ON Semiconductor Corporation

ON Semiconductor — Business Economics

Ticker: ON | Currency: USD | Most recent data: FY2025 (10-K, Dec 31, 2025)

ON Semiconductor sells power and sensing semiconductor chips. The economic engine is currently weakening cyclically — FY2025 revenue fell materially from the FY2022–2023 peak — but the company is restructuring aggressively to reposition toward higher-value secular growth vectors (SiC for EV/data centers, AI power management, ADAS imaging).

How it makes money: Revenue comes from selling physical semiconductor components — power MOSFETs, SiC devices, analog ICs, and image sensors — to automotive OEMs, industrial customers, and (increasingly) AI data center operators. Three segments: PSG (47% of FY2025 revenue), AMG (38%), and ISG (15%). Distribution is split between direct OEM relationships and third-party distributors. This is a capital-intensive, design-win-driven business where revenue visibility extends 1–3 years via socket wins but volume is tied to end-market cycles.

Direction of travel: The business is in a painful cyclical trough. The 2025 Manufacturing Realignment Program cut ~2,400 jobs, took $496M in asset impairments, and wrote off $268M in inventory. Revenue declined from approximately $8.3B (FY2022) to roughly $5.6–5.9B in FY2025. However, the strategic pivot toward SiC power (vertically integrated from substrate to module), AI data center power delivery, and ADAS sensing represents a legitimate structural growth thesis. The Qorvo SiC JFET and Aura Semiconductor Vcore acquisitions both target the AI power tree.

Win-win assessment: Genuinely win-win. ON's chips enable EVs to drive farther, data centers to consume less energy, and vehicles to be safer. Customers need these components; ON is not extracting rent from a captive base.

Key governing metrics:

• SiC revenue growth (proxy for EV/energy transition traction)
• Gross margin % (target: 50%+; tests mix shift away from commodity silicon)
• Automotive revenue concentration (currently ~50%; higher exposure = higher cyclicality risk but also higher ASPs)
• Design-win pipeline (forward revenue indicator)
• Free cash flow conversion (given heavy capex in SiC fabs)

Signs of deterioration: ISG (sensing) is clearly the weakest segment — $230M inventory write-off signals demand materially below expectations. The broader revenue decline is cyclical, not structural, but the depth of the cut (~30% peak-to-trough) is notable even by semiconductor standards.

Market Overview — ON Semiconductor

ON Semi plays in two structural growth markets — power semiconductors (silicon + SiC) and intelligent sensing — with 80%+ revenue tied to automotive and industrial end markets. The combined addressable market ON targets is ~$30–35B today, growing at high-single-digit CAGR through 2030.

Tailwinds are real but not uncontested. EV electrification drives SiC demand (800V architectures require 2–3× more SiC content vs. silicon IGBTs). AI data center power delivery is an emerging growth vector. ADAS sensor content per vehicle continues compounding. These are durable, multi-year secular trends — not cyclical bounces.

Competition is fierce. The power semiconductor space is moderately consolidated but intensely competitive:

ON's vertical integration in SiC (substrate through module) is a differentiator, but pricing pressure is intensifying as capacity comes online industry-wide. The market rewards scale and yields — ON must execute on its East Fishkill SiC fab ramp to remain cost-competitive.

Net assessment: Large TAM with genuine structural tailwinds, offset by meaningful competitive intensity and looming SiC oversupply risk in 2026–2027.

ON Semi's moat rests primarily on vertical SiC integration and automotive design-in switching costs—both genuine but neither unassailable.

The SiC advantage is structural: ON controls substrate-to-device manufacturing (via the 2021 GTAT acquisition), one of only ~3 Western companies with internal SiC boule production at scale. This yields cost/supply advantages and 2–3 year qualification cycles lock customers in once designed-in. The 2025 Qorvo SiC JFET acquisition further deepened the portfolio for AI data center power.

However, Infineon and STMicro possess comparable capabilities, and Chinese SiC entrants are scaling rapidly. The legacy analog/discrete business (AMG) operates in a commodity-adjacent space where TI's scale dominance creates persistent pricing pressure. The massive ISG restructuring ($230M inventory write-off) reveals a segment that lacked defensible positioning entirely.

The moat is narrow but widening in SiC/power; eroding elsewhere.

Financial Strength — ON Semiconductor

Verdict: A cyclically strong balance sheet with prudent capital structure, temporarily obscured by massive one-time restructuring charges in FY2025.

ON Semi's through-cycle ROIC of ~20-25% (FY2022-2023 peak: 30%+) comfortably exceeds its cost of capital and ranks among the best in analog/power semis. The current trough compresses this to low-teens, but the structural improvement since 2020 (shift to differentiated products, fab-lite optimization) is real and durable.

The debt structure is remarkably capital-efficient: predominantly 0% and 0.50% convertible notes due 2027/2029, plus 3.875% senior notes due 2028 — total long-term debt ~$3.5B against normalized annual EBITDA of $2.5-3B. Cash interest expense is negligible relative to cash flow. The $1.5B undrawn revolver (fully repaid in Dec 2025) provides ample downturn cushion.

FY2025 earnings quality caveat: GAAP results are severely distorted by ~$980M in restructuring/impairment charges ($496M asset impairments, $268M inventory writedowns, $67M severance, $149M other). Stripping these, underlying FCF conversion remains healthy at 80-90%. The ISG inventory writedowns ($230M) signal management over-invested in image sensing capacity they're now abandoning — a capital allocation miss, not a going-concern issue.

The $1.375B in buybacks during a downturn year is aggressive but defensible given liquidity. No off-balance-sheet concerns, no auditor changes, no customer concentration above ~10%.

ON Semi's reinvestment runway is genuine but narrower than management's peak-cycle rhetoric suggested. The SiC and AI data center power markets offer legitimate multi-year growth vectors, but recent capital deployment reveals a management team that over-invested at the cycle peak and is now restructuring aggressively.

The company spent ~$1.4–1.9B annually in capex during 2022–2023 to build SiC capacity (Hudson, NH mega-fab), then slashed to ~$700M in FY2025 as utilization collapsed. ROIC peaked near 25%+ in FY2023 and has likely fallen to single digits in FY2025 after restructuring charges. Return on incremental capital from 2023→2025 is negative — earnings fell far faster than the capital base shrank.

The buyback-heavy capital return in 2023 (~$2.8B near peak earnings at ~$80+ prices) destroyed value. The 2025 buybacks ($1,375M at ~$49 average) are better-timed but funded during a period of weakened cash generation, adding leverage risk.

The runway exists — SiC for EVs, GaN and Vcore for AI power — but management's capital allocation discipline has been inconsistent, and the path back to 20%+ ROIC requires a revenue recovery that remains uncertain.

ON Semi is a credible #3 in power semiconductors globally but faces formidable scale disadvantages versus Infineon and increasingly aggressive competition in its key SiC growth vector.

Competitive Landscape: Infineon is the dominant power semi player (~€16B revenue, ~2.5x ON's scale), with the broadest SiC/IGBT portfolio and deepest automotive relationships. STMicroelectronics is the closest analog—similar automotive SiC exposure, similar Tesla reliance, and similar margin compression in the downturn. Texas Instruments competes in power management with a vastly superior analog manufacturing cost structure. Wolfspeed, the SiC substrate leader, filed for bankruptcy in late 2025, which paradoxically benefits ON by removing a vertical integration threat.

Share Trend: ON gained SiC share rapidly from 2021-2024 through its Hudson Fab investment and aggressive LTSA strategy. However, Infineon's massive SiC capex ramp and STMicro's Catania fab expansion are compressing ON's window of differentiation. The EV adoption slowdown exposed ON's customer concentration risk.

ON's premium valuation versus Infineon and STMicro is hard to justify given smaller scale, narrower portfolio, and similar cyclical exposure. Its SiC position is real but not differentiated enough to warrant a structural premium.

Management & Shareholder Orientation

Verdict: Competent operator-management with adequate alignment, but not owner-operators.

CEO Hassane El-Khoury (appointed Dec 2020) has driven a genuine strategic transformation—exiting commodity lines, pivoting to SiC/intelligent power—demonstrating long-term thinking over short-term revenue maximization. His compensation is heavily equity-based with PSUs tied to revenue growth, non-GAAP EPS, and relative TSR, which reasonably aligns incentives.

Skin in the game is moderate but not exceptional. Total insider ownership is ~1–2% of shares—adequate for a $25B+ market cap but not the concentrated conviction you see in founder-led businesses. No pledging concerns have surfaced.

Capital return is a genuine strength. ON has aggressively repurchased shares, shrinking the count from ~435M (2022) to ~394M (Feb 2026)—roughly 9% reduction. The $3B buyback authorization signals continued commitment. No dividend, which is appropriate given reinvestment needs in SiC capacity.

Board is majority independent with no notable related-party concerns. Institutional ownership exceeds 90% (Vanguard, BlackRock, T. Rowe Price among top holders), providing external governance discipline. No SEC enforcement actions against leadership.

Insider selling has occurred periodically (typical for equity-heavy comp plans), though at prices ranging $40–$75—no unusual front-running patterns.

Management Competence & Ethics

Hassane El-Khoury (CEO since Dec 2020) has driven a legitimate strategic transformation but over-promised on near-term targets. Capital allocation has been disciplined: bolt-on acquisitions (GTAT for SiC in 2021, Qorvo SiC JFET for $119M in 2025) are strategically coherent, not empire-building. Share count declined from ~430M to ~394M through buybacks. However, the 2021 Analyst Day target of 53%+ gross margins by 2025 was missed — the cycle intervened, and onsemi took a painful $496M impairment on manufacturing equipment in FY2025 alongside 2,400 layoffs. This suggests prior capex was partially mis-timed, though the restructuring is being handled transparently (charges taken upfront, not dripped). No financial restatements, no auditor disagreements, no fraud history. Clean Section 404(b) attestation. Management earns credit for honest communication about difficulties but loses marks for overly aggressive forward guidance.

ON Semiconductor — Valuation

Verdict: Fairly valued with recovery priced in. Not cheap enough given cyclical risk and execution uncertainty.

The market is paying $28.2B for a business that earned $121M in FY2025 (47x trailing) but is expected to earn ~$1.8B over the next 12 months (16x forward). The entire investment thesis rests on a margin recovery that hasn't fully materialized yet. At the current price, you're buying the promise of normalization, not proof of it.

What's embedded in the price? Forward P/E of 16x implies ~$4.50 EPS on ~390M shares = ~$1.75B net income. This requires revenue recovering to ~$7B with gross margins back near 45% — achievable but not guaranteed given the 2025 trough of 33% gross margins.

Management's long-term model targets 53% non-GAAP gross margin and 40% non-GAAP operating margin. Even at peak (FY2023: $8.25B revenue), ON achieved only 47% gross / 32% operating. These aspirational targets assume significant SiC mix shift and operating leverage. Credibility is moderate — CEO Hassane El-Khoury has executed well on portfolio rationalization but the margin targets have never been hit.

Liquidation value: Tangible book of $5.65B ($14.30/share) provides minimal downside protection at $72.51. This is an earnings-driven stock.

Normalized earnings power: Mid-cycle revenue of $8-9B at 30-32% operating margins yields $1.8-2.2B net income. At 17-18x (appropriate for a cyclical semi with secular tailwinds in SiC/AI power), fair value = $31-40B.

Probability-weighted expected value: ~$38B — implying ~6% annualized return from current levels, with no dividend. For a beta-2 semiconductor cyclical, this risk-reward is uninspiring.

Long-term Valuation

Verdict: Plausible 2–2.5x over 10 years if SiC and AI power execution holds, but this is NOT a high-confidence compounder. The business lacks the pricing power and recurring revenue characteristics that produce multi-baggers with certainty.

Compounding mechanics: ON Semi's flywheel is design-win-driven — automotive qualification cycles (3–5 years) create sticky revenue once won, and SiC vertical integration (own substrates via GTAT) provides cost advantage. But incremental returns on capital are declining: the heavy capex period ($1.5B in 2023) built capacity now underutilized, and normalized ROIC will likely settle at 15–18% versus 25%+ at peak. Each dollar reinvested widens the moat less than the prior dollar.

Multi-bagger math: At $72.51, you're paying ~17x normalized FCF (~$1.7B). If revenue recovers to $9–10B by 2030 (EV + AI power tailwinds), margins normalize at 45% gross/25% operating, and shares shrink 15% via buybacks, EPS reaches $5–6. At 20x = $100–120. Add buyback accretion through the decade: 1.8–2.5x total return — decent but not exceptional.

What erodes it first: Chinese SiC achieving automotive-grade qualification at scale, compressing ASPs. Observable signal: SiC gross margins falling below 40% for two consecutive quarters, or top-3 OEM design-win losses announced.

The filing confirmed key facts. Let me compose the risk assessment based on my knowledge and the filing data.

Risk Assessment — ON Semiconductor Corporation

The existential risk is a failed SiC bet: billions in capex committed to a technology that could be competitively neutralized or architecturally bypassed, leaving ON with stranded assets and insufficient returns on invested capital.

Business risk is the dominant concern. ON is making a concentrated wager on automotive SiC power at precisely the moment Infineon, STMicro, Wolfspeed, and Rohm are doing the same. If EV adoption disappoints or 800V architectures shift to GaN-on-SiC hybrids, ON's capital-intensive fab investments (East Fishkill, the $2B+ expansion plan) become impaired. Customer concentration adds fragility—top automotive OEMs represent outsized revenue share and possess enormous pricing leverage.

Financial risk is moderate. ~$4B in debt (including convertible notes) against ~$1.8B 2025 FCF is manageable at peak, but a prolonged cycle trough compresses margins to 40%+ gross margin territory, stressing coverage. The 0% and 0.50% convertibles maturing 2027/2029 require refinancing or conversion—neither is guaranteed at favorable terms in a downcycle.

Governance risk is low. El-Khoury's restructuring track record is strong. No fraud signals or related-party issues.

External risk is elevated. ~30% revenue exposure to China creates tariff/export-control vulnerability. Geopolitical disruption to automotive supply chains could hit ON disproportionately given its concentration in that end-market.

Probability of permanent impairment: ~15%. The most likely path to permanent value destruction is SiC technology commoditization combined with a multi-year EV demand shortfall—plausible but not probable given structural electrification trends.

Final Verdict: TRACK

ON Semiconductor is an average-quality cyclical business with a narrow moat, undergoing a credible but unproven strategic pivot to SiC/AI power, priced for recovery with insufficient margin of safety.

The investment case reduces to one question: does a ~6-7% expected CAGR (base case $39B market cap by 2031 vs. $28.2B today) adequately compensate for owning a Beta-2 semiconductor stock with narrow moat, concentrated SiC capex bet, and 15% impairment probability? It does not.

What's right: Structural EV/AI power tailwinds are real. Vertical SiC integration creates genuine switching costs. Management is disciplined and shareholder-oriented. $1.4B FCF on $6B revenue demonstrates the asset-light potential once capex normalizes. The business will likely earn $2-3B in net income at cycle peak.

What's wrong: Returns on the massive SiC capex remain unproven — the $496M impairment is evidence of over-optimism. Moat is narrow and segment-specific (SiC only); commodity analog/sensing faces intense competition from TI, Infineon, and Chinese entrants. At 16x forward earnings, the stock prices in a clean recovery already. Beta of 2.02 means you absorb enormous volatility for equity-like returns.

Inversion — how does this fail permanently? Chinese SiC producers (SICC, TanKeBlue) achieve quality parity by 2028, collapsing ON's pricing power. Or EV adoption slows materially, stranding $2B+ in SiC capacity. Neither is probable (combined ~20-25%) but neither is negligible.

For existing holders: Hold through the cycle recovery — the business isn't broken and selling at trough earnings destroys value. But don't add capital here; the risk/reward is symmetric at best.

Entry price: Below $55 (12-13x forward earnings) creates adequate margin of safety for the risk profile. At that level, this becomes a BUY.

What to research further:

• SiC revenue run-rate and design-win pipeline in Q3/Q4 2026 earnings
• Chinese SiC competitor quality certifications (IATF 16949 progress)
• AI datacenter power management revenue traction vs. Monolithic Power and Infineon