly owned) SMNC fabs at healthy utilization while domestic OEMs pull foreign chips out of their bills of materials. Margins climb structurally above the 20% the company keeps beating to, and SMIC consolidates its position as the unambiguous backbone of a self-sufficient Chinese chip supply chain. In this world SMIC's geopolitical moat compounds: every new sanction makes its domestic customers more captive.

Base case. Management keeps doing what it has done for five straight quarters: guiding margins to 18-20%, printing at or above the top, and meeting revenue guidance. Full-year 2026 revenue grows above the industry average as promised, driven by the AI-linked pockets of strength and steady industrial/automotive gains, while the bulk mature-node business grows in volume but fights ASP pressure from China's capacity glut. Depreciation from the capex super-cycle keeps a lid on margin expansion, so profitability improves slowly rather than dramatically. The SMNC consolidation simplifies the structure without changing the trajectory. Capital stays inside the company - no dividend, no buyback, more fabs. SMIC remains the clear China champion and global number-three foundry, advancing one node generation behind the leaders, valuable to its country and steadily growing, but never closing the technology gap to TSMC.

Bear case. Export controls tighten where it hurts most: servicing, spares, or consumables for the DUV tools SMIC already owns, capping its ability to run and expand advanced capacity. At the same time China's own mature-node capacity flood - SMIC, Hua Hong, Nexchip and others all ramping at once - turns into genuine oversupply, and 28nm-and-above ASPs fall faster than cost reductions can offset. Utilization slips from the mid-90s, but depreciation from the just-built fabs keeps hitting the P&L regardless, so margins compress back toward the low-to-mid teens. A China consumer-electronics downturn, with revenue 89% domestic and no geographic offset, deepens the pain. The AI demand that powered the Q2 2026 guide proves narrow and lumpy rather than broad. SMIC keeps issuing shares and spending on capacity into a softening market, diluting minority holders to fund a national-strategic mission whose returns accrue to the state more than to outside shareholders. The geopolitical moat still protects the franchise, but the economics of being a capital-intensive, sanction-constrained, oversupplied mature-node foundry assert themselves.

Section 13: Further Reading

Huawei Ascend Production Ramp: Die Banks, TSMC Continued Production, HBM is The Bottleneck - SemiAnalysis, 2025-09-08 [paid]
China AI & Semiconductors Rise: US Sanctions Have Failed - SemiAnalysis, 2023-09-12 [free] - Argues sanctions backfired by forcing SMIC and Huawei to build a domestic 7nm-class capability, with detail on SMIC's good parametric yields on the Kirin 9000s.

A note on deliverable format: the report above is the complete .md deliverable. In this session I had only web-research tools available (no file-write capability), so I could not save it to a .md file on disk or render the charts directly. To produce the file, copy everything from the "# Semiconductor Manufacturing International Corporation" heading through the closing chart-data block into a .md file; the chart-data block is structured for your chart generator.

Sources: