The global semiconductor industry is undergoing a structural upcycle driven by artificial intelligence. Unlike previous cycles fueled by consumer electronics or mobile internet, this round is powered by the exponential surge in AI computing demand—from GPU clusters for training large models to custom chips for large-scale inference deployments. AI is reshaping the entire semiconductor value chain, from design to manufacturing.
Gartner’s latest forecast projects that global semiconductor revenue will surpass $1.3 trillion in 2026, up roughly 60% from 2025, marking the fastest growth rate in nearly two decades. AI-related chips will account for 30% of industry revenue, becoming the primary growth engine. Against this backdrop, wafer foundries—the core of semiconductor manufacturing—are experiencing both a shift in demand structure and a new competitive landscape. As China’s largest foundry, where does SMIC (Semiconductor Manufacturing International Corporation) stand in this cycle? How does AI demand flow from design to manufacturing? And to what extent are geopolitical factors reshaping the company’s growth logic? This analysis explores these questions from the perspectives of industry data, financial performance, and supply chain structure.
The Global Semiconductor Cycle: AI-Driven Structural Upturn
The semiconductor market in 2026 displays characteristics that set it apart from previous cycles. Gartner data shows global semiconductor revenue at $805.3 billion in 2025, projected to leap to $1.32 trillion in 2026 and further to $1.55 trillion in 2027. This growth pace far exceeds previous industry expectations—back in 2021, most analysts predicted annual chip sales wouldn’t top $1 trillion until 2030.
The key driver behind this explosive growth is the concentrated release of AI infrastructure investment. Hyperscalers are expanding AI data centers at an unprecedented rate. According to CreditSights and Fortune, hyperscaler capital expenditure is expected to reach $602–700 billion in 2026. Alphabet, Amazon, Meta, and Microsoft together plan annual capex exceeding $700 billion. Much of this capital is flowing into AI accelerator procurement, which in turn boosts demand for wafer foundry services.
Looking at the industry chain, AI demand pulls on semiconductors in a clear hierarchy. At the top are AI chip design firms—NVIDIA, for example, posted $215.9 billion in revenue for fiscal 2026, up 65% year-on-year, with data center business contributing $193.7 billion, nearly 90% of total revenue. Entering fiscal 2027, NVIDIA’s quarterly revenue surpassed $81.6 billion, with data center business at $75.2 billion, up 92% year-on-year. This scale of chip shipments directly drives foundry capacity.
The second tier is the wafer foundry segment. Counterpoint Research reports that in Q1 2026, global foundry 2.0 market revenue grew 23% year-on-year to $86 billion. TSMC, as the main beneficiary of the AI-driven upcycle, saw Q1 revenue jump 41% year-on-year, with advanced nodes (7nm and below) accounting for 74% of wafer sales. DIGITIMES estimates global foundry revenue will exceed $230 billion in 2026, up 17% from 2025.
The third tier encompasses the broader chip manufacturing ecosystem, including advanced packaging, semiconductor equipment, and materials. The increasing complexity of AI chips is driving the "Foundry 2.0" concept—deep integration of wafer manufacturing, advanced packaging, and testing capabilities. Industry competitiveness now depends not just on process technology but also on the ability to deliver comprehensive solutions at scale.
SMIC’s Performance and Capacity Status
Amid this industry upturn, SMIC delivered a milestone first quarter in 2026.
According to its Hong Kong stock exchange filing, SMIC posted Q1 sales of $2.505 billion, breaking the $2.5 billion mark for the first time in a single quarter—up 11.5% year-on-year and 0.7% quarter-on-quarter. Under A-share accounting, revenue was RMB 17.617 billion, up 8.1% year-on-year, with net profit attributable to shareholders at RMB 1.361 billion. Gross margin reached 20.1%, up 0.9 percentage points quarter-on-quarter, exceeding the previous guidance range of 18% to 20%.
On the capacity front, monthly capacity at the end of Q1 reached 1.078 million wafers (8-inch equivalent), up 10.8% year-on-year. Capacity utilization remained high at 93.1%. Wafer shipments totaled 2.51 million, up 9.5% year-on-year. Average selling price (ASP) per wafer rebounded 0.9% quarter-on-quarter to $999. Revenue from 12-inch wafers accounted for 76.4% of the total, with the share of high-end capacity continuing to rise.
From an end-market perspective, consumer electronics remained the largest revenue source at 46.2%, up 27% year-on-year. The industrial and automotive segment jumped from 9.6% to 14% of revenue, up 63% year-on-year, making it the fastest-growing segment. Smartphone and PC/tablet revenues accounted for 18.9% and 13.6%, respectively.
A notable trend is the shift in regional revenue mix. China’s share rose to 88.9%, up from 84.3% a year earlier, while the US share fell from 12.6% to 9.3%. This reflects a growing trend among domestic customers to shift orders to local foundries to ensure supply chain security amid rising geopolitical uncertainty. At the earnings call, co-CEO Zhao Haijun noted that AI demand is crowding out mature process foundry and memory capacity globally, driving a significant order flow back to China.
For Q2, SMIC issued an upbeat outlook: revenue is expected to rise 14–16% quarter-on-quarter, with gross margin between 20% and 22%. Management stated they are "more optimistic about overall operations this year."
How AI Demand Flows to Foundries: SMIC’s Supply Chain Position
AI-driven demand does not flow evenly to foundries; instead, it exhibits pronounced structural characteristics. Understanding this mechanism is key to assessing how much SMIC stands to benefit in this cycle.
Scarcity Premium for Advanced Nodes. AI training chips (such as NVIDIA GPUs) require cutting-edge process nodes, currently relying on TSMC’s 5nm, 4nm, and 3nm technologies. In Q1, advanced nodes accounted for 74% of TSMC’s wafer revenue, with 3nm at 25% and 5nm at 36%. These capacities are expected to remain fully loaded throughout the year, with orders booked through 2027. SMIC’s most advanced N+3 node, according to SemiAnalysis, is roughly equivalent to TSMC’s N6. This means that in the short term, SMIC cannot directly replace TSMC for top-tier AI training chip foundry. However, the shortage of advanced node capacity is creating a "spillover effect"—as the world’s most advanced capacity is absorbed by AI training chips, other chips (such as high-end smartphone processors and AI inference chips) that previously used advanced nodes are shifting to slightly older nodes. This shift creates incremental demand for SMIC’s N+2 and N+3 processes.
The Rise of AI ASICs and Diversified Foundry Demand. A significant structural shift is underway as the AI chip market evolves from a "GPU-dominated" to a "GPU+ASIC dual-track" model. JPMorgan estimates the digital AI ASIC market will reach $60–70 billion by 2026, with a compound annual growth rate of over 40–50% in the coming years. Projections for 2026 put ASIC chip shipments at 7.7 million units, with a 45% market share, overtaking GPUs at 58% in 2027. Unlike NVIDIA’s general-purpose GPUs, ASICs are typically designed for specific customers and workloads, with more diverse process node requirements—not all ASICs need the most advanced 3nm or 5nm nodes. Some inference chips can achieve competitive power efficiency at 7nm or even more mature nodes. This trend creates differentiated opportunities for foundries with broad process capabilities.
Supply-Demand Reversal for Mature Nodes. The most unexpected shift in this cycle is in mature process nodes. TrendForce notes that as AI-related demand continues to crowd out foundry resources, and as major players like TSMC and Samsung gradually cut 8-inch and some 12-inch mature node capacity, the supply-demand balance for mature nodes is tightening rapidly. Some constrained nodes saw price hike intentions of 5–10% in Q2–Q3 2026, with the upward pricing trend expected to extend into 2027. SMIC began a new round of price increases for mature nodes in July 2026, with previous hikes already pushing Q2 ASP up by 5–6% quarter-on-quarter. The reversal in mature node supply-demand dynamics has improved SMIC’s pricing power in its largest capacity segment.
Structural Push for Domestic Substitution. Geopolitical factors are accelerating the "de-Americanization" of China’s semiconductor supply chain. Ongoing US export controls—including SMIC’s designation as a "restricted facility" and proposed bans on DUV lithography equipment exports to China—have limited access to advanced process equipment but also spurred domestic chip designers to shift orders to local foundries. SMIC’s China revenue share climbed from 84.3% in 2025 to 88.9% in Q1 2026, a trend expected to continue. According to Semicon China data cited by Reuters, Chinese fabs’ global capacity share at mature nodes (22–40nm) will rise from 32% in 2025 to 41% in 2027. Mainland China’s foundry capacity is projected to exceed 4 million wafers per month (8-inch equivalent) in 2026, accounting for 34.4% globally, with a compound annual growth rate of 13.4%—the fastest worldwide.
Structural Opportunities and Constraints
Based on the above, SMIC’s structural opportunities in this AI-driven semiconductor cycle can be understood from three angles.
First, the upturn in mature nodes is improving profitability. SMIC’s capacity is weighted toward mature nodes, and the shift from oversupply to tightness globally is keeping utilization high (93.1%) and ASPs on the rise. TrendForce expects the price hike cycle to continue through 2027, suggesting SMIC could benefit from rising prices for several quarters to come.
Second, the explosion in AI inference demand is creating new incremental markets. As the AI industry focus shifts from "model training" to "inference applications," the structure of chip demand is changing. Inference chips require less advanced nodes than training chips but place greater emphasis on cost, power efficiency, and shipment scale. This market profile aligns well with SMIC’s capacity and cost structure. According to market sources, Huawei has confirmed its DeepSeek V4 model is powered by the Ascend 950 PR AI chip, and SMIC is believed to be the only Chinese foundry capable of supporting this chip with its N+3 node.
Third, domestic AI chipmakers’ foundry demand is shifting structurally. Under geopolitical constraints, Chinese AI chip design firms (including HiSilicon, Cambricon, Horizon Robotics, and others) can no longer rely on advanced overseas foundry capacity and must seek domestic alternatives. As China’s leading advanced process foundry, SMIC occupies an irreplaceable position in this shift. A recent CMB International report initiating coverage of SMIC’s H-shares described the company as "China’s core wafer manufacturing platform, combining domestic scale, broad coverage of mature and specialty processes, advanced logic capabilities, and deep involvement in the domestic fabless ecosystem," assigning a "Buy" rating and a target price of HK$110.
However, these opportunities are also subject to clear constraints.
Catching up in advanced nodes faces significant bottlenecks. SMIC’s N+3 process is competitive on some metrics—SemiAnalysis measured its minimum metal pitch at 32.5nm, about 10% tighter than Intel’s 18A process at 36nm—but this was achieved using only DUV lithography, without EUV. In terms of process maturity, yield, and production cost, N+3 still lags behind TSMC’s N6. If US export controls tighten further, SMIC will face real equipment bottlenecks in developing post-N+3 nodes.
High capital expenditures will continue to weigh on margins. Depreciation and amortization reached $1.088 billion in Q1, up 25.7% year-on-year, with capex at $1.563 billion. The depreciation burden from capacity expansion is a major factor limiting further margin improvement. CMB International also noted, "Depreciation will remain the main constraint on margin recovery."
Sustainability of AI infrastructure investment is uncertain. Today’s AI chip demand boom is built on hyperscalers’ large-scale capex. If cloud providers slow their AI investments, or if AI commercialization lags expectations, the current tight supply-demand situation could reverse. DIGITIMES has also flagged the risk of a potential "AI infrastructure investment bubble."
Conclusion
The global semiconductor industry is in the midst of an AI-driven structural upcycle. Gartner’s projection of $1.3 trillion in industry revenue by 2026, with AI chips accounting for 30%, highlights the scale and intensity of this cycle. Against this backdrop, SMIC’s growth logic is shifting from "consumer electronics-driven capacity expansion" to "structural upgrades fueled by AI spillover and domestic substitution."
Q1 data already validates this trend in key metrics: revenue ($2.505 billion), capacity utilization (93.1%), and gross margin (20.1%). The reversal in mature node supply-demand is improving profitability, the boom in AI inference demand is creating new incremental markets, and geopolitics-driven supply chain restructuring is giving SMIC a strategic, irreplaceable position. However, technical bottlenecks in advanced nodes, high capex pressure on margins, and uncertainty around the sustainability of AI investment remain constraints.
SMIC’s ultimate gains in this cycle will depend on three evolving variables: the duration and magnitude of the mature node price upcycle, technological breakthroughs and yield ramp-up for N+3 and subsequent nodes, and the speed and scale at which domestic AI chipmakers shift orders to local foundries. The interplay of these factors will determine whether China’s largest foundry can turn a cyclical boom into a lasting competitive edge.
FAQ
Q1: What is SMIC’s most advanced process node, and can it handle AI chip foundry orders?
SMIC’s most advanced node is N+3, which SemiAnalysis equates to TSMC’s N6. This process is already used to manufacture Huawei’s Ascend series AI chips. While there is still a generational gap compared to TSMC’s 3nm and 5nm nodes, N+3 represents the highest level of domestic foundry capacity available for Chinese AI chipmakers unable to rely on overseas fabs, making it highly valuable.
Q2: How does AI demand affect the market landscape for mature process nodes in foundries?
The expansion of AI chip capacity is crowding out global foundry resources, while TSMC, Samsung, and others are cutting mature node capacity. This is rapidly tightening supply for 8-inch and 12-inch mature nodes. TrendForce expects the price upcycle to continue through 2027, with some nodes seeing 5–10% price hikes in Q2–Q3 2026. This directly benefits SMIC, whose business is centered on mature nodes.
Q3: How did SMIC perform in Q1 2026?
In Q1, SMIC achieved $2.505 billion in sales, surpassing $2.5 billion in a single quarter for the first time—up 11.5% year-on-year. Gross margin reached 20.1%, up 0.9 percentage points quarter-on-quarter, with capacity utilization at 93.1%. The company expects Q2 revenue to rise 14–16% quarter-on-quarter, with gross margin between 20% and 22%.
Q4: What is the long-term impact of US export controls on SMIC?
The US has designated SMIC as a "restricted facility" and plans to ban exports of semiconductor manufacturing equipment. This limits SMIC’s ability to acquire advanced EUV lithography tools, creating bottlenecks for post-N+3 process development. On the other hand, it is accelerating the shift of domestic chip design orders to local foundries—SMIC’s China revenue share has risen to 88.9%. In the short term, order reshoring is the dominant effect; long term, equipment embargoes create a technological ceiling.
