ON Semiconductor

Phase A — Understand the business

Lens 1 · Company Overview

What it is. ON Semiconductor Corporation (brand: onsemi, Nasdaq: ON, incorporated Delaware, CIK 0001097864) designs and manufactures power and sensing semiconductors — the unglamorous analog/discrete silicon that moves, converts, and senses energy in cars, factories, and increasingly data centers. It is not a logic/compute company (no CPUs, GPUs, or leading-edge digital); its value is in intelligent power (switching efficiency, thermal handling, wide-bandgap materials) and image sensing. As of Dec 31 2025 it runs three reportable segments:

• PSG — Power Solutions Group — $2,805.1M FY2025 rev, 46.8% of total. Discrete & integrated power: MOSFETs, IGBTs, and the crown jewel EliteSiC silicon-carbide power for EV traction inverters. The growth engine and the margin volatility source.
• AMG — Analog & Mixed-Signal Group — $2,261.9M, 37.7%. Analog, power management, the Treo scalable analog platform, connectivity. The profit engine — 51.1% gross margin, expanding even through the downturn.
• ISG — Intelligent Sensing Group — $928.4M, 15.5%. CMOS image sensors (CIS) for ADAS/automotive vision, plus SiPM/SPAD/SWIR. The most damaged segment — margin collapsed to 15.1% in FY2025.

How it makes money. Sells chips to automotive and industrial OEMs and their contract manufacturers, both direct and (majority) through distributors. Distributors were ~54% of FY2025 revenue, and one distributor was ~11% of total revenue. A meaningful slice of automotive volume runs through long-term supply agreements (LTSAs) — multi-year, sometimes fixed-price commitments that lock customers in but expose onsemi to cost inflation and, critically, can be amended or cancelled by customers when demand softens (explicitly flagged as a risk). No dividend has ever been paid since IPO; capital return is 100% buyback.

End-market concentration. Automotive is the single largest end-market and was the dominant driver of the FY2024-25 decline across all three segments. The thesis is and always has been a levered bet on auto + industrial semiconductor content growth (electrification, ADAS, factory automation), now with an emerging AI-datacenter power kicker.

Lens 2 · Supply Chain

Map: substrate/material → wafer fab (largely in-house) → assembly/test → distributor → OEM/Tier-1 → end product. onsemi is an IDM (integrated device manufacturer) — it owns fabs, which is the whole point in power/SiC where process is the moat.

• Upstream inputs: silicon wafers, silicon-carbide (SiC) substrates (the scarce input for EliteSiC — onsemi vertically integrated here via the 2021 GT Advanced Technologies acquisition, $415M, giving it captive SiC boule/substrate growth), GaN, rare gases, photomask/consumables. SiC substrate supply is the historic chokepoint of the entire industry.
• Fabs (chokepoints, named): East Fishkill, NY (EFK — 300mm, AMG/ISG/PSG); Gresham, Oregon; Rožnov pod Radhoštěm, Czech Republic; Seremban, Malaysia (Site 2); plus South Korea (largest PP&E base at $1,175.3M net) and Philippines back-end. PP&E by geography: South Korea $1,175.3M, US $1,046.8M, Czech Republic $398.0M, China $186.0M, Philippines $169.8M.
• Manufacturing strategy — "Fab Right": onsemi is deliberately consolidating its fab footprint to drive a structurally higher-margin model. The 2025 Manufacturing Realignment Program cut ~1,500 employees and took a $496.0M non-cash equipment impairment. This is the source of both the FY2025 margin pain (underutilization) and the FY2026+ margin recovery thesis.
• Downstream: distributors (~54% of revenue, one at ~11%) → OEMs and Tier-1 suppliers (Bosch, Continental, ZF, etc.) and auto OEMs directly. Named strategic customer: Volkswagen Group (multi-year EliteSiC traction-inverter sourcing agreements, 2023 and expanded 2024).
• Single-source dependencies: captive SiC substrate is a strength (supply security) but also a fixed-cost millstone in a downturn. The pending Synaptics acquisition adds a new downstream surface (consumer/IoT/PC connectivity) outside the auto/industrial supply chain.

This lens passes the "names or it didn't happen" test: real fabs, real geographies, real customer (VW), real distributor concentration — all filing-sourced.

Lens 3 · Competitive Advantages (moats)

Moat profile: moderate, process-and-design based, not fortress-grade. onsemi competes in a brutal, cyclical, capital-intensive oligopoly.

1. Process & wide-bandgap IP (the real moat). EliteSiC is a top-tier SiC franchise with vertical integration from substrate to module — few competitors own the full stack. The Treo analog platform (single scalable architecture, wide voltage range, high temperature) is a genuine design-cycle accelerator. AMG's durable 51% gross margin through a brutal downcycle is the clearest proof of a real moat in the analog business.
2. Switching costs / design-in lock. Automotive power and sensing parts are designed in for the life of a vehicle platform (5-7+ years) and qualified under AEC-Q100 — once you win the socket, you keep it. LTSAs formalize this. This is onsemi's strongest moat: high, but it cuts both ways (losing a next-gen socket is equally sticky in reverse).
3. Scale in SiC. ~25% global SiC power market share — the clear #2 behind STMicroelectronics (~32-35%), ahead of Infineon (~15%) and Wolfspeed (~11%). Scale matters for yield-learning and capacity.

Bargaining power: weak-to-balanced. Against auto OEMs (VW, Bosch) onsemi is one of several qualified suppliers — and onsemi is described in the SiC market as the most aggressive price-competitor, willing to trade margin for sockets. That is a tell of limited pricing power, not strength. Distributor concentration (one at 11%) is a mild dependency. The moat is real in AMG, contested in PSG (SiC price war), and weakening in ISG (Sony pressure).

Lens 4 · Segments

Revenue, gross profit, and gross margin by segment — every figure:

The story the segment table tells:

• Two-year revenue erosion of ~27% ($8.25B → $6.0B) — a textbook auto/industrial semiconductor downcycle (inventory correction after the 2021-23 shortage bubble).
• The margin damage is concentrated and explained: PSG GM fell 16.8pts and ISG GM collapsed 31.6pts — both hit by (a) underutilization as volumes fell against a fixed fab base, and (b) a $268.2M excess-and-obsolete inventory charge, of which $230.3M landed in ISG and $37.9M in PSG. ISG's collapse is largely a one-time inventory event layered on cyclical weakness, not permanent.
• AMG is the anchor: revenue fell with the cycle but gross margin actually rose (50.1% → 51.1%) — the structural profit engine and the reason the whole company didn't go cash-flow negative.

Geography (revenue booked by ship-to, i.e. distribution hubs — NOT end-demand): Hong Kong 27.3%, UK 22.5%, Singapore 20.9%, US 20.5%. These reflect distributor/EMS locations, not where the cars are sold — read end-demand through the automotive/industrial mix, not the geography line.

Trend: decelerating decline → inflecting. Q1 FY2026 shows PSG re-accelerating and ISG margin snapping back (see Lens 5). The segment data is the spine of the bull case: the damage was cyclical + one-time, the recovery is underway, and AMG never broke.

---

Phase B — Measure performance

Lens 5 · Earnings Result — Q1 FY2026 (quarter ended April 3, 2026) is the inflection print

GAAP P&L:

Non-GAAP: revenue $1,513M beat the guidance midpoint; non-GAAP EPS $0.64 vs $0.61 consensus (+5%); non-GAAP gross margin 38.5%; non-GAAP operating margin 19.1%.

What drove it:

• Automotive +5% YoY — the first annual growth after seven straight quarters of decline. This is the single most important data point in the dossier: the auto inventory correction is over.
• AI-datacenter power revenue doubled YoY and grew >30% QoQ — a new, secular, non-cyclical demand vector layered onto the auto recovery. onsemi power content is being designed into AI server power-delivery.
• PSG revenue +14.2% YoY ($736.6M, now 48.7% of total) with gross margin recovering to 27.2% (from 19.1% in the prior-year quarter).
• ISG gross margin swung from (55.6)% to +39.4% YoY — the prior-year quarter carried the big inventory charge; the snap-back confirms the FY2025 ISG collapse was largely non-recurring.
• AMG steady at ~53.6% gross margin — the anchor holds.

Balance-sheet flags: cash $2,003.6M (down from $2,147.6M at year-end); inventory ticked back up to $2,049.2M (from $1,989.6M) — worth watching, but consistent with restocking into a recovery; LT debt $2,982.9M (flat); operating cash flow $239.1M (down from $602.3M Q1'25 — weaker, partly working-capital timing on the inventory build). Continued buyback: $346M repurchased in the quarter.

Guidance — Q2 FY2026: revenue $1.535-1.635B (midpoint ~$1.585B, slightly above the ~$1.59B consensus); non-GAAP EPS $0.65-0.77 (midpoint $0.71); management flagged exiting $30-40M of non-core revenue (the Fab-Right portfolio pruning continues). Tone: cautiously confident — calling Q1 an "inflection."

Market reaction: the Q1 beat itself drew a muted −4.6% on May 4 (recovery was already priced after the YTD run). The real move came later — see Lens 8 / Synaptics.

Verdict on the print: unambiguously the bottom. Revenue growth, margin recovery, the auto turn, and a brand-new AI-datacenter vector all confirm in one quarter. Still GAAP-loss-making only because of the (deliberate, winding-down) restructuring charges.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the research-layer shelf (transcripts/ empty); this lens is ``, labeled.

• Tone trajectory (FY2024 → Q1 FY2026): a clean arc from "managing the downturn / disciplined under-shipping" (2024-25 calls emphasized inventory digestion, utilization cuts, "Fab Right" discipline, defending margin) → "inflection" on the Q1 FY2026 call (May 2026), where management explicitly called the bottom and pivoted the narrative to growth vectors: AI datacenter, auto re-acceleration, and Treo/EliteSiC design wins.
• Recurring phrases that intensified: "intelligent power and sensing," "Physical AI," "AI datacenter," "structural margin model," "Fab Right." The "Physical AI" framing is new and is the strategic throughline that the Synaptics deal is meant to complete.
• What they stopped saying: the defensive "inventory correction / utilization headwind" language that dominated 2024-25 receded in Q1 2026.
• Credibility check: management has consistently guided to a structurally higher gross-margin model (mid-40s+ target) coming out of the trough; Q1's 38.5% and the Q2 guide (implied ~39-40%) are tracking toward that — guidance has been roughly met, not serially missed. The Synaptics announcement, however, is a narrative risk — it muddies the clean "focused intelligent-power" story management spent five years building.

Lens 7 · Comps

Peer set: large analog/power/auto-semi IDMs and the SiC cohort. Multiples are ; market caps . Where a specific figure was not sourced cleanly, it is marked n/a.

* onsemi's trailing ROE is distorted to near-zero by the FY2025 trough ($121M net income on ~$7.7B equity ≈ 1.6%); a normalized mid-cycle ROE on ~$1.8B operating income would be ~18-20%.

Reading the table:

• On forward P/E, onsemi (~35x) screens expensive vs NXP (~19x), roughly in line with TXN (~35x) and Infineon (~34x), and cheaper than STM (~56x). But forward P/E is contaminated by where each name sits in its own earnings recovery — onsemi's denominator (FY2026 EPS ~$3.18 consensus) is depressed/recovering, so the multiple overstates richness.
• On EV/EBITDA (~17.8x), onsemi is the cheapest of the group alongside NXP (~17.2x) and well below TXN (~31x), Infineon (~25x), STM (~21x). This is the bulls' anchor: on cash earnings, onsemi is the value name in the cohort.
• The honest read: onsemi trades at a mid-cycle-trough discount on EV/EBITDA but a recovery-inflated forward P/E. The valuation case rests entirely on the EPS recovery being real (Lens 11) — at FY2027 consensus EPS $4.39, the multiple compresses to ~20x, which is cheap for the #2 SiC franchise if the cycle and the AI-datacenter vector deliver and if Synaptics doesn't break the story.

Lens 8 · Stock-Price Catalysts (5-year pattern)

52-week range: $74.52 – $133.93 (all-time high $133.93 on June 3, 2026). Stock ~$88.58 at June 29 close; +113.8% YTD 2026 before the June drawdown. What moves this stock:

• June 26 2026 — Synaptics deal: −22% to −24% intraday ("worst day since March 2024"). The single largest move — a strategy/dilution shock, not an earnings event.
• Q1 FY2026 earnings (May 4 2026): −4.6% despite the beat — recovery already priced.
• The 2026 YTD melt-up (+114%): a pure cyclical-recovery + AI-datacenter-power re-rate, mirrored by STM (+177% YTD) — the whole auto/industrial-semi complex re-rated as the bottom came into view.
• March 2023 — Tesla SiC-reduction scare: Tesla's announcement it would cut SiC content ~75% in next-gen powertrains hammered the SiC cohort (onsemi, Wolfspeed, STM). Shows the stock's sensitivity to single-customer/SiC-demand headlines.
• 2023-24 downcycle grind: steady de-rating on the auto/industrial inventory correction and seven quarters of declining auto revenue.
• 2021-22 shortage boom: the stock ran with the semiconductor shortage and the SiC/EV narrative (GT Advanced acquisition, VW EliteSiC win Jan 2023).

Pattern: the market reacts to (1) the cycle (auto/industrial demand inflections — the dominant driver), (2) SiC-demand and single-customer headlines (Tesla, VW), (3) AI-datacenter content news (the new 2026 vector), and now (4) capital-allocation/strategy shocks (Synaptics). It is a high-beta cyclical that the market prices on forward demand inflection, which makes the current moment — bottom confirmed, but strategy clouded — genuinely two-sided.

---

Phase C — Judge people & books

Lens 9 · Management

CEO: Hassane El-Khoury (President & CEO since December 2020; ~5.5-year tenure).

1. Track record — strong, proven value-creator. Before onsemi he was CEO of Cypress Semiconductor, where he ran the "Cypress 3.0" transformation — exiting slow-growth commodity memory, refocusing on automotive/industrial and higher-margin franchises — and sold Cypress to Infineon for ~$9B (April 2020). He has run exactly this playbook before and exited it at a premium.
2. At onsemi — the same playbook, executed. Repositioned a sprawling, low-margin discrete house into a focused "intelligent power + sensing" company: divested non-core fabs and businesses, built EliteSiC into the #2 SiC franchise, won VW, and drove the "Fab Right" footprint consolidation. Gross margin went from low-30s pre-El-Khoury toward a mid-40s structural target (peaked 47% in 2023 before the cycle). This is a genuine, quantified turnaround, not a story.
3. Capital allocation — aggressive, shareholder-return-tilted, now making a big swing. No dividend ever; capital return is all buyback — $1,377.6M repurchased in FY2025 alone (27.9M shares), plus $654M FY2024, $564M FY2023 — i.e. buying back stock through the trough, which is contrarian and (if the recovery holds) value-accretive. Bolt-on M&A has been disciplined and on-strategy (GT Advanced $415M for SiC substrate; Qorvo SiC JFET $118.8M). The Synaptics deal is a sharp departure in scale and direction — $7B (vs prior sub-$500M bolt-ons), all-stock, and off the pure auto/industrial focus into consumer/edge. It is the single biggest test of his capital-allocation judgment to date.
4. Red flags — modest. No related-party / promotional-behavior flags surfaced. The one tension: the Synaptics pivot partially contradicts the five-year "focus on high-margin auto/industrial" narrative — and importing Synaptics' cyclical consumer exposure is exactly what Cypress 3.0 / onsemi's strategy was built to avoid. Insider-ownership detail not on the shelf (insider-transactions.csv absent) — n/a.
5. Archetype: professional operator-CEO / serial transformer (not founder). For this stage — a cyclical IDM trying to compound out of a trough into secular AI-power demand — that is the right archetype. The question is whether a proven focuser should be making a TAM-expanding diversification bet.

Net: high-quality, proven management with a credible turnaround already banked. The Synaptics deal is where the track record gets re-underwritten — bulls trust the operator; bears think he just blinked.

Lens 10 · Forensic Red Flags

Acting as a forensic analyst over the income statement, balance sheet, and cash-flow statement [all figures research-layer: filings unless labeled]:

• Restructuring/impairment as a recurring "below the line" item. FY2025 carried $666.9M in "Restructuring, asset impairments and other, net" (incl. the $496.0M equipment impairment and $268.2M inventory write-offs), and Q1 FY2026 carried another $329.3M. These are real cash and non-cash costs that management consistently excludes from non-GAAP. The gap between GAAP EPS ($0.29 FY2025; $(0.08) Q1'26) and non-GAAP EPS (positive) is large and persistent — the single most important thing to watch. It is currently defensible (a genuine multi-year footprint consolidation), but if "one-time" charges keep recurring quarter after quarter, the non-GAAP number is fiction. Track whether the charges actually wind down through FY2026 as guided.
• Cash flow vs earnings — reassuring. FY2025 operating cash flow was $1,759.8M against GAAP net income of just $123.6M — the divergence is favorable and explained (the $496M impairment is non-cash, D&A was $686.0M, working capital released). This is the opposite of the classic red flag (earnings > cash). OCF resilience through the trough is a quality signal.
• Inventory — watch item. Inventory was worked down to $1,989.6M at FY2025 year-end but ticked back up to $2,049.2M in Q1 FY2026. In a recovery this is normal restocking; if revenue stalls, it becomes the next write-off. Days-of-inventory remain elevated vs history. Yellow flag, not red.
• Receivables: fell to $908.0M (from $1,160.1M) — moving with revenue, no channel-stuffing signal.
• Goodwill/intangibles: goodwill $1,679.9M, intangibles $343.9M — modest relative to $12.5B assets; no impairment taken on goodwill (only on equipment), reasonable. The Synaptics deal will add several billion in goodwill/intangibles — a future impairment risk if the deal underperforms.
• Stock-based comp / non-GAAP flattering: SBC is added back to non-GAAP as standard; the dominant non-GAAP adjustment here is restructuring, not SBC, so SBC isn't the main distortion. Diluted share count is falling (buybacks > SBC issuance) — a positive.
• Leverage: LT debt ~$2.98B against ~$2.0-2.5B cash and ~$1.76B OCF — comfortable; net debt ~$0.4-1.0B. The all-stock Synaptics structure avoids adding debt (a point in its favor).

Regulatory findings (required sub-section):

• SEC Litigation Releases: None.
• SEC AAERs: None.
• Item 3 (Legal Proceedings), FY2025 10-K: generic ordinary-course-litigation language; the company notes complex proceedings are hard to predict and damages may be unquantifiable, but discloses no specific material adverse proceeding. No reportable contingency of note.
• Non-SEC / civil: a shareholder class action was filed (the Rosen Law firm solicitation, dating to ~2023) — routine post-decline securities solicitation, no indication of material liability. No FTC / DOJ / antitrust action found for onsemi in 2024-2026.
• Conclusion: No material regulatory or forensic accounting findings — verified via SEC EDGAR EFTS (LR, AAER = 0), 10-K Item 3, and web search as of 2026-06-30. The one genuine forensic watch-item is the persistence of "one-time" restructuring charges widening the GAAP↔non-GAAP gap — a quality-of-earnings flag, not a fraud flag.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 → FY2028)

Built bottom-up from Q1 FY2026 actuals + Q2 guidance, extended through the cyclical recovery. All EPS figures are non-GAAP (the metric Street and management use); every input labeled.

Revenue path FY2026: Q1 $1,513.3M (actual) + Q2 ~$1,585M (guide midpoint) + Q3 ~$1,650M + Q4 ~$1,700M (continued cyclical recovery + AI-datacenter ramp, net of ~$30-40M/qtr non-core exits) ≈ $6.45B, +7.6% vs FY2025.

Cross-checks: Base FY2026 ~$2.98 sits just below Street consensus $3.18 — I'm deliberately conservative on the H2 ramp. Base FY2027 ~$4.30 ≈ Street $4.39. Synaptics caveat: the deal closes mid-2027, adds ~12% share dilution but also Synaptics' earnings + a guided $200M cost synergies within 18 months; the net FY2027-28 EPS effect is genuinely uncertain and I have not baked a precise Synaptics number into the base — flag it as a wide error bar.

Valuation at $88.58: ~30x base FY2026 / ~21x base FY2027 / ~17x base FY2028. The forward multiple normalizes fast if the recovery delivers — ~21x FY2027 for the #2 SiC franchise with an AI-datacenter option is reasonable-to-attractive, but it requires the cycle to keep turning and the deal not to break the model.

Tracked forecast (Brier): per --watchlist rules, not logging a forecast.ts create in unattended breadth mode. Suggested base call to log later if promoted to a thesis: "ON FY2026 non-GAAP EPS ≥ $2.90, p≈0.62, resolves 2027-02-28."

Lens 12 · Bull vs Bear

Bull case (narrative). onsemi is a confirmed cyclical bottom dressed up as a falling knife. Q1 FY2026 proved it: revenue grew, gross margin snapped back 18 points, automotive turned positive for the first time in seven quarters, and a genuinely new secular vector — AI-datacenter power, doubling YoY — appeared on the income statement. The "Fab Right" restructuring that gutted FY2025 margins is precisely the lever that drives the structural margin model higher on the way out. AMG's unbroken 51% margin proves the analog moat is real. Management has run this turnaround playbook before and sold the last one to Infineon for $9B. On EV/EBITDA (~17.8x) it's the cheapest quality name in the cohort. And the Synaptics deal, bought at a 19% premium in an all-stock structure that adds no debt, expands the TAM by $30B to $243B and plants onsemi at the center of "Physical AI" — power + sense + compute + connect — for cents on the dollar of a panicked tape. The 24% drop is the opportunity.

Bear case (narrative). This is a high-beta cyclical that has already run 114% YTD to an all-time high pricing in a full recovery — and management chose that exact moment to make a $7B all-stock bet that nobody asked for. The Synaptics deal imports cyclical consumer-electronics exposure into a business whose entire five-year thesis was escaping consumer cyclicality, dilutes holders ~12%, won't close until mid-2027, comes with $200M of synergies but conspicuously thin revenue detail, and will bury several billion of goodwill on the balance sheet to impair later. Meanwhile the core is not invincible: onsemi is the price-taker in SiC (the aggressive discounter, not the leader — STM has ~33% share to onsemi's ~25%), Sony is pressuring image sensors (ISG), Chinese power-semi capacity is flooding in, and the auto "recovery" is one VW order-cut headline away from reversing. The GAAP↔non-GAAP gap is enormous and the "one-time" charges keep coming. At ~30x recovering EPS, you're paying a growth multiple for a cyclical commodity-power business whose CEO just signaled he'd rather diversify than focus.

Pre-mortem (18 months out, thesis broke): It's December 2027. Auto recovery stalled at "L-shaped" — content growth couldn't offset flat unit volumes and Chinese SiC undercut EliteSiC pricing, holding PSG margin in the mid-20s. The AI-datacenter revenue, real but small, never scaled into a needle-mover. The Synaptics integration absorbed two years of management attention, the synergies slipped, consumer end-markets dragged the blended margin down, and the market re-rated onsemi to a lower multiple as a now-less-focused, more-cyclical conglomerate. The stock sits in the $60s.

Are multiples too high? On forward P/E (~30x recovering EPS), yes — vulnerable. On normalized FY2027 (~21x) and EV/EBITDA (~18x), no — fair-to-cheap. The valuation is a bet on which earnings number is real.

Contrarian view (what the market is refusing to see): The market is treating the Synaptics deal as a 24% negative — pure dilution and distraction. The contrarian read: management is using an over-valued, post-melt-up stock as currency to buy a depressed asset that completes the "Physical AI" platform — the textbook right time to do an all-stock deal (issue expensive equity, buy cheap). If the integration works and the cycle holds, the deal that crashed the stock will look prescient. The market is also under-weighting how non-cyclical the AI-datacenter power vector could become — it's the one part of the story that doesn't depend on car volumes.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• Revenue concentration / single-customer fragility. ~54% of revenue runs through distributors, one at ~11%, and a chunk of auto volume sits under LTSAs that customers can amend or cancel. A single VW or distributor de-stock event resets the recovery. The 2023 Tesla SiC scare showed how violently this stock reacts to one customer's powertrain decision.
• The moat is thinner than bulls claim — in the segment that matters most. PSG (47% of revenue, the "growth" engine) is where onsemi is the #2 SiC player and the acknowledged price-discounter — "willing to trade margin for sockets". That is not pricing power; that is a margin-at-risk commodity fight against STM (the share leader), Infineon (300mm + Wolfspeed substrate leverage), and a wall of subsidized Chinese SiC capacity. The 24.5% PSG gross margin in FY2025 is what a price war looks like.
• The most dangerous competitor bulls underestimate: Sony (in ISG) and the Chinese SiC bloc (in PSG). Sony's CIS dominance is squeezing onsemi's image-sensing franchise; Chinese power-semi makers are commoditizing the low-to-mid end. ISG already collapsed to 15% gross margin once.
• Worst capital-allocation move — happening right now. The $7B all-stock Synaptics deal is the bear's gift: it dilutes ~12%, pivots away from the focused high-margin strategy, drags in cyclical consumer exposure, offers $200M synergies with thin revenue logic, and won't close for a year — a year of integration distraction at the exact moment the core cycle is turning. Buying "cash-strapped Synaptics" with your own stock is not obviously brilliant.
• What must hold for today's price (~$88.58 / ~30x): (1) auto recovery continues and accelerates; (2) AI-datacenter power scales into a real number; (3) PSG margin recovers despite the SiC price war; (4) Synaptics closes, integrates, and accretes rather than dilutes. All four must work. Break any one and the ~30x forward multiple is indefensible.
• If growth disappoints 20-30%: FY2026 revenue ~$5.5-5.8B instead of ~$6.4B → non-GAAP EPS toward the bear ~$2.40-2.70 → at a de-rated ~18x cyclical multiple, the stock is $45-50 — i.e. ~40-45% downside to a plausible disappointment case.
• The single scenario that permanently impairs the business: a structural SiC price collapse driven by Chinese over-capacity + a fundamental shift away from SiC in EV powertrains (the Tesla 2023 thesis, generalized) — which would strand onsemi's vertically-integrated, fixed-cost SiC investment (the GT Advanced substrate base, the dedicated fabs) as a margin millstone. Plausibility: moderate — SiC content per EV is still rising industry-wide, but the pricing is the vulnerability, and onsemi has positioned itself as the price-follower.

Lens 14 · Management Questions (ordered by information value)

1. Synaptics — revenue synergy math: You guided $200M of cost synergies. What is the revenue synergy thesis, quantified, and why does owning Synaptics' consumer/HMI/connectivity franchise beat a partnership or licensing deal — given five years spent reducing consumer-cyclical exposure?
2. The deal's effect on the structural margin model: Synaptics' consumer mix carries different margins and cyclicality than auto/industrial. What is the pro-forma blended gross-margin target for the combined company, and does it raise or lower your mid-40s structural goal?
3. PSG pricing in SiC: You're described as the most aggressive SiC price-competitor. What is your floor on PSG gross margin, and how do you defend EliteSiC pricing against Chinese capacity and STM's scale without conceding share?
4. AI-datacenter power — durability and size: Datacenter power doubled YoY. What is the FY2027-28 revenue run-rate you see for this vector, who are the design-win customers, and how cyclical/concentrated is it?
5. The GAAP↔non-GAAP gap: Restructuring charges have run $666.9M (FY2025) + $329.3M (Q1'26). Give a hard date by which these "one-time" charges end and GAAP and non-GAAP converge.
6. Inventory: Inventory ticked back up in Q1 to ~$2.05B. What days-of-inventory are you targeting by year-end, and at what demand level does the current inventory become a write-off risk?
7. LTSA durability: What fraction of revenue is now under LTSAs, what is the weighted duration, and how many have customers sought to amend or cancel in the last 12 months?
8. Capital allocation priority post-Synaptics: With ~12% dilution and an all-stock structure, how do buybacks (>$1.3B/yr recently) and any future M&A get re-prioritized against integrating Synaptics and deleveraging the share count?
9. ISG / Sony: Image-sensing margin collapsed and Sony is pressing. Is ISG a long-term core franchise or a divestiture candidate in a future "Fab Right" round?
10. Fab Right end-state: What is the target fab footprint and utilization at mid-cycle, and how much more impairment/restructuring is left in the program?
11. Automotive recovery shape: Q1 auto grew 5% YoY. Is this content-driven (electrification/ADAS dollars per car) or unit-volume-driven, and what's your visibility into LTSA backlog vs spot demand?
12. Treo platform monetization: How much of AMG's 51% gross margin and design-win pipeline is attributable to Treo, and what's the attach/cross-sell trajectory?
13. China exposure both ways: China is a demand source and a competitive threat. What share of revenue is China end-demand, and how exposed is the EliteSiC franchise to Chinese local-for-local substitution?
14. Goodwill risk: Synaptics will add several billion in goodwill/intangibles. Under what scenarios would that be impaired, and what's the sensitivity?
15. Long-term strategy / 5-year picture: Five years from now, is onsemi a focused intelligent-power-and-sensing pure-play, or a broader "Physical AI" platform company — and which valuation multiple should the market apply?

---