SMIC’s Rally Has Already Cashed the Guidance Check

The market’s story on SMIC fits on one line: AI demand is real, domestic substitution is accelerating, Q2 guidance pointed to 14-16% sequential revenue growth with gross margin stepping up to 20-22%, and Huawei’s chip noise in mid-May confirmed that China’s foundry champion is the cleanest way to own the trade. Up roughly 25% in three months, with most of that move packed into the weeks after Q1, the stock now trades at 85.20 HKD against a consensus average target of 79.97. The price has already crossed the line analysts drew for it.

The question is no longer whether the guidance is good. The question is what is left to pay you for owning it.

Trailing earnings growth sits at minus 9.7%, operating margin slipped to 6.00% from 6.83%, and free cash flow ran negative 3.84 billion CNY for FY2025. The rally was not built on trailing numbers. It was built on a forward sentence from management and a sentiment jolt from a competitor’s announcement.

The number the rally is leaning on, and what it actually says

Utilization is the KPI doing the heavy lifting in the bull case. The progression is real: 89.6% in Q1 2025, 92.5% by Q2, 95.7% by Q4. A foundry running near reticle-limit utilization can push pricing, bin yields more aggressively, and let gross margin breathe. On the first two links of that chain, SMIC is delivering.

The third link is where the arithmetic turns ugly. Capex was 8.1 billion USD for FY2025, against revenue of roughly 9.6 billion CNY. That is not a fab harvesting mid-cycle earnings power. That is a fab still building, with the depreciation tail attached to every new tape-out window it opens for fabless designers in Shanghai and Shenzhen. Operating margin compressed even as utilization climbed nearly six points year over year, which tells you the depreciation load is doing exactly what front-loaded capex does: it eats the operating line before the revenue ramp catches up. The market is paying mid-cycle multiples for what is structurally a pre-mid-cycle margin profile.

And the export-control backdrop matters here not as headline risk but as margin pressure. BIS enforcement has tightened through 2026, with penalties on equipment suppliers for shipments into SMIC’s fabs and fresh legislative proposals targeting advanced tooling. Beijing’s subsidy and self-reliance push offsets some of this on demand. On cost, every workaround, every domestically substituted tool, every parallel supply chain carries a bill that does not show up in revenue but does show up in capex intensity and depreciation per wafer. The rally has priced the demand story. It has not, by any visible math, priced the cost of self-reliance.

The strongest version of the other side

The bull case worth taking seriously is not the AI-demand-forever line. It is the structural one: SMIC is the only scaled foundry inside a closed domestic system being walled off from TSMC and Samsung at the advanced-node frontier by policy, not competence. In that frame, 95.7% utilization is not a peak but a floor, because Chinese fabless designers have nowhere else to tape out leading-edge designs at volume. Normalized earnings power, on that view, looks nothing like the 0.09 CNY diluted EPS the FY2025 statements show. The market made the same assumption with TSMC during earlier capex super-cycles, when the depreciation drag looked punishing in the moment and trivial in hindsight once volume caught the fixed-cost base.

The trouble with that case is the math of the entry point, not the destination. SMIC currently posts negative 3.84 billion CNY of free cash flow, 14.5 billion CNY of total debt, negative 2.2 billion CNY net cash, and 2.7 CNY book value per share against an 85.20 HKD price. Replacement-cost arguments can justify a rich multiple on a foundry. They do not justify paying up for next year’s guidance after the stock has already moved 25%. This company’s peak-to-trough EPS range has historically been wide enough that anchoring to a single forward quarter is how the market rediscovers gravity.

There is also the matter of the consensus target. Average sits at 79.97 against a current 85.20. The high end reaches 137.01, the low end 25.01. That is not a tight distribution around a confident base case. That is analysts modeling two different companies: one where the self-reliance flywheel compounds cleanly, one where depreciation and sanctions grind the operating line. The current price sits above the average and assumes the upper distribution is the right one to read.

If utilization holds above 95% through the next two quarters and gross margin lands inside the 20-22% guided band, the key test is whether operating leverage finally breaks through depreciation drag. Short of that, the price is doing analysts’ work before the operating numbers have done theirs.

Priced like the ramp is already in the P&L. Operates like a foundry still spending 84% of revenue on capex. Not the same thing.

Figures reflect the most recent available data and may differ slightly from live market prices.

© Mathstock