Year-over-year revenue growth is 3% and the AI business contributed zero dollars to a call that opened with an AI headline.
Thesis: The quarter is operationally real: 17.1% operating margin, inventory draining 27 days to 265, channel back inside the 7-8 week target, OpEx revised down while funding growth bets. But the P&L drivers are appliance seasonality, industrial mix, a yen tailwind, and a restructuring charge already absorbed. AI is a press-release line item with revenue at earliest 2028. Own the execution or pay for the narrative, but the execution alone does not re-rate the multiple.
Verdict: HOLD — Conviction: MEDIUM
Catalyst: Q3 at or above the $126M midpoint with gross margin holding 54-55% despite the yen headwind, plus TinySwitch-5 hitting meaningful revenue in H2 2026. That would prove the margin expansion is volume and mix, not currency and cost cuts.
Key Risk: Two of three margin drivers reverse: yen flips to a headwind in Q3 (guide already shows GM dropping to 54-55%) and the restructuring benefit is spent. Only the OpEx discipline persists. If appliance seasonality fades in Q4, the sequential growth engine stalls before any data center revenue arrives.
The Tell: The first analyst question deflated the call's own headline. Needham asked when 2,200V GaN produces revenue; the CEO answered: 'It's really just a technology demonstration. It's not a product... It'd be the next couple of years before we see anything meaningful there.' The opener was the demo; the first answer admitted no product.
Friction Level: MODERATE_FRICTION — Both sides agree the operations are clean: inventory down to 265 days, channel at 7.3 weeks, $18M FCF, real beat-and-raise. The disagreement is whether the industrial mix shift (43% of sales, +16% YTD) justifies repricing the stock now, versus paying today for an AI/data center narrative that has zero revenue and no product until 2028-2030.
2026Q2 · Aug 5, 2026
3 of 5 readings agreed3 HOLD · 2 LONGRead the full call
Margins Up 540 Basis Points, Yen Reverses Next Quarter
The quarter is operationally real: 17.1% operating margin, inventory draining 27 days to 265, channel back inside the 7-8 week target, OpEx revised down while funding growth bets. But the P&L drivers are appliance seasonality, industrial mix, a yen tailwind, and a restructuring charge already absorbed. AI is a press-release line item with revenue at earliest 2028. Own the execution or pay for the narrative, but the execution alone does not re-rate the multiple.
The first analyst question deflated the call's own headline. Needham asked when 2,200V GaN produces revenue; the CEO answered: 'It's really just a technology demonstration. It's not a product... It'd be the next couple of years before we see anything meaningful there.' The opener was the demo; the first answer admitted no product.
Revenue +10% QoQ with OpEx down slightly to $45.2M (below the $47M midpoint) and gross margin up 160bps on 43% industrial mix. Operating leverage is demonstrable in the print, though 540bps of it rests partly on yen and a spent restructuring charge.
The call opens on 2,200V GaN, NVIDIA 800V reference designs, and a billion-dollar SAM by 2030, while CFO states aux power data center revenue could start in 2028 and main GPU power path is 'still earlier.' Zero AI dollars in $118.9M revenue.
Management explicitly pivots R&D toward data center and auto while 'calibrating' investment in appliance and low-power industrial, which they call 'the main drivers of revenue and cash flow.' Near-term cash generators get less, future revenue gets more.
Industrial is 43% of sales, grew 14% in Q2 and 16% YTD after 15% in 2025, spanning renewables, building automation, metering, and battery storage gate drivers. This is the real growth engine and it is not the consumer appliance cycle the street models.
Same PowiGaN architecture scaling from 750V through 2,200V means customers designing 800V systems today are not forced to switch technologies at 1,500V. Roadmap lock-in is being built years before revenue shows.
+30
NEUTRAL"Direct answers with specifics on near-term items (OpEx $45.2M vs $47M midpoint, yen timing in Q3 vs Q4, channel weeks). Hedged language on the distant roadmap: 2,200V is 'just a technology demonstration,' main power path is 'still earlier,' aux power 'could be revenue in 2028.' No deflection, no overcommitment."
Full transcript with specific figures on gross margin drivers (mix, volume, yen), inventory days, channel weeks, OpEx revision, and a clean CFO answer on supply chain. Only gap is valuation, which is not in evidence.
Q3 at or above the $126M midpoint with gross margin holding 54-55% despite the yen headwind, plus TinySwitch-5 hitting meaningful revenue in H2 2026. That would prove the margin expansion is volume and mix, not currency and cost cuts.
Two of three margin drivers reverse: yen flips to a headwind in Q3 (guide already shows GM dropping to 54-55%) and the restructuring benefit is spent. Only the OpEx discipline persists. If appliance seasonality fades in Q4, the sequential growth engine stalls before any data center revenue arrives.
The verdict is public. The case behind it is Pro: the forensic reading of guidance and language, the bull and bear cases, the red flags, and the play. It opens to everyone on Nov 3, 2026.