Gallium nitride is now a mainstream choice for high-power 5G radio units, especially in massive MIMO systems where efficiency, power density, and thermal performance directly affect the size and operating cost of the equipment. That does not mean gallium arsenide has become obsolete. GaAs remains highly competitive in lower-power transmit chains, driver stages, small cells, microwave front ends, and receive-side functions where noise figure, linearity, integration, cost, or proven availability matter more than maximum power density.

For procurement teams, the practical question is not whether GaN is technically superior in isolation. The real question is whether the complete RF architecture benefits enough from GaN to justify a new bias network, different thermal design, higher qualification burden, limited second-source options, and potentially longer replenishment cycles. A poorly planned technology change can improve amplifier efficiency while creating a major production risk elsewhere in the bill of materials.

This guide explains how to compare GaN and GaAs at component level, where each technology fits in a 5G radio, which specifications should be normalized before quotations are compared, and how purchasing teams can reduce supply risk before a power amplifier is locked into the design.

GaN and GaAs Solve Different RF Problems

GaN and GaAs are both compound semiconductor technologies, but their electrical properties lead to different strengths. GaN devices can operate at higher electric fields and higher supply voltages, allowing substantial RF output power from a relatively small die. GaN-on-SiC devices also combine high power capability with a thermally conductive substrate, which is valuable in outdoor base stations and densely packed active antenna systems.

GaAs devices generally operate at lower voltage and lower output power, but the technology is mature, widely used, and available in many highly integrated RF products. GaAs pHEMT and HBT processes are common in power amplifiers, gain blocks, switches, low-noise amplifiers, and front-end modules. In applications where only a few watts or less are required, GaAs can offer an attractive balance of linearity, cost, package size, and design simplicity.

As a result, the decision should be made by stage and function rather than by applying one technology to the entire RF chain. A 5G radio may use a GaN final-stage PA, a GaAs or silicon pre-driver, a GaAs pHEMT or SiGe LNA, silicon switches, and dedicated power-management devices in the same front-end assembly.

Why GaN Is Expanding in 5G Base-Station Power Amplifiers

Higher power density

In a massive MIMO radio, dozens of transmit channels must fit behind or close to the antenna array. Higher power density can reduce the area occupied by each PA channel and make it easier to build compact radio units. The benefit is especially important when the equipment includes 32T32R or 64T64R architectures, because even a small saving per channel becomes significant across the complete array.

Efficiency under modulated conditions

5G signals have a high peak-to-average power ratio, so the amplifier rarely operates at saturated output power during normal service. Procurement teams should therefore avoid comparing devices only by saturated efficiency. The relevant figures are power-added efficiency, gain, adjacent-channel leakage, and error vector magnitude at the required average output power and operating back-off.

GaN devices are frequently selected for Doherty and other efficiency-enhanced PA architectures because they can provide useful gain and efficiency over a wide instantaneous bandwidth. In a large radio array, improved average efficiency reduces DC input power, heat generation, heat-sink mass, fan or cooling requirements, and long-term electricity cost.

Higher operating voltage

Broadband and high-frequency capability

GaN is well suited to sub-6 GHz infrastructure bands and can also support microwave and millimeter-wave applications. However, the word “GaN” alone does not guarantee adequate bandwidth or linearity. Buyers must verify the exact frequency range, matching condition, load-pull assumptions, modulation standard, test fixture, and thermal reference used in the manufacturer data.

Where GaAs Still Makes Commercial and Technical Sense

Lower-power small cells and indoor radios

In a low-power small cell, distributed antenna system, repeater, or indoor radio, the final output requirement may be too low to justify a discrete high-voltage GaN implementation. A compact matched GaAs PA module can reduce external component count and simplify production. Some small-cell PA families also offer pin-compatible variants across multiple bands, which can help manufacturers reuse the same PCB and shorten regional product development.

Driver and pre-driver stages

The final PA may be GaN while earlier gain stages remain GaAs, SiGe, or silicon. Pre-drivers are selected for gain, linearity, noise contribution, output level, stability, and interface compatibility. Replacing every stage with GaN can increase cost without delivering a meaningful system advantage.

Receive-side LNAs

Receive paths prioritize low noise figure, gain flatness, linearity, isolation, and protection against high input levels. GaAs pHEMT and SiGe devices are common choices for LNAs because they can achieve strong low-noise performance with mature, highly integrated processes. GaN LNAs do exist for specialized rugged or high-power environments, so it is too broad to say that GaN is never used on the receive side. For conventional commercial 5G receive chains, however, GaAs and SiGe are usually more economical and widely available.

Cost-sensitive, proven platforms

A mature GaAs design with stable field performance should not be replaced merely because GaN is newer. If the product already meets efficiency, temperature, output-power, and regulatory requirements, a technology migration may add qualification work and supply-chain exposure without increasing customer value. Procurement should ask engineering to quantify the expected system-level saving before supporting a redesign.

Do Not Compare Datasheet Headline Numbers Directly

RF amplifier specifications are highly dependent on test conditions. Two devices that appear similar in a distributor parametric table may have been characterized under completely different conditions. Before building a commercial comparison, normalize the following items:

• Frequency range: Confirm performance at the exact operating band, not only at a center-frequency demonstration point.

• Signal type: CW, pulsed, LTE, 5G NR, and WCDMA results are not directly interchangeable.

• Output definition: Separate saturated power, peak power, compressed power, and average modulated output power.

• Back-off condition: Compare PAE and linearity at the same average output and peak-to-average ratio.

• Supply voltage and quiescent current: Include the DC architecture and idle consumption.

• Linearization: Determine whether published results assume digital predistortion.

• Thermal reference: Check whether data is specified at ambient, case, flange, or channel temperature.

• Matching: Identify whether the component is internally matched, partially matched, or an unmatched transistor requiring a custom RF network.

• Package and evaluation fixture: PCB material, grounding, flange mounting, and heat spreading can materially affect measured performance.

A purchasing spreadsheet that ignores these differences can rank the wrong part as the lowest-cost option. The quoted unit price may be lower, but the device may require additional matching development, a more expensive PCB stack-up, a larger heat spreader, or a different power-supply module.

The Most Important Procurement Trap: Biasing and Sequencing

Many depletion-mode RF GaN devices require a negative gate voltage before the positive drain supply is applied. During power-down, the sequence may need to be reversed. Incorrect sequencing can create excessive drain current and damage the transistor. The precise requirements vary by device, so the manufacturer data sheet and reference design must control the implementation.

This is not simply a matter of buying a generic “GaN gate driver.” Power-switching drivers used for power-conversion GaN transistors are not automatically suitable for RF PA bias control. An RF PA may require a negative-voltage generator, drain-current sensing, active gate regulation, temperature compensation, over-current protection, power-good logic, fast enable control, and coordinated power-up and power-down sequencing.

Before approving a GaN PA for production, procurement should confirm that the BOM includes all required bias-control and protection components. It should also confirm that those supporting parts are available for the full product lifetime. A nominally second-sourced PA is not truly second-sourced if both alternatives depend on a scarce controller, custom converter, or proprietary module.

Thermal Design Is a Supply-Chain Issue

GaN can tolerate high junction temperatures, but that does not remove the need for disciplined thermal design. Reliability still depends on keeping channel temperature within the manufacturer’s limits under worst-case output power, ambient temperature, antenna mismatch, enclosure condition, and aging scenario.

Supply Risk: Why a Functionally Excellent GaN Part Can Still Be a Bad BOM Choice

High-power RF devices are often difficult to substitute because the transistor, package, bias point, matching network, thermal interface, and digital predistortion settings are interconnected. A replacement with similar frequency and output power may require a new PCB layout and extensive requalification. This makes source selection more important than the number of nominal alternatives found in an online catalog.

Lead times can also move sharply during operator deployment cycles. A part that is readily available during prototype development may become constrained when multiple radio programs ramp simultaneously. Procurement should not assume a fixed 20-week or 30-week lead time for an entire technology family. Lead time is manufacturer-, package-, capacity-, and demand-dependent and must be confirmed against the exact part number and required quantity.

Questions to ask before design lock

• Is the exact device in volume production, sampling, or pre-production?

• What wafer process, assembly site, and test location support the part?

• Is there a product-change-notification history or known package transition?

• What is the manufacturer’s standard lead time and current allocation status?

• Can the supplier support scheduled orders, bonded inventory, or a forecast agreement?

• Is there a pin-compatible alternative, or only a functionally similar device?

• How much redesign and requalification would an alternate require?

• Are evaluation boards, nonlinear models, reliability reports, and application support available?

GaN vs GaAs Procurement Decision Matrix

Calculate Total Implemented Cost, Not Device Price

The correct commercial comparison is the total implemented cost per compliant RF channel. Include the PA, pre-driver, matching components, bias controller, negative rail, DC-DC conversion, current monitoring, protection, PCB area, substrate requirements, heat spreader, thermal material, shielding, calibration time, yield impact, and qualification cost.

A Practical Sourcing Workflow for GaN and GaAs RF Components

1. Define the electrical operating point. Record frequency, bandwidth, average and peak output, modulation, back-off, gain, linearity, supply voltage, temperature, and load mismatch requirements.

2. Separate mandatory and preferred specifications. This prevents the team from rejecting viable alternatives over non-critical headline values.

3. Classify the component type. Identify whether the BOM requires an unmatched transistor, matched PA, PA module, gain block, LNA, switch, or complete front-end module.

4. Compare the full support circuit. Review matching, bias control, power sequencing, monitoring, thermal management, and PCB requirements.

5. Confirm lifecycle and production status. Do not base a production design solely on an evaluation sample or an online stock listing.

6. Build an alternate strategy before qualification. Determine whether an alternate can share the same footprint or whether a parallel PCB option is required.

7. Secure traceability. High-value RF semiconductors should be purchased with clear manufacturer, date-code, lot-code, packaging, and chain-of-custody records.

8. Plan incoming inspection. Depending on risk, this may include label and package inspection, X-ray, electrical verification, decapsulation through an approved laboratory, or comparison against known-good samples.

9. Align forecasts with the supply channel. Provide realistic demand timing so manufacturers and suppliers can reserve capacity or propose scheduled deliveries.

How Aurora Helps Reduce RF Component Procurement Risk

Aurora Components Co., Limited supports procurement teams sourcing RF, microwave, power-management, analog, connector, and control components for telecom and industrial applications. For GaN and GaAs requirements, our role is not limited to quoting a part number. We help customers review the purchasing context around the device, including exact suffix, package, lifecycle status, available documentation, requested date code, quantity, delivery schedule, and traceability expectations.

When a specified RF component is constrained, Aurora can help identify commercially available options for engineering review. Because RF substitutes are rarely drop-in replacements, any alternative must be validated by the customer’s engineering team for matching, bias, thermal behavior, linearity, stability, digital predistortion, regulatory compliance, and reliability. We clearly distinguish between pin-compatible, functionally similar, and redesign-required alternatives.

Final Recommendation: Switch When the System Benefit Is Measurable

GaN is the preferred direction for many high-power, high-density 5G transmit applications, particularly massive MIMO and infrastructure designs where efficiency and thermal performance create measurable system value. GaAs remains relevant in lower-power PAs, drivers, front-end modules, and low-noise receive functions where mature integration and cost are more important than maximum power density.

The best procurement decision is therefore not “GaN or GaAs” at platform level. It is a stage-by-stage decision based on operating conditions, total implemented cost, qualification effort, and supply continuity. Before switching, confirm the supporting power architecture, bias sequence, thermal stack, model availability, production status, and alternate-source strategy. A technically strong RF part only becomes a strong BOM choice when it can be sourced, assembled, qualified, and replenished throughout the product lifecycle.