Hua Hong Semiconductor (1347.HK) — 2026Q2 FY2026 Earnings Call Analysis

Orders At 2x Capacity, Return On Equity Still 2.4%

Gross margin expanded 560bps in a year and annualized ROE is still 2.4%. The pricing is real. The profits are not.

Thesis: Hua Hong has genuine pricing power at 100% utilization with 60% of Q3 growth coming from ASP increases, and management is backing it with $6B in committed capex. But 16.5% gross margin means every dollar of new fab depreciation crushes a razor-thin profit pool. The $545M in H2 depreciation against ~$236M in half-year gross profit makes this a bet on whether ASP increases can outpace the depreciation wall. ROE at 2.4% says they cannot yet. The stock only works if margins expand toward 20%+ as pricing flows through H2 2026 and into 2027.

Verdict: HOLD — Conviction: MEDIUM

Catalyst: Q3 earnings showing gross margin at or above 18% as last quarter's price increases flow through alongside $250M of new Fab 1 depreciation. Management says pricing actions 'manifest in H2 and into 2027.' If margins hold or expand through the depreciation ramp, the thesis holds.

Key Risk: H2 depreciation of approximately $545M ($250M Fab 1 + $210M Fab 2 + $30M Hua Li + $55M legacy) against quarterly gross profit of roughly $118M. If ASP momentum stalls while new fab depreciation hits full rate, margins compress below 12% and the $6B capex cycle generates no return.

The Tell: Dr. Bai volunteered that '2028 is where things, people start to have some debate' while simultaneously committing $6B into a fab that ramps through that exact timeframe. He is pre-blaming 2028 for potential weakness before the capex commitment is questioned. He then pivots to 'I do see a secular growth story' to soften the admission.

Detected Patterns

Pricing Power Signal: CFO confirmed 60% of Q3 revenue growth comes from ASP increases, not volume. Management says price increases taken last quarter flow through H2 and into 2027. Orders at 1.5x-2x capacity with 100% utilization means customers have no alternative. Standalone memory revenue up 149.3% YoY.
Beat and Raise Machine: All-time high revenue of $717.5M, gross margin up 560bps YoY, and Q3 guide of $770-780M implies another 7-9% sequential jump. Management guided margin up to 16-18% from 16.5% current, raising through a depreciation cycle.
Capital Conviction: Dr. Bai started construction of the third 12-inch fab in March before anyone confirmed 2027 demand. $6B over three years at roughly $2B annually, backed by an order book at 1.5-2x capacity. Management says 'even if a downturn comes, we keep building.'
High Utilization as Ceiling: 100% loaded with orders at 1.5-2x capacity, but Fab 9A output does not show until Q4 and Fab 9B first volume in Q1. New capacity to relieve the constraint arrives in 2027 at earliest. The constraint is real but the company cannot monetize beyond current capacity until 2027.
Backlog Quality: Orders at 1.5-2x capacity, but Chairman Peng admits lead times lengthen when demand is tight, meaning orders exceed deliverable output. Overbooked orders are cancellable. Peng flags 2028 as the year 'people start to have some debate.' Cancellable backlog normalizes fast on a cycle turn.
AI Label Without Substance: Products tied to AI are PMICs, MCUs, trench MOSFETs, and smart card ICs on mature specialty nodes. No CoWoS, no HBM, no advanced packaging, no leading-edge logic. Silicon photonics is 'early stages.' SiC power devices are 'planning.' The AI premium in the multiple is not backed by technology differentiation.

Friction Level: MODERATE_FRICTION — Both sides agree pricing power is real (60% of Q3 growth from ASP) and orders exceed capacity. The disagreement is whether that pricing power survives the $545M H2 depreciation ramp against a 16.5% gross margin, or whether it compounds into meaningful ROE improvement.

Report not found

The report data is no longer available. Please return to the archive.

Orders At 2x Capacity, Return On Equity Still 2.4% | Silicon Signal