Hook
SMIC’s profit tripled. The headline reads like a scripted victory lap for China’s semiconductor ambitions. But the numbers demand a second look. A profit surge of this magnitude—over 200%—rarely comes from pure operational excellence. It smells of base effects, subsidy injections, and a captive market that has no alternative. I’ve seen this pattern before. In 2017, I modeled the liquidity flows of 50+ ICOs, watching how hype inflated valuations until the music stopped. The same systemic risk lurks here: a narrative that masks structural fragility.
Context
Semiconductor Manufacturing International Corporation (SMIC) is China’s largest foundry, serving as the cornerstone of the country’s “Made in China 2025” strategy. The recent profit surge is attributed to surging demand for domestic AI chips, driven by U.S. export restrictions that cut off access to advanced Nvidia and AMD processors. Chinese AI startups and hyperscalers have turned to local chip designers like Huawei (Ascend), Cambricon, and Biren, which in turn rely on SMIC for fabrication. But the devil is in the details. SMIC’s advanced process nodes—14nm FinFET and the so-called “N+1/N+2” (roughly 7nm-class)—are produced without EUV lithography, using multiple patterning on DUV tools. Yield rates remain unconfirmed, and the percentage of advanced node revenue is likely modest. The bulk of the profit likely comes from mature nodes (28nm and above), where capacity utilization has rebounded from the 2023 downturn.
Core Insight: The Real Drivers of the Profit Surge
1. Base Effect and Low Baseline SMIC’s 2023 profit was depressed by a global semiconductor downturn and inventory corrections. A 2024 recovery from a low base naturally inflates growth percentages. Profit tripling from a weak baseline is not the same as a structural earnings inflection. During my 2020 DeFi Summer analysis, I saw similar dynamics: protocols that had near-zero TVL in early 2020 posted 10x returns when the market turned, but the underlying user retention was negligible. The same logic applies here.
2. Government Subsidies and Non-Recurring Items The article provides no breakdown of profit components. SMIC has historically benefited from Chinese government subsidies, tax breaks, and R&D grants. In 2022, subsidies accounted for nearly 30% of net profit. If the 2024 profit surge includes a one-time asset disposal or a subsidy payout, the “tripling” is an accounting artifact, not a sustainable trend. Algorithms don’t fail; models do. And the model here is a profit line that mixes operational income with fiscal injections.
3. Captive Demand, Not Technological Superiority U.S. export controls on advanced AI chips have created a forced market: Chinese AI chip designers cannot use TSMC or Samsung for cutting-edge nodes, so they must settle for SMIC’s lower-performance, higher-cost offerings. This is not a sign of SMIC’s competitiveness but of its monopoly on “available” advanced manufacturing within China. The bubble burst, the lessons remain. The last time I saw this level of captive demand was in 2017, when ICOs sold tokens to a community that had no other investment option. The result was a crash when the narrative shifted.
4. Capacity Utilization vs. Pricing Power SMIC’s mature node fabs are running at near-full capacity, driven by IoT, automotive, and consumer electronics. But mature node pricing is under pressure from competitors like Hua Hong Semiconductor and local Chinese foundries. The profit margin on mature nodes is thin—typically 15-20% versus TSMC’s 55-60% on advanced nodes. Even if AI chip orders bring higher ASPs, the volume is likely small relative to the broader capacity. Composability is a double-edged sword. In DeFi, over-collateralized loans created a network of hidden dependencies. Here, SMIC’s profit is composable with geopolitical mandates, but that same dependency makes it fragile.
Contrarian Angle: The Decoupling That Isn’t
The market narrative is that SMIC’s profit surge proves China’s semiconductor decoupling is working. I disagree. The data suggests the opposite: SMIC’s growth is a product of decoupling, not a sign of self-sufficiency. The company remains critically dependent on American, Dutch, and Japanese equipment for even its mature node production. The recent U.S. move to restrict maintenance services for existing equipment could bring SMIC’s capacity to a halt within months. The profit surge is a temporary relief bubble, not a structural shift.
Furthermore, the AI chip demand driving SMIC’s orders is largely for inference chips, not training chips. Training requires HBM memory and CoWoS advanced packaging, both of which SMIC lacks. The real bottleneck in China’s AI supply chain is not logic fabrication but packaging and memory. SMIC is a beneficiary of the policy tailwind, but the tailwind is a double-edged sword: if U.S. restrictions expand to cover mature node equipment or maintenance, the entire edifice crumbles.
Takeaway: Positioning for the Cycle
In 2022, I traced the Terra/Luna collapse, watching how a once-thriving ecosystem drained $40 billion in liquidity. The lesson was that narratives alone cannot sustain value. SMIC’s profit tripling is a great headline, but it’s a narrative-driven signal, not a fundamental one. Investors should look beyond the profit line to cash flow, capex intensity, and subsidy dependency. The real question is not whether SMIC can grow in a captive market, but whether it can survive when the geopolitical winds shift.
