Chip Types — Semiconductor Device Overview

Chip Types is the output layer of the semiconductor supply chain — the devices that emerge from the fabrication and assembly process and are deployed across every sector of the AI-industrial economy. Semiconductors span multiple device categories, each optimized for specific functions: computing and data processing, energy conversion and control, physical sensing and measurement, and wireless communication. Logic and memory chips dominate global revenue. AI accelerators, power semiconductors (SiC, GaN), and sensor ICs are driving the fastest growth through 2030.

Understanding chip types from a supply chain perspective means understanding not just what each device does, but what process node it requires, who manufactures it, what upstream inputs it depends on, and what cross-sector demand signals are pulling against its supply. A GPU and a gate driver IC are both chips — but they occupy completely different supply chains, face different concentration risks, and are constrained by different bottlenecks. SX organizes chip types into three clusters that reflect genuine supply chain groupings rather than arbitrary product marketing categories.

See the ranked view of which chip types face the most severe supply chain constraints: Semiconductor Bottleneck Atlas

Three Supply Chain Clusters

The three-cluster architecture reflects how chip types actually share supply chain infrastructure — which device families compete for the same foundry capacity, the same substrate supply, and the same qualification pipelines. Devices within a cluster share upstream dependencies. Devices across clusters generally do not.

Cluster	Device families	Process node range	Primary supply chain character	Dominant bottleneck
Compute & Memory	CPUs, GPUs, AI accelerators, edge inference SoCs, ASICs, FPGAs, MCU/MPUs, security silicon, memory and storage, HBM, neuromorphic, quantum compute	Leading-edge (N2-N5) for AI and high-performance logic; mature node (28nm-180nm) for MCU/MPU and embedded control	TSMC foundry concentration for leading-edge; CoWoS and HBM advanced packaging as separate supply constraints from wafer starts; EDA and PDK lock-in for fabless designers; AEC-Q100 qualification paradox for automotive MCU	TSMC N3/N5 wafer allocation; CoWoS packaging capacity; HBM supply (SK Hynix dominant); AEC-Q100 lock-in for automotive MCU
Power & Analog	Power semiconductors (SiC MOSFETs, GaN HEMTs, IGBTs, Si MOSFETs), SiC and GaN power modules, analog ICs, mixed-signal devices, optoelectronics, solar PV cells	Specialty (SiC PVT, GaN MOCVD epi); mature node (90nm-180nm for analog); 200mm fab dominant for automotive-grade analog and power	SiC substrate physics-limited throughput; nine-market SiC demand convergence against one wafer funnel; TI-ADI precision analog duopoly; 200mm fab capacity ceiling; Wolfspeed Chapter 11 as Western SiC chokepoint	SiC substrate boule growth rate; Wolfspeed restructuring; 200mm fab capacity; AEC-Q101 qualification lock-in
Sensing & Connectivity	Image sensors (CMOS BSI), LiDAR (InGaAs APD, VCSEL), radar (77GHz SiGe BiCMOS), IR/thermal, IoT/IIoT sensors, quantum sensors, sensor fusion, RF and networking, electromechanical sensors (position, current, temperature, force-torque, IMU)	Five distinct technologies — advanced CMOS (image sensor), 130-250nm SiGe BiCMOS (radar), III-V compound (LiDAR), MEMS specialty (inertial/position), mature analog CMOS (electromechanical measurement)	Sony CIS dominance (~50-55% automotive market); InGaAs APD compound semiconductor scarcity; SiGe BiCMOS radar oligopoly; proprioceptive sensor layer dominates unit count 10:1 to 100:1 over perception sensors at vehicle and robot scale; force-torque supply void at humanoid volume	Sony Japan concentration; InGaAs APD scarcity for LiDAR scale-up; GMSL SerDes lock-in; force-torque supply void at humanoid production volume

Segment Mapping — Device Categories, Functions, and Key Players

Category	Cluster	Primary function	Representative companies	Key applications
Logic & Compute	Compute & Memory	Instruction execution, system control, processor arithmetic	Intel, AMD, ARM (IP), RISC-V vendors, Renesas, NXP, Infineon (MCU)	PCs, servers, embedded systems, automotive control, industrial automation
Memory & Storage	Compute & Memory	Short- and long-term data retention — volatile (DRAM, SRAM) and non-volatile (NAND, NOR, emerging NVM)	Samsung, SK Hynix, Micron, Kioxia, Western Digital	Smartphones, data centers, AI training clusters, IoT, automotive storage
AI Accelerators & GPUs	Compute & Memory	Parallel compute for machine learning training and inference; high-throughput matrix operations	NVIDIA, AMD, Google (TPU), Amazon (Trainium), Microsoft (Maia), Meta (MTIA), Tesla (FSD/AI5/AI6)	AI training datacenters, inference clusters, AV compute, humanoid robot central inference
SoCs	Compute & Memory	Integrated CPU, GPU/NPU, memory controller, safety subsystem, and I/O on a single die	Apple (A/M-series), Qualcomm, MediaTek, Samsung Exynos, NVIDIA DRIVE, Mobileye EyeQ	Smartphones, automotive ADAS, robotics, industrial compute
Power & Compound Semiconductors	Power & Analog	High-voltage switching and power conversion; SiC and GaN enable higher efficiency than silicon at elevated voltages and frequencies	Infineon, Wolfspeed, STMicro, Onsemi, Rohm, Bosch, EPC, Navitas, Infineon/GaN Systems	EV traction inverters, BESS power conversion, EVSE DCFC, solar inverters, industrial VFDs, robot joint drives
Analog	Power & Analog	Signal amplification, conditioning, and voltage regulation for continuous signals; battery cell monitoring, current sensing, temperature measurement	Texas Instruments, Analog Devices, STMicro, Microchip, Renesas	Power management, battery monitoring, motor control, industrial sensing
Mixed-Signal	Power & Analog	Bridge analog inputs with digital processing — ADC/DAC conversion, RF front-ends, automotive safety ICs	NXP, Renesas, ADI, TI, Infineon	Automotive BMS and ADAS, telecom, IoT, industrial control
RF & Connectivity	Sensing & Connectivity	Wireless signal generation, amplification, and reception across 5G, WiFi, Bluetooth, GNSS, and satellite bands	Qualcomm, Broadcom, Qorvo, Skyworks, MediaTek, NXP	5G handsets and infrastructure, WiFi, IoT, automotive telematics, satellite connectivity
Sensors	Sensing & Connectivity	Convert physical phenomena (light, pressure, acceleration, magnetic field, temperature, force) to electrical signals for perception and control	Sony (CIS), Bosch (MEMS), ADI (IMU), ams-OSRAM (magnetic encoder), NXP (radar), TI (current/temp), Lumentum (VCSEL/LiDAR)	Automotive cameras and radar, humanoid robot joints and manipulation, BMS cell monitoring, ADAS, industrial automation
Optoelectronics	Power & Analog / Sensing & Connectivity	Light emission (LEDs, laser diodes, VCSELs) and light detection (photodiodes, APDs, image sensors); energy generation (PV cells)	Nichia, ams-OSRAM, Lumentum, Coherent, First Solar, LONGi	Displays, lighting, LiDAR emitters and detectors, optical communications, solar PV

Market Trajectory by Device Category (2024-2030)

Market size is not the same as supply chain risk. The highest-revenue categories are not necessarily the most constrained. The fastest-growing categories — AI accelerators, power semiconductors — are often where supply chain stress is most acute because demand growth is outpacing the 3-5 year lead time required to expand fab capacity and qualify new devices.

Category	Approx. 2024 global share	CAGR 2024-2030	Primary growth driver	Supply chain stress
Memory & Storage	~28%	6-7%	AI training cluster HBM demand; datacenter DRAM and NAND expansion	High for HBM (SK Hynix concentration); moderate for DRAM/NAND
Logic & Compute	~25%	5-6%	Server CPU; automotive MCU growth; RISC-V adoption in embedded and IoT	Very High for automotive MCU ($2 chip paradox, 200mm); moderate for server CPU
AI Accelerators & GPUs	~10%	>20%	AI training cluster buildout; AV compute; robot inference demand post-2026	Critical — stacked bottleneck (TSMC N3/N5 + CoWoS + HBM)
SoCs	~12%	8-9%	Automotive ADAS and SDV; smartphone volume; industrial IoT edge compute	High for automotive inference SoC (NVIDIA concentration + TSMC + AEC-Q100)
Power & Compound Semiconductors	~7%	12-14%	EV SiC adoption; BESS power conversion; solar inverters; VFDs; robot GaN drives	Critical — SiC physics-limited throughput; Wolfspeed restructuring; nine simultaneous demand markets
RF & Connectivity	~6%	7-8%	5G rollout; IoT device proliferation; satellite connectivity	Medium — SiGe BiCMOS oligopoly for radar; generally more distributed than other categories
Sensors	~5%	9-10%	Automotive camera and radar density; humanoid robot proprioceptive sensor demand; BESS monitoring	High — Sony CIS, InGaAs APD scarcity, force-torque supply void at humanoid scale
Analog	~4%	5-6%	Power management across every electrified system; battery cell monitoring; industrial automation sensing	High — TI-ADI duopoly; 200mm fab ceiling; AEC-Q100 lock-in mirrors $2 chip paradox
Mixed-Signal	~2%	6-7%	Automotive BMS and safety ICs; IoT ADC/DAC; telecom infrastructure	High for automotive-grade; moderate for industrial and telecom
Optoelectronics	~1-2%	4-5%	LiDAR VCSEL and APD demand with AV/robot scale-up; LED stable; solar PV tied to renewable buildout	Medium-High for LiDAR emitters (VCSEL) and detectors (InGaAs APD); LED moderate

Compute & Memory — Section Guide

AI Inference & Edge Compute SoCs — Supply Chain (Tier 1 interface page)
TSMC foundry concentration. Stacked bottleneck (TSMC N3/N5 + CoWoS + HBM + AEC-Q100). NVIDIA ~80% AV program concentration as structural supply chain risk. Custom ASIC programs (Tesla FSD, Waymo Albatross). Robot inference as the second AI SoC demand wave.

Mature Node MCUs — The $2 Chip Paradox (Editorial deep-dive)
Why a $2 microcontroller stops a $55,000 vehicle. ISO 26262 qualification lock-in. 200mm fab economics. Six-supplier landscape (Infineon AURIX, Renesas RH850, NXP S32K, TI TMS570, ADI, STMicro). China mature-node capacity as the most asymmetric geopolitical lever in the semiconductor trade conflict.

Power & Analog — Section Guide

SiC & GaN Power Modules — Supply Chain (Tier 1 interface page)
Crystal growth physics as the primary SiC capacity ceiling. Nine-market demand convergence against one wafer funnel. Wolfspeed Chapter 11 and Western OEM program risk. 150mm to 200mm transition as the volume multiplier. Full supplier landscape. GaN robot joint drive as the new demand signal post-2026.

Sensing & Connectivity — Section Guide

Sensor Semiconductors Overview (cluster hub)
Two-population framework — perception sensors vs. electromechanical sensors. Per-platform unit count comparison (10:1 to 100:1 electromechanical-to-perception). Supply chain risk heat map across eight sensor categories.

Perception & Environment Sensors — Supply Chain
Sony CMOS structural dominance. InGaAs APD LiDAR scarcity. VCSEL supply. 77GHz SiGe BiCMOS radar oligopoly. GMSL and FPD-Link concentration.

Electromechanical & Control Sensors — Supply Chain
Battery cell monitor IC duopoly. Isolated current sense amplifiers. Motor position encoder ICs. IMU supply. Force-torque supply gap. TI-ADI duopoly.