Power Semiconductor / Rail

Sitting in a café, let me walk through my view on the next two quarters. First, on the revenue cadence. Looking at the historical pattern across the past four quarters, Q1 is the off-season (Spring Festival, low utilization), Q2 ramps up, Q3 pulls back slightly, and Q4 is the locomotive delivery peak—it's a "seasonal pattern" baked into our rail transit DNA, and that's largely the framework I work from. We're already sitting on roughly 5.1 billion booked for 2026 Q1, so year-over-year should look decent; I'm modeling 2026 Q2 at around 8.8 billion (approximately +15% YoY), driven by two engines: one, rail transit equipment deliveries recovering from the Q1 trough; two, our power semiconductors—particularly the SiC production line—starting to ship meaningful volume and contribute incremental growth. For Q3 I pencil in 7.5 billion. Per the usual pattern, Q3 cools a bit versus Q2, but year-over-year should still post growth because the SiC second curve is still climbing. Now gross margin. I'm assuming Q2 steps up to about 34% and Q3 to roughly 34.2%—a clear step up from 2025 Q2's 31.1%. The logic underneath is a "product mix reshuffle"—higher-margin SiC modules, automotive-grade IGBT, and new-energy power devices take up more share, gradually offsetting the traditional IGBT price war. This is where the IDM model really shines: integrated wafers, modules, and end products. Margin improvement comes in gradually, nothing like the rollercoaster ride Fabless plays on. On net profit, I'm working off a 13% net margin, landing at roughly 1.14 billion and 0.975 billion respectively. Operating margin, I'm holding near 17%—R&D spend still needs to keep flowing into SiC and the 8-inch production line, so I don't see it expanding much beyond the 2025 Q4 high water mark. Two key assumptions: ① SiC production line ramps on plan and hits yield targets; ② major rail transit customer orders release on schedule under the annual framework. The biggest risk I want to flag: **SiC moving from lab to fab—timing could be half a beat behind the model.** Substrate yield, epi-wafer supply, or automotive-grade certification—if any one of those bottlenecks, the slope of the entire second curve flattens, and the margin improvement gets diluted. If Q3 revenue slips below that 7.5 billion line, that's the signal this risk is materializing. The other concern is uncertainty in the rail transit bidding cadence, but that's relatively manageable. Overall: revenue up, earnings up, margins up—that's the "gradual ramp" path I see. Not a rocket-ship takeoff.