Dossier · AOSL · Dormant
AOSL
Last analysed · · source: theme_discovery
Current thesis
Power-discrete AI-content second-derivative play; B. Riley PT raise to $25 (2026-04-13) + two-session momentum tape (2026-04-14, 2026-04-20) = narrative discovery phase into FY26 Q3 print (~2026-05-06). Binary on earnings; no confirmed AI design win yet.
Invalidation trigger
Weekly close below $20 on no news, OR FY26 Q3 revenue miss vs consensus, OR Q4 guide implies <5% YoY growth, OR China export-control headline hitting Chongqing JV.
Thesis status
Open commitment catalyst duescored if the trigger above fires How this is scored →Current Thesis
Power-discrete second-derivative name on AI-server + AI-PC silicon content. B. Riley PT hike to $25 (2026-04-13) and intraday strength 2026-04-14 + inclusion in "12 IT stocks moving" 2026-04-20 tape suggests a multi-day narrative leg forming into the May FY26-Q3 print. Still DORMANT-leaning-ACTIVE: the bull leg is real but sell-side rating stayed Neutral, and we have zero company-confirmed AI-server design wins. Watch for a confirmed clean above-$25 breakout on RVOL >1.5 OR an earnings beat-and-raise; otherwise it's a B-tier semicap-equipment proxy trading on peer-group heat, not stock-specific alpha.
Bull Case
- 2026-04-13: B. Riley Securities maintained Neutral but raised PT to $25 — first tangible sell-side datapoint validating AI-server power-content narrative on AOSL specifically.
- 2026-04-14: Benzinga flagged "AOSL soaring" + grouped with Avanos/Travere/Globalstar as big Tuesday movers — momentum body-language consistent with a retail narrative discovery phase.
- 2026-04-20: Re-appearance in "12 IT stocks moving in Monday's intraday session" (Benzinga) — not a one-day pop, narrative has a second leg.
- Power-semi content per AI rack is expanding: Nvidia Blackwell GB200/GB300 boards require ~30-50% more 100V/150V MOSFET content vs H100 gen due to higher current density; AOSL's portfolio overlaps that socket.
- JV-fab operating leverage: Chongqing 12" JV hits incremental GM drop-through above ~75% utilization; any sequential recovery print compounds into EPS.
- Small float asymmetry: ~$700M mkt cap, ADV light enough that one clean earnings beat + guide raise can re-rate 25-40% in a session (cf. AOSL's own history — multiple 15%+ gap-and-go sessions post-print over past 6 quarters).
Bear Case
- Sell-side rating still Neutral — $25 PT implies recovery earnings, not multiple expansion. Low-quality bull signal; Buy-list upgrades from ≥2 firms is the actual confirmation we want.
- No disclosed AI-server design win — the entire AI-content narrative is market-implied from peer-group extrapolation. If Q3 commentary is silent on data-center/AI, the leg breaks.
- China / Chongqing JV risk — any re-escalation of export controls or Taiwan tension hits AOSL's supply chain + Asia distribution disproportionately vs. ON/Infineon/STM.
- Commodity pricing pressure — low-voltage MOSFETs are the most commoditized corner of power discrete; ASP erosion is the base case, not the exception.
- Peer-group proxy, not leader — AOSL tends to move AFTER MPWR/ON/POWI; buying AOSL in April means you're the second or third wave, with weaker price structure and higher crash-back risk.
Setup & Price Structure
- No live tape in this synthesis — operator must pull real-time quote + 20/50/200-day MAs + RVOL before any entry decision.
- Anchor levels: $25 = B. Riley PT (2026-04-13) and the sell-side "fair value" ceiling. A weekly a daily close below the thesis-invalidation level on RVOL >1.5 = thesis-confirming breakout. A a daily close below the thesis-invalidation level invalidates the April leg.
- Entry discipline: only trigger if price is above the 50-day MA AND within 7% of a 20-day high on confirming volume. Do NOT chase a >8% gap-up — reversal base rate is high on small-cap Benzinga-mention names.
- Archetype-2 sizing rules apply: probe 1-2%, can scale to 3% post-earnings confirmation. No SUPREME sizing until company-confirmed AI design win.
- No averaging down — if first entry trips the [entry redacted] invalidation, exit and wait for a fresh base.
Catalyst Calendar (next 30 days)
- ~2026-05-06 (est., confirm via IR): FY26 Q3 earnings (March-quarter). Historical pattern = first week of May post-market. This is the binary. Beat + guide = re-rate; in-line = drift; miss = stop out.
- ~2026-05-19 to 2026-05-22: Computex Taipei. Nvidia/AMD/Intel platform announcements historically create 2-3 day reflexive bids in power-semi peer group; AOSL typically moves with group even without direct mention.
- Ongoing through May: Sell-side channel checks for May earnings are the main pre-print information vector — watch for any AOSL-specific note from Craig-Hallum, Needham, or Stifel (not just B. Riley) as the marginal upgrade signal.
What Would Change Our Mind
- Q3 revenue beat >5% AND Q4 guide implies >10% YoY growth → upgrade DORMANT → ACTIVE, initiate 2-3% probe, scale on [entry redacted] breakout retest.
- AOSL-specific AI-server design-win disclosure (Nvidia/AMD reference board, hyperscaler sole-source) → conviction jumps to HIGH, size to 4%.
- Second sell-side upgrade to Buy within 14 days (not just PT raise) → confirms narrative clustering, thesis accelerating.
- a daily close below the thesis-invalidation level on no news → thesis broken, remove from active watch, re-seed only on a new base.
- Guide-down or China-export headline → instant skip, theme reassessment.
- Theme transition: if semicap-equipment flips from ACCELERATING to SATURATED (sell-side universal upgrades, retail EFT flow peak), AOSL as a laggard proxy becomes a trim, not an add.
Correlation Notes
- High beta to: MPWR (Monolithic Power), ON (ON Semi), POWI (Power Integrations) — AOSL lags these by 1-3 sessions and amplifies the move.
- Theme peers: IFX.DE (Infineon), STM (ST Micro), ALGM (Allegro MicroSystems) — watch for relative-strength confirmation; if peers fade while AOSL holds, that's actually a bullish divergence.
- Upstream correlation: NVDA/AMD data-center revenue prints create reflexive bids in power-semi supply chain 1-5 sessions after; NVDA next print is the macro tell.
- Negative correlation: rising USD-CNY + China export-control headlines = immediate de-rating risk.
- Liquidity risk: ADV low enough that a single institutional exit can move price 4-6% intraday; scale entries, don't market-buy size.
Pipeline notes
- "Earnings blackout: no new entry within 3 trading days of FY26 Q3 print (~2026-05-06) — confirm exact date via IR before committing capital.", Archetype-2 proxy name, not a leader — AOSL lags MPWR/ON/POWI; only add on relative-strength confirmation against peer group., "Small-cap liquidity: ADV light; scale entries, avoid market orders on size.", B. Riley rating still Neutral — $25 is sell-side ceiling, not a base. Need a second firm upgrade to Buy for narrative cluster confirmation., No AOSL-specific AI-server design win disclosed — entire AI narrative is peer-group implied. Q3 call commentary is the key tell.
Related · shared themes
MRVL
Google-Marvell TPU/AI-chip partnership talks (2026-04-20) just added a third hyperscaler ASIC customer on top of AWS Trainium + MSFT MAIA — narrative velocity stepped up overnight, and Q1 FY27 print ~2026-05-28 is the binary that either validates the custom-silicon royalty story or breaks it.
CLS
AI-hyperscaler networking picks-and-shovels narrative re-accelerating as sell-side chases higher (BofA $430, JPM $410, Cowen $350 all raised in last 4 sessions), but Q1 2026 print inside 7 trading days = binary gate. Defer aggressive sizing until post-print reaction confirms.
SMTC
AI-connectivity pick-and-shovel: CopperEdge ACC attaches to every GB200/GB300 rack. Two PT hikes in 72h (B. Riley $122, Benchmark $120) + spotlight-list mention signal pre-print re-rate; 2026-06-04 FQ1 FY27 print is the binary, with hyperscaler capex commentary 2026-04-23 to 2026-05-01 as the accelerant.
AMKR
Picks-and-shovels AI-packaging play (#2 global OSAT, Arizona CHIPS-Act fab + Apple multi-year deal) but narrative is MATURING not accelerating, OSAT margin structure caps re-rating, and Q1 '26 earnings are inside the 3–10 trading-day binary window (historical print late-April). Defer until post-print setup.