PM Portfolio Manager Reports

From Portfolio Manager reports

Edge Compute, Power, Connectivity, And Robot Semiconductors

This node ranks the semiconductors that sit inside a humanoid robot's torso or head compute box, joint and hand control boards, sensor hubs, battery and power tree, charging electronics, and robot-network links. The pieces are edge AI modules, GPUs, NPUs, embedded processors, MCUs, analog front ends, PMICs, power discretes, motor drivers, sensor interfaces, Ethernet, wireless, timing, and safety controllers. They run local inference, close 5-10 ms control loops, translate force, vision, and IMU signals, regulate voltage rails, move data, and keep the robot safe when cloud latency is too slow. The conversion path is design win to qualification to production socket to content per robot to revenue, margin, and free cash flow. NVDA is the platform anchor; TXN, NXPI, QCOM, MPWR, ADI, ON, MCHP, and STM carry the direct board-content basket; AMBA is the specialist edge-vision SoC addition; AMD and AVGO are adjacent AI infrastructure monitors with weaker on-robot attribution.

Current Setup

Production socket gate Robot semiconductor content matters when design wins become qualified production sockets.

Ten xhigh primary agents and one xhigh merge agent added AMBA, kept AMD and AVGO only as adjacent monitors, and ranked direct endpoint silicon above broad AI infrastructure read-throughs.

Platform anchor NVDA has the clearest robot compute platform evidence. Jetson, Thor, Isaac, and GR00T map to local inference and simulation, but reported revenue is still dominated by Data Center.
Board-content core TXN, NXPI, QCOM, MPWR, ADI, ON, MCHP, and STM cover embedded control, signal chain, connectivity, and power. The proof is not a generic chip cycle. It is robot socket evidence plus segment revenue, margin, inventory, and cash conversion.
Attribution limit Most public revenue is still reported through data-center, auto, handset, industrial, or broad analog categories. AMBA is more direct but subscale; AMD and AVGO stay in the basket only as adjacent AI infrastructure monitors.
  1. Paper priorCompute and connectivity are quantified robot cost pools
  2. Design winRobot OEM chooses the endpoint silicon stack
  3. QualificationThermal, battery, latency, and safety tests clear
  4. ProductionSocket volume becomes content per robot
  5. EconomicsRevenue, margin, inventory, and FCF prove conversion

Basket

The ranking below is the tournament merge result from ten xhigh primary research agents plus one xhigh merge agent. It ranks robot endpoint economics and source-backed exposure first, operating proof second, and weekly setup third. Market caps use 2026-07-02 local closes times local weighted shares outstanding. The right rail chart reads the local report API; the page does not embed OHLC payloads or static chart levels.

NVDA NVIDIA Corporation

Best robot-compute platform route through Jetson, Thor, Isaac, GR00T, and the local inference toolchain.

Market cap$4.73T
Latest qtr revenue$81.615B
Weekly setupPullback to rising average

Role in stack

NVDA supplies edge AI compute modules, GPU acceleration, robot simulation, and software tooling that can sit in the robot compute box and in the training and simulation loop around it.

Revenue mix

Q1 FY2027 revenue was $81.615B. Data Center was $75.246B, about 92% of revenue, while Edge Computing was $6.4B and includes robotics, automotive, PCs, workstations, consoles, and AI-RAN.

Proof burden

Robot and edge shipments need clearer isolation from Data Center. Watch named OEM deployments, Jetson/Thor volumes, software attach, and whether robotics becomes visible revenue rather than platform evidence only.

TXN Texas Instruments Incorporated

Broad analog and embedded-control exposure maps to robot power, sensing, motor control, isolation, and board-level management.

Market cap$266.7B
Latest qtr revenue$4.825B
Weekly setupBase forming

Role in stack

TXN sells analog power, signal-chain, isolation, sensing, and embedded processors used across control boards, motor drivers, battery systems, industrial interfaces, and robot support electronics.

Revenue mix

Q1 2026 revenue was $4.825B, with Analog at $3.924B and Embedded Processing at $723M. Q2 guidance was $5.00B-$5.40B, making analog recovery and embedded breadth the first read.

Proof burden

Industrial and automotive sell-through, inventory days, fab utilization, capex downshift, free cash flow, and Silicon Labs integration need to show recovery without forcing a robot-specific revenue assumption.

NXPI NXP Semiconductors N.V.

Strong endpoint-control and connectivity fit through processors, MCUs, secure connectivity, automotive, and industrial control silicon.

Market cap$69.0B
Latest qtr revenue$3.181B
Weekly setupBase forming

Role in stack

NXPI supplies embedded processors, MCUs, networking, secure connectivity, RF, and safety controllers that can support robot control boards, gateways, battery systems, and industrial communication.

Revenue mix

Q1 2026 revenue was $3.181B. The lane highlights Automotive, Industrial and IoT, Mobile, and Communication Infrastructure growth, with channel inventory at 11 weeks and Q2 guide midpoint at $3.45B.

Proof burden

Q2/Q3 guidance, Industrial and IoT demand, automotive breadth, data-center control-plane disclosure, channel inventory, and non-GAAP margin need to validate the rerated embedded-control setup.

QCOM QUALCOMM Incorporated

Direct edge AI and wireless-connectivity candidate for robots, with automotive and IoT disclosure that is more relevant than pure data-center AI proxies.

Market cap$185.8B
Latest qtr revenue$10.599B
Weekly setupBase forming

Role in stack

QCOM brings low-power SoCs, wireless links, edge AI, automotive platforms, IoT processors, and custom silicon options that can sit in robot connectivity, perception, and local compute modules.

Revenue mix

Fiscal Q2 2026 revenue was $10.599B. QCT was 85.6% of revenue; Automotive revenue was $1.326B, IoT was $1.726B, and handsets remained the largest QCT stream.

Proof burden

Auto and IoT growth must offset handset pressure, Apple displacement, memory constraints, and broad custom-silicon claims. Robot relevance improves only with named endpoint platforms or clearer robotics/industrial IoT evidence.

MPWR Monolithic Power Systems, Inc.

Highest-value power-density route for compute boards, voltage rails, battery conversion, and thermal-constrained robot electronics.

Market cap$63.3B
Latest qtr revenue$804.185M
Weekly setupPullback to rising average

Role in stack

MPWR sells high-density power management, regulators, modules, and conversion products that can support local AI compute boards, battery power trees, sensor rails, and charger or motor-control electronics.

Revenue mix

Q1 2026 revenue was $804.185M. The strongest current proof is power-management demand around AI/server and communications equipment, with robot power sockets still mostly an extrapolation from the product role.

Proof burden

Revenue, inventory days, gross margin, enterprise data and communications mix, and disclosed endpoint socket wins need to prove that power density converts into durable revenue outside the AI-server narrative.

ADI Analog Devices, Inc.

Signal-chain, sensing, power, industrial automation, and precision-control exposure fits robot sensor and control boards.

Market cap$183.7B
Latest qtr revenue$3.623B
Weekly setupPullback to rising average

Role in stack

ADI supplies high-performance analog, mixed-signal, RF, power, sensor, DSP, and system solutions for industrial control, sensor translation, data acquisition, communications, and precision motion feedback.

Revenue mix

Q2 FY2026 revenue was $3.623B. Industrial was 50% of revenue, Automotive 24%, Communications 15%, and Consumer 11%; Communications was the clearest AI-linked channel in the filed lane.

Proof burden

Q3 FY2026 revenue versus the $3.9B guide, Industrial breadth, Communications durability, inventory quality, Empower integration, margin, and free cash flow need to justify the premium continuation setup.

ON ON Semiconductor Corporation

Power discretes, SiC, sensing, image-sensing, and motor-control adjacency fit robot power, perception, and motion boards.

Market cap$35.7B
Latest qtr revenue$1.513B
Weekly setupBase forming

Role in stack

ON supplies power semiconductors, SiC devices, sensors, image-sensing, and intelligent power products that can support robot motor drives, battery conversion, perception modules, and industrial controls.

Revenue mix

Q1 2026 revenue was $1.513B. The robot route sits inside automotive, industrial, power, and sensing categories, so the page treats the exposure as direct but still proof-heavy.

Proof burden

Gross margin, GAAP margin, industrial recovery, auto cycle pressure, SiC demand, free cash flow, and absolute AI or power revenue disclosure need to improve together for a stronger robot electronics read.

MCHP Microchip Technology Incorporated

Embedded MCUs, mixed signal, analog, connectivity, timing, FPGA, Ethernet, and PCIe products are highly relevant to deterministic robot control.

Market cap$45.8B
Latest qtr revenue$1.311B
Weekly setupBase forming

Role in stack

MCHP sells MCUs, analog, mixed-signal, connectivity, security, FPGA, Ethernet, and timing products that can sit on joint control, sensor, battery, gateway, and industrial network boards.

Revenue mix

Q4 FY2026 revenue was $1.311B. FY2026 revenue was $4.713B, with product mix at 50.0% mixed-signal microcontrollers, 28.2% analog, and 21.8% other.

Proof burden

Distributor inventory at 26 days, June-quarter FY2027 revenue guide, gross margin, leverage, FCF after dividends, and identifiable AI/data-center or industrial connectivity revenue are the main checks.

STM STMicroelectronics N.V.

Broad MCU, MEMS, power, sensor, custom-processing, RF, and automotive-industrial exposure, with a heavier execution burden.

Market cap$60.7B
Latest qtr revenue$3.095B
Weekly setupOverextension

Role in stack

STM sells MCUs, MEMS and imaging sensors, analog, power discretes, custom processing, RF/optical, and automotive-industrial products that can support robot sensing, control, power, and connectivity boards.

Revenue mix

Q1 2026 revenue was $3.095B. Segment evidence includes AM&S, P&D, EMP, and RFOC, with Q2 revenue guidance of $3.45B and gross margin guidance of 34.8%.

Proof burden

STM needs utilization, P&D repair, gross margin, FCF, data-center revenue quality, MEMS integration, and policy/FX resilience to catch up with the strong stock move.

AMBA Ambarella, Inc.

Specialist edge-vision and low-power AI SoC candidate with direct robotics relevance and a high scale, loss, and concentration proof burden.

Market cap$3.4B
Latest qtr revenue$100.4M
Weekly setupBase forming with volume

Role in stack

AMBA sells low-power vision, video, perception, radar, and edge AI SoCs for IoT, security, robotics, telematics, ADAS, automotive, and edge infrastructure applications.

Revenue mix

Q1 FY2027 revenue was $100.4M. FY2026 revenue was $390.7M, edge AI was about 80% of FY2026 revenue, IoT was about three-fourths of Q1 revenue, and automotive was about one quarter.

Proof burden

Q2 FY2027 revenue versus the $105M-$111M guide, inventory conversion, gross margin, GAAP losses, SBC, WT/customer concentration, CV product ramps, and LTA conversion decide whether design wins become durable economics.

AMD Advanced Micro Devices, Inc.

Adjacent compute monitor: strong AI accelerator and embedded-adaptive exposure, but the current evidence is mostly data-center infrastructure.

Market cap$844.4B
Latest qtr revenue$10.253B
Weekly setupOverextension

Role in stack

AMD provides CPUs, GPUs, AI accelerators, adaptive/FPGA, embedded, and semi-custom compute. The robot endpoint link is mainly embedded and adaptive compute, while the stock debate centers on Data Center AI.

Revenue mix

Q1 2026 revenue was $10.253B. Data Center was $5.775B, 56.3% of revenue, and Embedded was $873M, 8.5% of revenue.

Proof burden

AMD needs Data Center AI margin, OpenAI/Meta milestones, cash conversion, export-control resilience, and Embedded recovery. Robot endpoint content is too small in current disclosure to rank above the direct board-content names.

AVGO Broadcom Inc.

Adjacent AI custom-silicon and networking monitor with large AI revenue, but weak direct humanoid endpoint attribution.

Market cap$1.71T
Latest qtr revenue$22.187B
Weekly setupBase forming

Role in stack

AVGO supplies custom silicon, networking, accelerators, switching, optical and connectivity infrastructure, plus VMware software. The robot link is mostly offboard AI infrastructure and network read-through.

Revenue mix

Q2 FY2026 revenue was $22.187B. Semiconductor Solutions was $15.009B and AI semiconductor revenue was $10.8B, but the lane ties most of that to data-center custom silicon and networking.

Proof burden

AI revenue, RPO, custom silicon customer concentration, VMware margin, and FCF quality are central. The node keeps AVGO for continuity, while robot endpoint exposure remains indirect.

Latest-quarter revenue fields use the freshest local knowledge-lane evidence available during the 2026-07-05 build. Next earnings dates were not confirmed, so the metric strip uses weekly setup instead of estimated event dates.

What Confirms Or Weakens

Area What confirms What weakens or invalidates Watch next
01Robot endpoint platformNamed sockets
What confirms

Named robot OEM wins, Jetson/Thor, Qualcomm, AMBA, MCU, analog, power, or connectivity deployments, and production qualification evidence appear in filings, releases, or call materials.

What weakens or invalidates

Robotics stays in demos, pilots, developer kits, or mixed segment language while production orders and content per robot remain undisclosed.

Watch next
  • OEM wins
  • Qualification
  • Production sockets
  • Content per robot
02Board contentControl and signal chain
What confirms

MCUs, analog, signal chain, PMICs, motor drivers, timing, Ethernet, wireless, safety controllers, and sensor interfaces are named in robot or control-board use cases.

What weakens or invalidates

Exposure remains generic industrial, auto, handset, or data-center language without robot socket evidence or segment revenue that maps to endpoint electronics.

Watch next
  • TXN
  • NXPI
  • QCOM
  • ADI
  • MCHP
03Power and thermalBattery and heat
What confirms

Higher endpoint compute content pulls through PMICs, power modules, discretes, battery management, motor-drive content, and measured power-density improvement.

What weakens or invalidates

Heat, battery life, cost, or efficiency limits force lower endpoint compute content, cloud dependence, or simpler control boards.

Watch next
  • MPWR
  • ON
  • TXN
  • ADI
  • BOM cost
04Revenue conversionDesign win to FCF
What confirms

Design wins convert into qualification, production sockets, content per robot, latest-quarter revenue, segment margin, inventory normalization, and free cash flow.

What weakens or invalidates

Segment revenue remains dominated by data center, handset, auto, analog recovery, or broad industrial restocking while inventory, margin, or FCF deteriorates.

Watch next
  • Revenue
  • Margins
  • Inventory
  • FCF
05Connectivity and loopsLatency requirement
What confirms

Robot architectures require local inference, deterministic communication, industrial Ethernet, wireless, safety isolation, and 5-10 ms control-loop performance.

What weakens or invalidates

Cloud tolerance, simpler autonomy, lower sensor count, or centralized controllers reduce endpoint silicon content and weaken the per-robot semiconductor path.

Watch next
  • Latency
  • Ethernet
  • Wireless
  • Safety controllers
06Specialist edge AIAMBA test
What confirms

AMBA robotics, edge infrastructure, automotive, or vision SoC programs turn into revenue growth, inventory conversion, high gross margin, lower losses, and clearer LTA purchase evidence.

What weakens or invalidates

Customer concentration, inventory, advanced-node costs, SBC, or weak scale outweigh design-win claims, and LTAs remain potential revenue rather than firm commitments.

Watch next
  • Q2 FY2027
  • Inventory days
  • CV ramps
  • LTA terms
07Policy and supplyCross-border risk
What confirms

Export controls, tariffs, foundry access, memory/input cost, and customer-country exposure remain manageable while production ramps and gross margin hold.

What weakens or invalidates

Export controls, ICTS rules, tariffs, foundry shortages, memory scarcity, Taiwan or China exposure, or customer concentration disrupt supply, margin, or customer access.

Watch next
  • Export controls
  • Tariffs
  • Foundry
  • Memory
08Stale conditionRefresh trigger
What confirms

Knowledge lanes remain current, local daily_ohlc rows stay current, and the report API serves weekly charts for all ranked tickers without embedding chart payloads.

What weakens or invalidates

A new market session, earnings release, 10-Q, guidance revision, chart failure, robot design-win disclosure, or source-lane update arrives before this node is refreshed.

Watch next
  • Daily OHLC
  • API charts
  • New filings
  • Source lanes

Source Trail

Routes used for this node
Route Use it for Primary links
Paper priorsRobot compute and connectivity cost pools, edge AI, local inference, 5-10 ms control loops, heat, battery draw, regulation, and semiconductor partnership constraints.Citrini humanoid primer; Roland Berger 2026 extraction; Roland Berger PDF
Knowledge posteriorCompany setup, revenue mix, segment evidence, guidance, inventory, margin, FCF, valuation proof burden, and sector transmission paths.NVDA; TXN; NXPI; QCOM; MPWR; ADI; ON; MCHP; STM; AMBA; AMD; AVGO
Sector and conceptsAI infrastructure capex, analog and embedded recovery, memory/input-cost pressure, export controls, policy risk, valuation dispersion, and physical AI routing.Information Technology; AI capex cycle; Memory and components; Physical AI boundary; Trade and sourcing
Monitor listImportant suppliers or read-throughs that were rejected as core because local chart coverage, direct endpoint attribution, or source support was weaker for this node.Infineon; Renesas; VICR; POWI; ALGM; SITM; CEVA; ARM; MRVL; MU; AMKR; SLAB; LSCC.
Discovery and DuckDB methodsRead-only local coverage, computed market caps, weekly setup labels, and selected-security chart provenance. Ranked tickers had daily_ohlc rows through 2026-07-02.node metadata; discovery status NVDA --json; discovery lineage-status --ticker NVDA --dataset daily_ohlc --limit 5 --json; read-only DuckDB on daily_ohlc, instruments, and ingestion_runs.
API chart routeRight-rail chart rendering. Each rank button and metrics strip calls the selected-security panel through data-chart-ticker. Weekly charts are expected from daily_ohlc, grouped by calendar week./api/securities/TICKER/chart?frequency=weekly&window=3y&as_of=latest; three-year visible horizon; no embedded OHLC chart payloads; all ranked tickers were API-checked on 2026-07-05.
Ranking processThe final order was selected by ten xhigh primary subagents and one xhigh merge agent. The merge kept NVDA first, promoted direct endpoint silicon, added AMBA, and demoted AMD and AVGO to adjacent monitors.Final order: NVDA, TXN, NXPI, QCOM, MPWR, ADI, ON, MCHP, STM, AMBA, AMD, AVGO.
Data limitationsDiscovery supports price, volume, chart routing, and computed market-cap context. It does not prove humanoid revenue exposure. Papers are priors; filed lanes are the posterior. Direct humanoid revenue is not assigned without source support.Refresh if a new market session, filing, earnings release, guidance update, robot design win, or local API/chart behavior changes.

Categories