Autonomous Machines & Automotive Semiconductor Deep Dive

Overview & Framework

This research examines two complementary positions in the autonomous machine supply chain: ON Semiconductor (the eyes — image sensors, depth sensing) and NXP Semiconductors (the brain and nervous system — MCUs, radar, networking, edge AI). Both companies derive 50-58% of revenue from automotive, but the thesis extends directly into robotics, drones, and industrial automation — a TAM expanding from ~$70B to $200B+ by 2030.

All financial data is sourced from SEC EDGAR (10-K, 10-Q, 8-K filings) via XBRL company facts API and direct filing text extraction. Competitive intelligence is synthesized from public secondary research, industry reports, and product announcements.

Toll-Road Summary

Part 1: ON Semiconductor — Sensing & Vision

Ticker: ON (Nasdaq) | CIK: 0001097864

Q2 2026 Earnings: Reports Aug 3, 2026 (after close)

Executive Summary

ON Semiconductor (onsemi) is a Tier 3 WEAK PROXY under the toll-road framework (composite 7 → downgraded one tier for conglomerate dilution). The sensing business (ISG — Intelligent Sensing Group) is a genuine bottleneck in automotive ADAS image sensors with ~35-40% market share, but it represents only 15.7% of total company revenue. The dominant power solutions (PSG) and analog (AMG) segments carry significant cyclical, competitive, and restructuring headwinds that dilute the sensing thesis.

1.1 SEC EDGAR Financials

All data sourced directly from SEC filings (CIK 0001097864) via XBRL company facts API and 10-K/8-K exhibits.

Revenue Trajectory

Filing URLs:

• 10-K FY2025: Open SEC filing
• 10-Q Q1 2026: Open SEC filing
• 8-K Q1 2026 earnings: Open SEC filing

Revenue by End-Market (FY2025 vs FY2024 vs FY2023)

Auto revenue fell 21.0% from FY2024 to FY2025 ($3.90B → $3.08B). Auto was 51.4% of FY2025 — ON remains heavily auto-dependent.

Revenue by Segment

ISG (Sensing) Detail: FY2025 revenue $928.2M (15.7% of total), gross profit ~$139.9M → 15.1% GAAP GM. The GM collapse was driven by $230.3M excess & obsolete inventory charge and $496.0M non-cash impairment from the 2025 Manufacturing Realignment Program. Underlying post-restructuring ISG GM: estimated ~40%.

Gross Margin Trajectory

EPS Trajectory (GAAP Diluted)

Operating Cash Flow & CapEx

FCF surged to $1.4B in FY2025 as CapEx was slashed post-restructuring. ON bought back $346M in Q1 2026 (160% of FCF).

Q1 2026 Segment Revenue (from 8-K EX-99.1)

Q2 2026 Guidance

1.2 Q2 2026 Earnings Preview — What to Watch

ON reports Q2 2026 today (Aug 3) after market close. Key items:

1. AI data center revenue doubling trajectory — Management targeted doubling in 2026 vs. >$250M FY2025 baseline. Q1 showed >30% QoQ growth.
2. Auto end-market commentary — Is the inventory correction definitively over?
3. SiC utilization and margins — 150mm→200mm transition and "Fab Right" consolidation critical for margin recovery
4. Q3 and full-year 2026 guidance — Beat-and-raise vs. beat-and-maintain
5. China exposure commentary — Any update on Chinese OEM demand, OmniVision dynamics
6. Share buyback pace — $346M in Q1 at 160% of FCF

Sentiment: Elevated call buying ahead of the print. ON has beaten in 5 of last 7 quarters. Requires convincingly strong guide to justify upside given rich valuation.

1.3 Automotive Image Sensor Market Share & Competitive Position

The automotive CMOS image sensor (CIS) market is a three-player oligopoly:

Sony — The Biggest Threat:

• Targeting 43% auto CIS market share by FY2027
• ¥1T+ Kumamoto fab investment + Japanese government ¥60B subsidy (April 2026)
• TSMC JASM partnership (55K wpm, targeting 100K+ wpm)
• Key sensor: IMX728 (8.4MP) directly competes with ON's AR0820/Hayabusa

Chinese Competitors:

• SmartSens: Growing auto revenue (107% cited in 2025), releases vehicle-grade sensors
• GalaxyCore: Primarily mobile — limited auto design wins
• OmniVision: ~35% of Chinese auto CIS market; geopolitical headwinds in Western markets

1.4 Content-Per-Vehicle Thesis

• Auto CIS market CAGR: 9-16% through 2035 (2-4x total expansion)
• ON's estimated per-vehicle sensor content: ~$9-10 today → $14-16 by 2030
• ISG FY2025 auto revenue estimated at ~$650-700M

1.5 Auto Cycle Recovery Signal

• ON's auto revenue fell from $4.32B (FY2023) to $3.08B (FY2025) — -28.7% peak-to-trough
• Q1 2026: First YoY growth (+4.7%) after 7 declining quarters
• CEO: "We have moved beyond the cyclical trough on a path to recovery"
• Peer read-throughs: TXN, NXPI, MPS, STM all beat and raised
• Verdict: Inventory correction is ending. Recovery is real but measured.

1.6 Key Risks

1. Chinese Competition (HIGH — structural): ON's China auto power share fell from 5.9% → 3.2%. BYD vertical integration. Chinese SiC substrates at $400/wafer vs $1,200 Western.
2. Sony Encroachment (HIGH — secular): 43% share target by FY2027, ¥1T+ capex, IMX728 competing directly.
3. SiC Division Drag (MODERATE-HIGH): PSG GM ~24.5% in FY2025. 150mm→200mm transition costs. Peer STM had -21.4% Power & Discrete op margin in Q2 2026.
4. ISG Margin Sustainability: Can ISG return to 45%+ GM vs. Sony competition?
5. Concentration Risk: One distributor = 11% of revenue. Auto = 51.4%.

1.7 Toll-Road Classification

Standalone ISG: Would be Tier 1 DIRECT BOTTLENECK (Q1:3, Q2:3, Q3:3 = 9) if it were a pure-play. The SiC/power cyclicality and AMG commoditization drag down the composite.

1.8 Bull vs Bear

Bull Case:

• Auto inventory correction definitively ended (Q1 2026 inflection)
• ISG GM recovers toward 45%
• AI data center power doubles in 2026 ($250M → $500M+)
• Fab Right + 200mm SiC transition delivers H2 2026 margin expansion
• Geopolitical decoupling shifts Western ADAS share from OmniVision to ON + Sony
• $6B buyback at 160% of FCF is highly accretive

Bear Case:

• Sony's 43% share target + ¥1T capex = structural share loss over 3-5 years
• ISG is only 15.7% of revenue — thesis diluted by commoditized power/analog
• SiC margins remain under severe pressure
• China exposure is structural, not cyclical
• GM recovery to 38.5% still well below FY2023's 47.1%
• One distributor = 11% of revenue = concentration risk

Part 2: NXP Semiconductors — The Nervous System

Ticker: NXPI (Nasdaq) | CIK: 0001413447

Q2 2026 Earnings: Reported July 28, 2026 (beat) | Q3 Guidance: Raised

Executive Summary

NXP Semiconductors is a Tier 1 DIRECT BOTTLENECK under the toll-road framework (composite 8 — Q1:3, Q2:2, Q3:3). NXP is the dominant supplier of automotive microcontrollers (MCUs), vehicle networking, radar, and secure connectivity — the "nervous system" of every modern vehicle. With automotive representing ~58% of FY2025 revenue ($7.1B of $12.3B), NXP is the most auto-pure-play among the major analog/power semiconductor peers. Q2 2026 results delivered a comprehensive beat-and-raise: revenue $3.50B (+19% YoY), non-GAAP GM 58.0%, and Q3 guidance of $3.65-3.85B. The stock sold off ~13% post-earnings on inventory concerns — creating a potential entry point.

2.1 SEC EDGAR Financials

All data sourced directly from SEC filings (CIK 0001413447) via XBRL company facts API, 10-K, and 8-K exhibits.

Revenue Trajectory

Filing URLs:

• 10-K FY2025: Open SEC filing
• 10-Q Q2 2026: Open SEC filing
• 8-K Q2 2026: Open SEC filing

Revenue by End-Market (FY2025)

Auto held remarkably stable through the downturn — down only 0.5% vs. peers' 15-28% declines. This is the empirical proof of content-growth decoupling.

Q2 2026 Segment Performance

Revenue by Geography (FY2025)

China grew +6.0% YoY — the ONLY geography to grow in FY2025. Q2 2026 direct China: $630M (+25.5% YoY). Total China exposure estimated at 30-36%.

Gross Margin Trajectory

Operating Margin

*Q1 2026 GAAP op margin was 47.3% — inflated by ~$627M MEMS sensors divestiture gain.

EPS & Free Cash Flow

2.2 Q2 2026 — The Sell-Off Puzzle

NXP delivered a comprehensive beat: revenue $3.496B (+19% YoY), EPS $3.61, GM 58.0%, raised Q3 guidance. The stock fell -13%. Why:

1. Channel inventory rose to 11 weeks (from 9 weeks YoY)
2. Internal DIO elevated at ~156 days
3. GAAP EPS declined sequentially ($4.43 → $3.02 — Q1 had a one-time divestiture gain)
4. Mobile softness (-10% QoQ)
5. $750M debt repayment consumed cash
6. High expectations — stock was +26% YTD

CEO Quote: "AI is moving from the cloud into the physical world — into vehicles, factories, and robots — and it lands directly in the markets where NXP has leadership positions."

2.3 Competitive Position — The "Nervous System"

NXP does NOT make SiC power devices (ON/STM/Infineon), image sensors (ON/Sony), or high-power analog (TI). It makes the compute, networking, and sensing processing:

The S32 Platform Moat: Unified architecture across MCU/processor lines, common software framework, S32 CoreRide pre-integrated platform, cloud-based virtual development. Once an OEM adopts S32 for one domain, incremental adoption is much easier.

Reported Ambarella Talks (Unconfirmed): Secondary reporting indicated talks around a potential ~$3.3B transaction for CVflow AI vision SoCs. Neither company had announced a transaction as of the research date. If completed, it would fill NXP's biggest gap: dedicated computer vision/ADAS perception processing.

2.4 Content-Per-Vehicle Thesis

Q2 2026 auto revenue +12% YoY despite flat global vehicle production. Drivers:

• NXP's estimated dollar content: ~$250-450 per vehicle
• 2026 volume ramps: S32N (5nm), SAF8444 single-chip radar, S32K5 MCUs

2.5 Auto Cycle Recovery — Further Along Than ON Semi

NXP's auto business was essentially flat through the worst semiconductor downturn since 2008 — the strongest empirical evidence for the toll-road thesis.

2.6 Key Risks

1. Qualcomm/NVIDIA on S32 (MODERATE): Compete at central compute layer, not NXP's zonal/networking/radar layer
2. China Localization (MODERATE): Horizon Robotics, BYD Semi gaining. But NXP's China revenue is GROWING (+25.5% in Q2)
3. TI Fab Buildout (MODERATE): $5B/year on 300mm fabs. NXP competes on integration/software, not unit cost
4. Distribution Inventory (MODERATE): 11 weeks channel inventory could be over-ordering
5. Potential Ambarella Transaction Risk: Reported talks, if they result in a transaction, would add execution risk in a crowded ADAS vision market

2.7 Toll-Road Classification

2.8 Bull vs Bear

Bull Case (Tier 1 — Up to 20% Depth):

• Proven toll-road: auto revenue flat through worst downturn since 2008
• Content growth decoupling: S32, radar, SDV driving 5-10% annual growth
• Margin expansion: GM approaching 60%, op margins approaching 37%
• Physical AI tailwind: edge processing + connectivity are picks-and-shovels
• Q2 sell-off (-13%) creates potential entry point
• Q3 guidance implies acceleration: $3.65-3.85B (+15-21% YoY)

Bear Case:

• Channel inventory at 11 weeks — could signal over-ordering
• Internal DIO at 156 days — persistent manufacturing overhang
• China localization: Horizon Robotics, BYD Semi threats
• TI fab capacity: $5B/year could pressure MCU pricing
• Qualcomm/NVIDIA could capture central compute, reducing NXP to lower-value networking

Part 3: Beyond Automotive — Robotics, Drones & Autonomous Machines

Executive Summary

The automotive sensing/compute thesis extends directly into a much larger and faster-growing market: autonomous machines — robots, drones, AMRs, humanoids, and industrial automation. Both NXP and ON Semiconductor have clear migration paths, playing complementary roles in the same NVIDIA Jetson-centric ecosystem. Combined, auto + robotics/industrial represents a TAM expansion from ~$70B to $200B+ by 2030.

3.1 Market Sizing — Non-Automotive Autonomous Opportunity

The critical insight: sensing content per robot is structurally similar to sensing content per vehicle, but unit volumes could be 10-100x larger. ~70M vehicles/year vs. potentially hundreds of millions of robots/drones over the next decade.

3.2 ON Semiconductor — The Eyes of Autonomous Machines

Migration Path: Auto Sensors → Robot Sensors

Key Robotics Products

Hyperlux ID — 3D Depth (Launching 2025-26):

• AF0130/AF013x: 1.2MP iToF depth sensor, range up to 30m
• On-chip dual laser drivers, 200 MHz modulation
• Applications: Factory AMRs, drone mapping, warehouse robots, metrology
• Integrated in e-con Systems DepthVista Helix
• Eliminates external PCs — single-sensor depth solution

Hyperlux SG — High-Speed 2D:

• AR0235: 2.3MP global shutter, 120 fps
• For drones at speed, pick-and-place robots, conveyor inspection

Hyperlux LP — High-Res for Jetson:

• AR2020 (20MP): Standard on NVIDIA Jetson AGX Orin via e-con Systems e-CAM200_CUOAGX
• AR0830 (8MP): Jetson Orin via Leopard Imaging LI-AR0830 modules
• ON is the #1 image sensor supplier to the Jetson robotics ecosystem

The Jetson Ecosystem Lock-In

• NVIDIA Jetson Orin (67-70 TOPS, $249) and Jetson Thor (7.5× Orin, $3,499) are dominant robot compute platforms
• Early Thor adopters: Amazon Robotics, Boston Dynamics, Figure, Agility Robotics, Caterpillar, Meta, Medtronic
• ON sensors have first-class driver support in the Jetson stack (MIPI CSI, GStreamer, TensorRT)
• Every robot built on Jetson is a potential ON sensor socket

Estimated robotics/industrial sensing revenue: $150-250M today, growing at 12-20%+ CAGR

3.3 NXP — The Brain & Nervous System of Autonomous Machines

Migration Path: Vehicle Controllers → Robot Brains

i.MX + Kinara Edge AI Platform

• Kinara Ara-2 NPU: 40 TOPS — supports vision, transformers, generative AI
• i.MX 95 / i.MX 8M Plus: Application processors with integrated NPUs
• M.2 AI accelerator modules (Geniatech): Drop-in 40 TOPS for existing industrial systems
• This is NXP's direct answer to NVIDIA Jetson at the lower-power/lower-cost end

The NXP-NVIDIA Complementary Dynamic

• NVIDIA Jetson = AI brain (GPU-heavy, high TOPS, expensive)
• NXP i.MX + Kinara = Efficient edge brain (lower power, safety-certified, 10-15 year lifecycle)
• NXP MCUs/PMICs/Ethernet = Nervous system surrounding BOTH
• Many robots use NVIDIA for perception + NXP for control in the same system

NXP Non-Auto Revenue — The Robotics Signal

The Industrial & IoT segment surged +38% YoY in Q2 2026 — the non-auto growth engine is firing. Annualized run-rate approaching $3B.

3.4 The Architecture Convergence

A modern EV and a warehouse AMR share nearly identical architectures:

3.5 Competitive Dynamics

ON Semi vs Sony in Robot Vision:

• ON leads in practical, deployable robot vision (Hyperlux ID depth, global shutter, Jetson ecosystem)
• Sony leads in absolute image quality and overall CIS R&D
• They serve different market segments

NXP vs NVIDIA in Robot Compute:

• NVIDIA dominates heavy AI perception (2,070 TOPS on Thor)
• NXP dominates efficient, safety-certified edge compute (40 TOPS, 10-15 year lifecycle)
• Complementary, not competitive

Chinese Competition:

• OmniVision: Robotics/drone sensors but geopolitical headwinds
• DJI: Vertically integrated for drone cameras
• Horizon Robotics: Competes with NXP in Chinese factory edge AI
• Risk: Chinese robot/drone makers prefer domestic suppliers

3.6 TAM Expansion

3.7 Investment Implications

For ON Semiconductor: The robotics opportunity partially addresses ON's biggest weakness — ISG at 15.7% of revenue. If Hyperlux adoption grows ISG from $928M to $1.5B+ (17-25% CAGR), ISG crosses the 20% threshold, eliminating the conglomerate dilution penalty. Catalyst: ISG exceeding $1.1-1.2B.

For NXP: The robotics opportunity reinforces the Tier 1 thesis. Industrial & IoT already 18.5% of revenue, growing +38% YoY. The i.MX + Kinara platform creates a second toll-road beyond automotive — same switching costs apply. Catalyst: Industrial & IoT crossing $3.5-4B annual run-rate.

Part 4: The Competitive Landscape — Why ON & NXP Win

The autonomous machine semiconductor space has eight major players. Six compete with or complement ON and NXP. Here's how they stack up — and why ON and NXP are our preferred exposures.

Competitive Map

4.1 Infineon (IFNNY/IFX) — The Auto Semiconductor King

Position: #1 in automotive semiconductors by revenue (13.5% share), #1 in auto MCUs (~36% share via AURIX), leader in SiC power.

Why Not Our #1 Pick:

• Conglomerate: Infineon is a €50B European industrial conglomerate. Auto semis are the largest division but it's not a pure play.
• Accessibility: Trades primarily on Xetra (IFX.DE). US ADR (IFNNY) has lower liquidity. Not ideal for a US-centric portfolio.
• SiC margin drag: Same transition pain as ON Semi (150mm→200mm), similar utilization headwinds.
• No sensing differentiation: Infineon doesn't make image sensors — it competes with NXP in MCUs and with ON in SiC, but doesn't have either company's unique sensing/compute moat.

Where Infineon Wins: If you want a diversified auto semiconductor leader with #1 share in both MCUs and SiC power, Infineon is that company. But it's a different thesis — broader, less concentrated, harder to access.

4.2 Texas Instruments (TXN) — The Margin King (Wrong Thesis)

Position: #1 in analog semiconductors. Auto is ~20-25% of revenue. GM 61.4%, op margin 42.3%.

Why Not Our Pick:

TI is arguably the best-run semiconductor company on the planet. Its margins (61% GM, 42% op margin) destroy NXP (58% GM, 35% op margin) and ON (38% GM). But:

• TI is an analog company, not an auto company. Auto is only ~20-25% of revenue. You're buying exposure to industrial, data center, personal electronics, and everything else. The auto thesis is diluted.
• No image sensors, no radar, no MCU leadership. TI competes in power management and embedded processing — not in the sensing/compute layers that define the autonomous machine thesis.
• TI's auto content is generic. Power management chips and signal chain analog are essential but commoditized — they don't create the software lock-in and platform moats that NXP's S32 or ON's Hyperlux ecosystem provide.
• Capex intensity: TI is spending ~$5B/year on 300mm fabs. This is a long-term strategic advantage but creates near-term FCF drag.

Verdict: TI is a magnificent company but the wrong vehicle for an autonomous machine thesis. If you want broad analog exposure with auto tailwinds, buy TXN. If you want concentrated auto/robot sensing and compute, buy ON and NXP.

4.3 STMicroelectronics (STM) — The SiC Drag Is Worse

Position: #2 in SiC power, auto MCUs (STM32), MEMS sensors. Q2 2026 revenue $3.49B (beat).

Why Not Our Pick:

• Gross margin is terrible: 34.8% vs NXP's 58% and TI's 61%. The SiC transition drag is worse at STM than at ON.
• SiC segment is loss-making: Reported negative operating margins in power/discrete.
• Post-earnings massacre: -15% to -18% despite a beat — the market is punishing SiC-exposed names.
• MEMS is a commodity: STM's MEMS sensors (accelerometers, gyroscopes) are important but not differentiated — NXP and ON have stronger sensing moats.
• No radar leadership, no S32-equivalent platform, no image sensor #1 position.

Verdict: STM has the worst of both worlds — SiC margin drag like ON but without ON's sensing differentiation, and MCU exposure like NXP but without NXP's platform moat. Pass.

4.4 Sony (SONY) — The Threat, Not the Opportunity

Position: ~50% overall CMOS image sensor market share. Targeting 43% auto CIS by FY2027. IMX728 competes directly with ON's AR0820.

Why Not Our Pick:

• Sony is a $120B+ conglomerate. Image sensors are within Sony Semiconductor Solutions — one division of a massive entertainment/electronics/gaming/financial services company. You cannot buy Sony as an auto sensing pure play.
• Sony IS the existential threat to ON's thesis. It's the competitor, not the investment. Understanding Sony's trajectory is critical to sizing the ON position correctly, but Sony itself is an uninvestable vehicle for this thesis.
• Auto CIS is still small for Sony: Mobile CIS dominates Sony's sensor revenue. The auto ramp is a growth story but not the primary driver of Sony's stock.

Verdict: Track Sony to monitor the competitive threat to ON Semi. Do not buy Sony for auto sensing exposure — it's far too diluted by PlayStation, movies, music, and mobile sensors.

4.5 Renesas (RNECY) — NXP's Closest Competitor

Position: #2 in automotive MCUs (competing with NXP's S32 via R-Car). Q1 FY2026: GM 59.2%, op margin 33.7%. Benefiting from Wolfspeed share gains.

Why Not Our Pick:

• Japan-headquartered: Trades primarily on TSE (6723.T). ADR (RNECY) has thin liquidity. Yen exposure adds currency risk.
• Japanese OEM concentration: Heavy dependence on Toyota, Honda, Nissan — less diversified across global OEMs than NXP.
• R-Car vs S32: Renesas competes credibly but NXP's S32 platform has broader adoption, better software ecosystem (CoreRide), and a clearer SDV roadmap on 5nm.
• Wolfspeed gains are one-time: The ¥63B Q2 finance income from WOLF shares is non-recurring — flatters earnings.
• NXP's networking/radar moats: Renesas doesn't have NXP's #1 auto radar position or industry-leading vehicle networking portfolio.

Verdict: Renesas is NXP's most credible competitor in auto MCUs — strong margins, similar thesis. But inferior platform moat, less diversification, and worse accessibility for US investors. NXP wins on breadth and ecosystem.

4.6 Qualcomm (QCOM) — The Central Compute Challenger

Position: Snapdragon Ride + Cockpit. BMW named QCOM lead compute provider. Auto revenue $1.33B (+38% YoY), approaching $7B annualized.

Why Not Our Pick:

• Auto is ~5% of Qualcomm's revenue. The company is primarily a smartphone modem/SoC company ($35B+ total revenue). The auto thesis is exciting but it's a tiny slice of a much larger, handset-cyclical business.
• Competes at the WRONG layer. Qualcomm's Snapdragon Ride targets the central ADAS compute socket — competing directly with NVIDIA Drive Orin/Thor. This is the most competitive, most expensive, and most commoditizing layer of the stack. NXP competes at the zonal/networking/radar layer — complementary, less contested.
• NVIDIA is winning central compute. Jetson Thor has Amazon Robotics, Boston Dynamics, Figure, Agility. NVIDIA's ecosystem and developer mindshare in AI/robotics are overwhelming. Qualcomm is a distant #2.
• China headwinds: Chinese OEMs (BYD, Xiaomi, Huawei) are developing in-house ADAS SoCs — Qualcomm's share in China is eroding.

Verdict: Qualcomm's auto story is real and growing fast ($7B run-rate). But it's the wrong layer (central compute vs. zonal/networking), too small relative to the handset business, and facing the NVIDIA buzzsaw. NXP is the better way to play auto compute without the NVIDIA/Qualcomm bloodbath.

4.7 Summary — The Winner Matrix

The Thesis in One Sentence

ON Semiconductor gives you the eyes of every autonomous machine — the #1 auto image sensor supplier, deeply embedded in the NVIDIA Jetson robotics ecosystem, with a recovery tailwind and a robotics growth option. The SiC/power drag keeps it at Tier 3 (5% max), but if ISG crosses 20% of revenue, it upgrades to Tier 2.

NXP Semiconductors gives you the brain and nervous system — the #1 auto MCU/radar/networking supplier, proven toll-road through the worst downturn since 2008, with Industrial & IoT surging +38%. Tier 1 DIRECT BOTTLENECK (20% max). The cleanest expression of the autonomous machine thesis in public markets.

Together they cover every chip in a modern vehicle or robot except the central AI processor (that's NVIDIA's domain) and the high-power analog (that's TI's).

The other six are either too diluted (TXN, QCOM, SONY), too hard to access (IFNNY, RNECY), too margin-challenged (STM), or competing at the wrong layer. They're useful to track — but for a concentrated portfolio, ON and NXP are the right pair.

Combined Portfolio Framework

Position Sizing

Complementary Coverage

A combined position gives you:

• ON Semi: The eyes of every autonomous machine (auto + robot + drone)
• NXP: The brain and nervous system (compute + networking + control)
• Combined: Full autonomous machine semiconductor stack

Watch Items

Data Quality & Sources

Primary Sources (HIGH confidence)

• SEC EDGAR: All revenue, GM, EPS, segment, end-market, and geographic data from 10-K, 10-Q, and 8-K filings via XBRL company facts API (SEC XBRL data)
• ON Semi CIK 0001097864: 10-K FY2025, 10-Q Q1 2026, 8-K Q1 2026 earnings
• NXP CIK 0001413447: 10-K FY2025, 10-Q Q2 2026, 8-K Q2 2026 earnings

Direct links to the filings, earnings releases and official product evidence are collected in the report source library.

Secondary Sources (MEDIUM confidence)

• Public secondary-source synthesis: Market share estimates, competitive dynamics, China exposure, SiC/peer economics, robotics ecosystem, content-per-vehicle estimates
• Consistent across multiple analyst/investor sources (Yole Group, Counterpoint, Strategy Analytics)
• NVIDIA Jetson ecosystem data from vendor announcements and partner pages

Methodology

• Toll-Road Classification: Per the PXS Research public-market methodology
• Financial Analysis: XBRL API for exact GAAP figures + 8-K exhibits for non-GAAP reconciliations and guidance

This research is for informational purposes and does not constitute investment advice. All data sourced from SEC filings and public market commentary as of August 3, 2026.