5G/6G & Wireless
Semiconductors

The 5G and wireless sector occupies a structurally distinct position in the semiconductor supply chain map. Unlike compute-dominant sectors where leading-edge CMOS foundries (TSMC N3/N5) define the supply chain, wireless infrastructure and RF front-end semiconductors are heavily dependent on compound semiconductor processes - gallium arsenide (GaAs), gallium nitride (GaN), silicon germanium (SiGe) BiCMOS, and indium phosphide (InP) - manufactured at specialty foundries with supply chains that operate largely independently of the mainstream silicon logic supply chain. This compound semiconductor dependency is the defining supply chain characteristic of the wireless sector and the source of its most durable bottlenecks.

The sector splits into two functionally distinct sub-markets with different semiconductor profiles. The infrastructure sub-market covers 5G base stations (gNodeB), remote radio units (RRUs), massive MIMO antenna arrays, and the baseband processing units that drive them - dominated by Ericsson, Nokia, and Huawei on the system side, with semiconductors from Qualcomm, Marvell, NXP, and TI on the ASIC and DSP side. The device sub-market covers the RF front-end modules (RFFE) inside smartphones, fixed wireless access (FWA) routers, private 5G enterprise systems, and satellite connectivity terminals - dominated by Qualcomm on baseband SoCs and Qorvo, Skyworks, and Broadcom on RF front-end modules. A third and growing sub-market is private 5G for industrial, logistics, and robotics applications - where 5G connectivity becomes the communication layer for autonomous systems and robot fleet coordination.

Related Coverage: RF & Networking | Compound Wafers | Critical Elements | Robotics & IoT | Mobile & Consumer | Bottleneck Atlas

Semiconductor Device Map — 5G/6G & Wireless

The Compound Semiconductor Dependency

The wireless sector's most fundamental supply chain characteristic is its dependency on compound semiconductors - GaAs, GaN, SiGe BiCMOS, and InP - that cannot be manufactured at standard silicon CMOS foundries. This is not a temporary limitation that will be resolved by process scaling; it is a physics constraint. Gallium arsenide provides higher electron mobility than silicon, enabling power amplifiers that operate efficiently at the microwave frequencies used in cellular RF front-ends. Gallium nitride on silicon carbide provides higher breakdown voltage and power density than silicon, enabling base station power amplifiers that deliver 10-100W of RF output. Silicon germanium BiCMOS enables low-noise amplifiers and transceiver circuits that operate at millimeter-wave frequencies (24-100GHz) with noise figures that silicon CMOS cannot match.

The compound semiconductor supply chain is structurally narrower than the silicon supply chain. WIN Semiconductors in Taiwan is the world's largest GaAs foundry and manufactures the GaAs pHEMT and HBT wafers used in the majority of handset RFFE power amplifiers globally. Wolfspeed in Durham, North Carolina grows the SiC substrates and operates GaN-on-SiC device fabs that supply a significant share of Western 5G base station PA supply. GlobalFoundries Fab 9 in Malta, New York is one of a small number of fabs capable of SiGe BiCMOS production at the process generations required for mmWave circuits. Each of these represents a concentration point with no near-term silicon foundry equivalent.

The gallium supply dimension compounds this further. Gallium - the base material for GaAs, GaN, and GaAs substrates - is primarily produced as a byproduct of aluminum smelting from bauxite, with China accounting for approximately 80% of global refined gallium production. China's 2023 export controls on gallium and germanium exports, while not yet causing acute supply disruption to the Western wireless semiconductor supply chain, represent a potential leverage point in the US-China technology competition that has no equivalent in silicon supply chains. Western gallium supply independence - from primary smelting through GaAs and GaN wafer production - is a multi-decade investment requirement, not a near-term policy fix.

Huawei Kirin — The Export Control Case Study

Huawei's Kirin SoC program is the most instructive case study in semiconductor sector geopolitics on the SX site. Huawei's HiSilicon division designed leading-edge Kirin SoCs - including the Kirin 990 5G (TSMC N7+, 2019) and Kirin 9000 (TSMC N5, 2020) - that were competitive with Qualcomm Snapdragon on both application SoC performance and 5G modem integration. In May 2020, BIS (Bureau of Industry and Security) imposed export controls requiring TSMC to obtain a license before manufacturing chips for Huawei, effectively ending TSMC's ability to manufacture Kirin SoCs. Huawei exhausted its pre-ban chip inventory by late 2021 and was forced to use older Kirin generations and MediaTek SoCs for its handset business through 2022-2023.

The Kirin 9010 (also labeled 9000S in some markets), revealed in the Mate 60 Pro in August 2023, demonstrated that SMIC had achieved sufficient yield at its N+1 process node (DUV multi-patterning, approximately 7nm equivalent density) to manufacture a functional 5G-capable SoC for Huawei without TSMC. The chip was manufactured without EUV lithography, relying instead on multiple DUV exposures to achieve sub-10nm feature density. The performance gap to Qualcomm's Snapdragon 8 Gen 2 (TSMC N4) is real - approximately one to two generations in compute density and power efficiency - but the Kirin 9010 demonstrated that China's domestic semiconductor industry could produce a commercially viable 5G handset SoC under export control conditions. This is the most significant single data point in the ongoing assessment of China's semiconductor self-sufficiency trajectory.

The supply chain implication extends beyond handsets. Huawei's Kirin program at SMIC validates the viability of a China-domestic 5G silicon supply chain for smartphones, which reinforces Beijing's confidence that domestic semiconductor development can eventually replace Western-foundry-dependent supply chains across multiple device categories. It also demonstrates that export controls, while effective at slowing capability development, cannot prevent it entirely when the target nation has sufficient engineering talent, state investment, and domestic market scale to absorb the performance penalty during the catch-up period.

5G Infrastructure — The Three-Vendor Market

Global 5G radio access network (RAN) infrastructure is effectively a three-vendor market: Ericsson, Nokia, and Huawei. Samsung is a fourth player with meaningful share in South Korea, the US (as a Huawei alternative post-ban), and Japan. ZTE holds a position in China and some emerging markets. The semiconductor supply chains for each vendor are structurally distinct - Ericsson and Nokia source components from Western semiconductor suppliers and design proprietary ASICs using Western EDA and foundry supply chains; Huawei designs and sources from Chinese domestic suppliers where possible and is actively reducing dependency on Western semiconductor components in its RAN systems.

The US and allied-nation restrictions on Huawei equipment in 5G infrastructure (US Clean Network initiative, UK ban effective 2027, European operator decisions) have effectively split the global 5G infrastructure market into two parallel supply chains. In Western and allied markets, Ericsson and Nokia compete on the same base stations with similar semiconductor architectures. In China, Huawei and ZTE dominate with domestically designed semiconductor content. This bifurcation is structurally parallel to the AI compute split but at the infrastructure rather than chip level - and it creates two independently evolving RAN technology roadmaps that will increasingly diverge in both capability and supply chain composition through 2030.

Private 5G — The Industrial and Robotics Convergence

Private 5G networks - standalone 5G deployments within factory floors, warehouse logistics operations, port facilities, and industrial campuses - represent the fastest-growing new demand vector for 5G semiconductor supply. The driver is the requirement for deterministic low-latency wireless connectivity that WiFi cannot reliably provide at industrial scale. A warehouse with 500 AMRs operating simultaneously needs guaranteed sub-10ms latency and reliability levels (99.9999% or higher) that WiFi 7, despite its performance improvements, cannot consistently achieve in dense multi-path RF environments with moving metal obstacles.

The semiconductor stack for private 5G differs from macro cellular in several ways. Small cell RAN hardware (using Qualcomm FSM RAN SoC or Marvell OCTEON Fusion) replaces macro base station infrastructure. CBRS (Citizens Broadband Radio Service) spectrum in the US and similar shared spectrum bands in Europe and Japan enable private network deployment without licensed spectrum acquisition. The connected endpoints - AMRs, AGVs, humanoid robots, industrial sensors, and wearable worker terminals - each carry a 5G modem (typically Qualcomm Snapdragon X series or MediaTek T750 for industrial devices) plus a GaAs or GaN front-end module.

The robotics intersection is particularly significant from a supply chain perspective. A humanoid robot operating in a factory environment connected via private 5G carries both the robotics semiconductor stack (GaN joint drives, position encoders, inference SoC) and the 5G connectivity semiconductor stack (5G modem, GaAs PA, BAW filters) simultaneously. The combined semiconductor content of a 5G-connected humanoid robot is substantially higher than either a standalone humanoid or a standalone 5G device - and neither supply chain was sized for this combined demand profile.

6G — The 2030 Supply Chain Horizon

6G standardization (3GPP Release 20 and beyond) is targeting commercial deployment in the 2030-2033 timeframe. The semiconductor supply chain implications are already visible at the research level. 6G is expected to operate in sub-THz frequency bands (100GHz-300GHz) for the highest-throughput applications, which requires semiconductor processes - InP HEMT, GaN, SiGe BiCMOS at sub-130nm nodes - that are currently in research-grade production. The move to sub-THz frequencies demands transistors with transition frequencies (ft) above 300GHz, achievable today only in InP HEMT and advanced SiGe BiCMOS, not in standard silicon CMOS.

Key 6G semiconductor research programs are underway at DARPA (6G program), the European Commission (Hexa-X initiative), Japan's NICT (Beyond 5G), South Korea (Samsung and LG Electronics 6G R&D), and China (IMT-2030 promotion group). Each program is developing compound semiconductor device technology for sub-THz operation in parallel with new air interface specifications. The supply chain for 6G sub-THz silicon will require significant expansion of InP and advanced SiGe BiCMOS foundry capacity starting in the 2026-2028 window to support prototype system development ahead of 2030 commercial deployment. This represents a new and distinct compound semiconductor demand wave layered on top of 5G infrastructure deployment that is still ongoing in emerging markets.

Supply Chain Bottlenecks and Risk Factors (2026-2030)

Key 5G/6G & Wireless Semiconductor Suppliers

Cross-Sector Convergence

The 5G/6G wireless sector intersects three significant cross-sector supply chain dynamics. First, the gallium supply convergence: gallium is the common upstream material for GaAs (handset RFFE PAs), GaN (base station PAs and EV OBC and robot joint drives), and InP (6G sub-THz devices and datacenter coherent optical transceivers). China's dominance of refined gallium production creates a single geopolitical leverage point that cuts across the wireless, automotive, robotics, and datacenter sectors simultaneously. A tightening or weaponization of Chinese gallium export controls would affect all compound semiconductor supply chains concurrently - not sequentially - with no rapid Western supply response possible.

Second, the private 5G and robotics convergence: as humanoid robots and AMRs deploy in factory and logistics environments connected via private 5G networks, the semiconductor demand profile of each robot unit includes both the robotics-specific analog and power stack and the 5G connectivity stack. Private 5G network infrastructure serving robot fleets requires small cell RAN hardware (open RAN baseband ASICs, GaN small cell PAs) scaled to facility size. The combined demand from robot fleet 5G connectivity is a new demand vector for private 5G infrastructure semiconductors that is not reflected in standard cellular infrastructure market forecasts.

Third, the satellite and automotive V2X convergence: 5G NTN (non-terrestrial network) integration in automotive creates a supply chain intersection between the wireless sector's satellite modem supply chain and the automotive sector's V2X connectivity supply chain. 3GPP Release 17 NTN supports direct-to-device satellite connectivity using standard 5G NR air interface, which means automotive telematics SoCs with 5G NTN support can access satellite connectivity via the same modem used for terrestrial 5G. Qualcomm's Snapdragon Auto 5G Modem-RF Gen 2 with NTN support and automotive-grade qualification is the primary device enabling this convergence - and it must satisfy AEC-Q100 automotive qualification requirements on top of the standard 5G modem qualification regime.

Related Coverage: RF & Networking | Compound Wafers | Critical Elements | Bottleneck Atlas | Robotics & IoT | Automotive & Mobility | Datacenter / HPC | Mobile & Consumer | U.S. Reshoring

Cross-Network: ElectronsX Demand Side

5G wireless infrastructure demand intersects EX coverage at the autonomous vehicle V2X connectivity layer and the robot fleet wireless communication infrastructure layer.

EX: AV Platforms Directory | EX: Humanoid Robots | EX: Supply Chain Convergence Map

Key Questions — 5G/6G & Wireless Semiconductors

Why can't BAW filters be manufactured at TSMC or Samsung Foundry? Bulk acoustic wave filters rely on piezoelectric thin films - typically aluminum nitride (AlN) or scandium-doped aluminum nitride - deposited on silicon and patterned to resonate at specific RF frequencies. The resonant frequency is set by the film thickness to atomic-level precision. This is a fundamentally different manufacturing process from CMOS transistor fabrication - it requires different deposition equipment, different materials, and different process control than anything in the standard silicon foundry toolkit. Qorvo, Broadcom (Avago heritage), TDK, and Murata each developed their BAW processes over decades as proprietary IP. There is no BAW process design kit available at TSMC, Samsung, or GlobalFoundries because the process has never been standardized or licensed to those foundries. Adding BAW capability to a standard CMOS foundry would require years of process development investment with uncertain yield outcomes - which is why new entrants have not emerged despite the obvious supply concentration risk.

What is the significance of the Huawei Kirin 9010 at SMIC? It demonstrated that China's domestic semiconductor industry can produce a functional 5G-capable SoC at approximately 7nm equivalent density using DUV multi-patterning without EUV - approximately two process generations behind TSMC's leading edge at time of production. The performance gap is real and the supply chain limitations (SMIC cannot access ASML EUV scanners) are durable through at least 2030 given current export control trajectories. But the Kirin 9010 showed that the performance gap does not prevent commercial viability for the domestic Chinese market, where consumers and enterprises have no access to NVIDIA or TSMC-dependent alternatives regardless of preference. This matters for supply chain analysis because it validates Beijing's confidence in the domestic semiconductor development path and supports continued state investment in SMIC capacity expansion and domestic EDA tool development.

What is open RAN and why does it matter for supply chains? Open RAN (O-RAN) is an architecture that disaggregates the radio access network into standardized, interoperable components - radio units (RU), distributed units (DU), and centralized units (CU) - that can be sourced from different vendors rather than from a single proprietary system supplier like Ericsson or Nokia. The supply chain significance is twofold. From a geopolitical standpoint, open RAN is the primary policy mechanism for enabling Huawei alternatives without being locked into a two-vendor (Ericsson/Nokia) market - any vendor whose hardware meets the O-RAN Alliance specifications can participate. From a semiconductor standpoint, open RAN creates a new commercial market for baseband ASICs (Qualcomm FSM, Marvell OCTEON Fusion) that would otherwise be captive to Ericsson and Nokia internal design programs. Open RAN adoption at scale creates a more open wireless infrastructure semiconductor supply chain, but current performance parity gaps with proprietary systems mean this transition will take until 2027-2028 to be commercially definitive.

How does 6G affect compound semiconductor supply chain requirements? 6G's sub-THz frequency bands (100-300GHz target for the highest-throughput links) require transistors with transition frequencies above 300GHz. Only InP HEMT and advanced SiGe BiCMOS currently achieve these speeds in production-grade devices. The supply chain consequence is a new wave of compound semiconductor foundry demand starting approximately 2026-2028 for prototype system development, scaling to commercial deployment demand post-2030. InP wafer production - primarily from Sumitomo Electric and AXT in the US - will need to expand. InP device foundry capacity at IQE, Qorvo, and WIN will need to expand. GlobalFoundries Fab 9 and IHP SiGe BiCMOS capacity will need to expand for the sub-100GHz portions of 6G. None of these expansions are in the current publicly disclosed capital investment plans of the relevant foundries at the scale 6G commercial deployment will require.

Related Coverage

Sectors Hub | RF & Networking | Compound Wafers | Critical Elements | Bottleneck Atlas | SiC & GaN Power Modules | Mobile & Consumer | Robotics & IoT | Automotive & Mobility | Datacenter / HPC | Space / Defense | U.S. Reshoring