Radar Sensor Semiconductors

Radar detects objects by transmitting radio waves and measuring the reflected return signal — extracting range from round-trip time, velocity from Doppler frequency shift, and angle from antenna array geometry. Unlike cameras and LiDAR, radar operates through fog, rain, snow, and darkness with no performance degradation, making it the all-weather redundancy layer in every automotive ADAS and AV perception stack. Modern automotive radar runs at 77–81GHz (millimeter wave) using SiGe BiCMOS monolithic radar SoCs that integrate transmitter, receiver, ADC, and digital signal processor on a single die — a level of integration that has compressed what was once a rack of discrete RF components into a module smaller than a deck of cards.

The radar semiconductor market spans three distinct supply chain populations: automotive radar (SiGe BiCMOS, AEC-Q100 qualified, NXP/TI/Infineon dominant), imaging radar (high-channel-count MIMO ICs and antenna-in-package substrates, emerging for L3+ AV), and defense phased-array radar (GaN-on-SiC transmit/receive modules, entirely separate vendor ecosystem). Industrial and drone radar draws from the automotive IC ecosystem. These populations share the 77GHz frequency band but diverge sharply in semiconductor process, power level, qualification requirement, and supply chain.

Automotive Radar ICs — Products & Process

Deployment & Supply Chain Risk

Imaging Radar — The 4D Resolution Architecture

Standard automotive radar produces a sparse point cloud — enough spatial information to detect a vehicle 200m ahead and measure its closing velocity, but not enough angular resolution to distinguish a pedestrian from a fence post at the same range. Imaging radar (4D radar) addresses this by deploying large MIMO (Multiple Input Multiple Output) antenna arrays that synthesize a much larger virtual aperture than the physical antenna footprint. A 3TX/4RX standard radar produces 12 virtual antenna elements. Arbe's Phoenix chipset deploys 48TX/48RX, producing 2,304 virtual elements — a 192× increase in virtual aperture that translates directly into angular resolution.

The semiconductor requirement for imaging radar is substantially higher than standard radar: more TX/RX channels demand more RF front-end circuits, higher ADC sampling rates for the wider instantaneous bandwidth, and significantly more digital signal processing to handle the increased data rate from 2,304 virtual channels simultaneously. The imaging radar IC is therefore a more complex and power-hungry device than a standard radar transceiver. The antenna-in-package (AiP) substrate — integrating the 77GHz antenna array directly into the chip package rather than etching it on PCB — is required because at 77–81GHz, the antenna geometry tolerance (±tens of micrometers) exceeds what standard automotive PCB manufacturing can achieve at the required production volume and consistency.

Imaging radar is positioned as a potential partial substitute for LiDAR in AV perception stacks — not because it achieves LiDAR spatial resolution (it does not), but because it provides all-weather 3D spatial information at radar cost and reliability without LiDAR's InGaAs APD supply chain constraints. Continental, Bosch, and Aptiv are all developing imaging radar modules for volume automotive production. The semiconductor IC underlying these modules comes from NXP (cascaded TEF82xx for more channels), TI (AWR2944), and dedicated imaging radar chip companies (Arbe, Vayyar).

Defense Phased-Array Radar — Separate Supply Chain

Defense and aerospace radar operates at power levels, frequency ranges, and performance specifications that are fundamentally incompatible with automotive SiGe BiCMOS radar ICs. An AESA (Active Electronically Scanned Array) radar on a fighter aircraft or naval vessel uses GaN-on-SiC transmit/receive (T/R) modules — one per antenna element — operating at kilowatts of peak power, controlled by custom beamformer ICs, and fed by high-speed ADCs with 12–16-bit resolution. The semiconductor supply chain is entirely distinct from automotive radar.

SiGe BiCMOS — The Automotive Radar Process Constraint

Every NXP and Infineon automotive radar IC is fabricated on SiGe BiCMOS — a process that integrates silicon-germanium heterojunction bipolar transistors (HBT) with standard CMOS on the same die. SiGe HBTs achieve the transition frequency (fT) required for 77GHz operation — a transistor speed that standard CMOS cannot match at the noise floor required for long-range radar. The two production-scale SiGe BiCMOS foundries for automotive radar are GlobalFoundries (BiCMOS9, 0.13µm, Malta NY) and Tower Semiconductor (0.18µm SiGe, Israel and US fabs). These are the same two foundries supplying SiGe for RF transceivers and analog mixed-signal — a shared constraint pool.

Tower Semiconductor's situation — independent following Intel's failed acquisition, blocked by Chinese regulatory approval in 2023 — adds strategic uncertainty to a foundry that supplies automotive radar ICs for most Western ADAS platforms. Tower's long-term capital investment commitment under independent ownership is an open question for ADAS radar supply chain planners with 10-year vehicle platform design horizons. GlobalFoundries Fab 9 in Malta NY provides geographic diversification from Tower's Israel-primary operations but has its own capacity constraints shared across BiCMOS customers.

Supply Chain Bottlenecks

Related Coverage

Sensor Semiconductors Overview | Perception Sensors Supply Chain | LiDAR Sensors | Automotive & Robot Image Sensors | RF & Networking (SiGe BiCMOS, GaN RF) | SiC & GaN Power Modules | Semiconductor Bottleneck Atlas

Cross-Network — ElectronsX Demand Side

Every ADAS-equipped vehicle ships with a minimum of four radar units; L2+ platforms deploy six or more. The EU NCAP 2024 mandatory radar requirements create a volume step function that is already in the AEC-Q100 qualification pipeline. Imaging radar is the emerging perception technology for AV platforms that cannot depend on LiDAR at scale — its semiconductor content per vehicle is significantly higher than standard radar. Smart infrastructure traffic monitoring and smart intersection radar are growing industrial radar demand vectors tied to EX smart infrastructure content.

EX: ADAS/AV Compute Architecture | EX: EV Semiconductor Dependencies | EX: Humanoid Robots