SMIC's profit more than tripled. Headlines scream "China's AI chip demand drives foundry growth."
Bullish?
Not so fast.
Dig into the order flow. The surge is a liquidity event—forced demand from captive customers, not a technological breakthrough.
This is not innovation. It's a toll booth.
Context: The Only Game in Town
SMIC is China's largest semiconductor foundry. It's also the only domestic option for advanced logic chips. US export controls have blocked access to EUV lithography and high-end DUV tools. SMIC's 7nm-class N+2 process is limited, with low yields and high costs.
Yet Chinese AI chip designers—Huawei, Cambricon, Biren—need production. They can't go to TSMC (Taiwan, sanctioned). They can't go to Samsung (US-controlled). So they come to SMIC.
This is captive demand. It's not a vote of confidence in SMIC's technology. It's a forced choice.
Core: The Order Flow Analysis
Let's break down the profit components.
First, the base effect. In 2023, SMIC's net profit fell sharply due to the global semiconductor downturn. A 2024 recovery from a low base amplifies the percentage growth. "More than triples" sounds impressive, but the absolute numbers matter. SMIC's net profit in 2023 was around $900 million. Triple that is $2.7 billion. TSMC's 2024 net profit? Over $30 billion. The gap is not closing.
Second, government subsidies. China's state-backed initiatives—Big Fund, local government grants—flow directly into SMIC's income statement. These are non-recurring or semi-recurring. They mask the underlying operational profitability. Strip them out, and the margin story weakens.
Third, captive demand pricing. SMIC can charge a premium because Chinese AI chip firms have no alternative. But this is a monopoly rent, not a sign of superior technology. The moment export controls ease or domestic competitors emerge, that pricing power evaporates.
Now, look at the cost side. SMIC's capital expenditure remains high—over 50% of revenue. New fabs in Beijing, Shanghai, Shenzhen are under construction. Depreciation will hit margins hard once utilization rates drop. The current high utilization (thanks to AI orders) masks the looming depreciation headache.
This is a classic liquidity trap. The profit surge is real, but it's a one-time adjustment to a new equilibrium of captive demand. The long-term trend is rising costs, capped technology, and regulatory risk.
And that's not all. The demand itself is fragile. Chinese AI chip orders may include panic buying, double-ordering, and policy-driven stockpiling. Real end-user demand for Chinese AI chips (like Huawei's Ascend) is still unproven at scale. If the AI bubble pops, SMIC's order book will shrink faster than a DeFi TVL after a rug pull.
Contrarian: Retail Sees a Boom, Smart Money Sees a Fragile Toll
Retail investors read the headline and think "China's semiconductor ambitions are paying off." They buy the stock. They push the valuation to 50x P/E or higher.
Smart money? They see a different picture.
SMIC is a toll booth. It collects fees from captive customers who have no other road. But toll roads are regulated. The government can cap prices. The government can also build competing roads (e.g., Huawei's own foundry, or new state-backed fabs).
More importantly, the toll booth is under constant attack. US export controls are not static. The next round could target mature-node equipment maintenance, or even software updates. SMIC's existing fabs rely on American and Dutch parts for maintenance. A spare parts ban would be a liquidity kill switch.
And the technology gap? It's widening. TSMC is already mass-producing 3nm with GAA. SMIC is stuck at 7nm (with low yields). The gap is not 2-3 nodes; it's a different generation. Chinese AI chips that need high performance will always be hobbled by SMIC's limitations.
So the profit surge is a mirage. It's a short-term liquidity event driven by a temporary monopoly. The long-term trend is systemic fragility.
Gas is the toll for chaos.
Takeaway: Watch the Maintenance, Not the Profit
SMIC's profit tripled. But the real signal is not in the P&L. It's in the equipment maintenance logs. If a single ASML DUV tool goes down and can't be repaired, a whole production line stops.
Liquidity dries up when fear sets in.
In this market, the smart money is not buying the narrative. It's shorting the volatility.
Ask yourself: If SMIC's profit triples again next year, will it be because of technology or because of more captive demand? If the answer is the latter, then this is a trade, not an investment.
Code is law, but bugs are fatal.
And SMIC's biggest bug is that it's a captive toll booth on a road that could be rerouted at any moment.