Advanced computing grew 100% sequentially and still only represents $20M of a $164M quarter while losses tripled.
Thesis: Advanced computing is a real structural driver but at $20M per quarter it cannot offset Power IC declining 14% YoY while opex rises 12% YoY. The June guide to 23% gross margin with half from mix is the single data point that matters. Until that prints with positive operating cash flow, the bottom call is management's forecast, not a fact. The stock is a bet on the June confirmation.
Verdict: HOLD — Conviction: MEDIUM
Catalyst: June quarter print of $168M at 23% non-GAAP gross margin with positive operating cash flow would confirm the bottom. Management explicitly guided both revenue and margin to sequential improvement for the first time in three quarters.
Key Risk: June guide range bottoms at $158M, below this quarter's $163.8M. If the print comes in at the low end, the trough call fails and the market reprices the three-quarter margin decline as structural rather than cyclical.
The Tell: CEO states 'our backlog provides us with good visibility' for medium voltage expansion, then later in the same prepared remarks says 'visibility into the second half of the calendar year remains limited given ongoing macro and component-related uncertainties.' He is confident about capacity he is expanding into but cannot see what happens six months out.
Friction Level: MODERATE_FRICTION — Both sides agree advanced computing is growing fast and the P&L is deteriorating. The disagreement is whether the June margin guide to 23% confirms a real trough or is a one-quarter utilization bounce masking structural weakness in the higher-margin Power IC line.
2026Q3 · May 6, 2026
3 of 5 readings agreed3 HOLD · 2 SHORTRead the full call
AI Doubles Off a Base Too Small to Move Earnings
Advanced computing is a real structural driver but at $20M per quarter it cannot offset Power IC declining 14% YoY while opex rises 12% YoY. The June guide to 23% gross margin with half from mix is the single data point that matters. Until that prints with positive operating cash flow, the bottom call is management's forecast, not a fact. The stock is a bet on the June confirmation.
CEO states 'our backlog provides us with good visibility' for medium voltage expansion, then later in the same prepared remarks says 'visibility into the second half of the calendar year remains limited given ongoing macro and component-related uncertainties.' He is confident about capacity he is expanding into but cannot see what happens six months out.
Advanced computing revenue more than doubled sequentially and grew 40% YoY, reaching 25% of the computing segment. Medium voltage MOSFETs for 48V-to-12V conversion moving into build phase at ODMs serving hyperscalers. Management is expanding medium voltage capacity ahead of demand.
Management is actively expanding medium voltage capacity with backlog providing visibility, and CapEx guidance stepping up from $12.1M actual to $15-17M for June quarter to support the ramp.
CEO states backlog provides good visibility for medium voltage, then states visibility into second half of calendar 2026 remains limited due to macro and component uncertainties, both in the same prepared remarks. Backlog conviction and forward visibility are contradictory.
Advanced computing represents roughly $20M of $163.8M revenue, about 12%. It doubled off a small base. The higher-margin Power IC line fell 20.3% QoQ while lower-value DMOS grew 13.9%. The AI narrative carries 12% of revenue into a claimed 2027 acceleration.
CFO confirms ASP erosion continues, just slower than December. Foundry and material cost increases are absorbed into guidance with no pass-through. A capacity-constrained supplier would not suffer continued ASP erosion.
-30
CAUTIOUS"CEO pairs 'backlog provides us with good visibility' with 'visibility into the second half of the calendar year remains limited' in the same prepared remarks. Added hedging language: 'can never be fast enough,' 'always anxious,' 'dynamic environment with some uncertainty.'"
Specific segment revenue, product mix splits, margin bridge (half utilization, half mix), and input cost details provided. Guidance range of ±$10M on $168M is wide, and key forward metrics like book-to-bill or backlog dollar value were not disclosed.
June quarter print of $168M at 23% non-GAAP gross margin with positive operating cash flow would confirm the bottom. Management explicitly guided both revenue and margin to sequential improvement for the first time in three quarters.
June guide range bottoms at $158M, below this quarter's $163.8M. If the print comes in at the low end, the trough call fails and the market reprices the three-quarter margin decline as structural rather than cyclical.
“Overall, total March quarter revenue was $163.8 million, down 0.5% year-over-year and up 0.9% sequentially.”
Changed less than 94% of calls
Wording compared with the Feb 5, 2026 call, ranked among 97 calls reporting on Q1 2026. A measure of how much the story moved, not a signal about the stock.
Figures are extracted from the transcripts and checked in code: every quote appears word for word. Language change is computed from word frequencies, with no model involved. Descriptive only: none of this feeds the verdict.
Advanced computing more than doubled sequentially and grew 40% YoY to 25% of computing. This was faster than management's own models predicted. Medium voltage MOSFETs for hot swap and IBC are in build phase at ODMs serving hyperscalers, and AOS holds the 48V-to-12V sockets today with a path to 800V by 2027. Management says these carry margins higher than some power IC products. June guides to $168M and 23% gross margin, first sequential improvement in both metrics in three quarters, with half from durable mix. Communication segment grew 18.7% YoY on BOM content expansion with tier one US smartphone customers on rising charging currents. ASP erosion slowing. Balance sheet is clean: inventory down, DSO improved to 20 days, $190M cash. December and March were the bottom.
The AI story is $20M of a $163.8M quarter. The business paying for R&D is falling apart: Power IC down 20.3% QoQ and 14.1% YoY while lower-value DMOS grows. Gross margin declined three straight quarters. Operating expenses up 12% YoY. Loss per share tripled from $0.10 to $0.28 in one year. Two quarters of negative operating cash flow. EBITDA down 60% YoY. CFO confirms continued ASP erosion and foundry cost increases with no pass-through. The CEO hedges visibility in the same breath as touting backlog. The guidance range itself contains a downside case below the current quarter. The 800V transition in 2027 is an unaddressed platform risk where AOS could lose content. Half the guided margin recovery is utilization, a cyclical fact.
Say/Do Gap
CEO claims backlog gives good visibility for medium voltage, then says second half visibility remains limited, both in the same prepared remarks. If backlog is strong, visibility should be clear.
Zero Pricing Power Despite Constraints
CFO confirms ASP erosion continues and foundry costs are rising with no pass-through mechanism. Chinese MOSFET competition presses the commodity side while input costs climb.
AI Label Without Substance
Advanced computing is $20M of $163.8M. It doubled off a base too small to offset Power IC declining 14% YoY. The 2027 acceleration promise is the only unhedged number in the call.
HOLD. Wait for the June print. The setup is real: advanced computing doubling, mix-driven margin guide, capacity expansion. But three quarters of margin decline, two of negative cash flow, and a guidance range that bottoms below this quarter mean I cannot size the position until the trough confirms. Full position requires $168M at 23% with positive operating cash flow and a second consecutive quarter of margin advance. If June prints at the low end of the range, the bottom call collapses and the P&L deterioration becomes the thesis.