The 6G community has effectively settled the spectrum question, and the answer is the 7 to 24 GHz range. 3GPP calls it Frequency Range 3 (FR3), while marketing departments have taken to calling it the “golden” band, which says more about the marketing departments than about the spectrum. The band is the last large block of cellular-suitable spectrum that isn’t already saturated by something else. The technical work is solvable. The coexistence work is harder.
Why this band and not another
Below 7 GHz, every desirable slice is allocated, refarmed, or aggregated. Above 24 GHz, the millimeter-wave deployment record from 5G has been chastening: high path loss, brutal building penetration, and a coverage economics problem that no amount of beamforming has solved cleanly. FR3 sits between these failure modes. At the lower end near 7 GHz, propagation is close enough to traditional sub-6 GHz cellular that conventional macro deployments remain plausible. At the upper end near 15 to 24 GHz, bandwidth opens up considerably while path loss remains within range of dense urban small cells equipped with larger arrays.
Samsung’s analysis frames FR3 as the band that can satisfy both capacity and coverage requirements, with EIRP for FR3 systems needing to be at least 6 dB higher than FR1 systems to match mid-band coverage. That number assumes the operator can use the band, which is where the difficulty starts.
The incumbents
The 7 to 24 GHz range is occupied by, among others, fixed satellite service downlinks and uplinks, Ku-band commercial satellite operations, government and military communications, fixed terrestrial microwave links, Earth exploration satellite service passive bands, radio astronomy bands, and radar systems including military X-band, weather radar, maritime navigation, and synthetic aperture radar.
The Federal Aviation Administration operates microwave links in the 7.125 to 8.4 GHz band to connect air traffic control centers with remote aeronautical radionavigation radars. Defense satellite communications systems use geostationary satellites in this range, with 7.25 to 7.7 GHz allocated for downlink and 7.9 to 8.4 GHz for uplink. Non-federal uses include unlicensed ultra-wideband devices operating in the 7.75 to 8.75 GHz range, covering item tracking, wall-penetrating radars, automotive radars, and wearable technology. The upper portion intersects the satellite Ku-band at 12 to 18 GHz, where commercial satellite systems are the incumbents, including Starlink and every other GEO and NGSO commercial satellite operator currently in service.
None of these incumbents will be evicted. Defense satellite communications operate under treaty obligations and serve mission-critical functions that cannot be replicated on alternative bands. Earth exploration satellite service passive allocations are protected under ITU Radio Regulation 5.340 with all emissions prohibited. Radio astronomy carries similar protections. Fixed satellite service operators have billions of dollars of orbital assets that will not relocate to accommodate a terrestrial mobile system. The only path forward involves sharing.
WRC-27 and what’s on the table
The 2027 World Radiocommunication Conference is scheduled for October 18 through November 12, 2027, in Shanghai, the first time the conference has been held in the Asia-Pacific region. Agenda Item 1.7 will consider IMT identification for three specific portions of the upper mid-band: 4.4 to 4.8 GHz in Regions 1 and 3, the 7.125 to 8.4 GHz range with regional variations, and 14.8 to 15.35 GHz globally.
The FCC’s IWG-2 advisory group preparing the U.S. position has been unable to reach consensus on Agenda Item 1.7. The pro-IMT view is supported by CTIA, AT&T, Ericsson, GSMA, Nokia, Qualcomm, T-Mobile, and Verizon. A more cautious view is supported by Apple, Boeing, Broadcom, Charter, Comcast, Lockheed Martin, Planet, and the Satellite Industry Association, among others. The European position has been similarly contentious.
The European Commission’s Radio Spectrum Policy Group has noted that certain frequency bands being considered for IMT identification at WRC-27, including 4.4 to 4.8 GHz, 7.25 to 8.4 GHz, and 14.8 to 15.35 GHz, may jeopardize usages relevant to the Common Security and Defense Policy or to the EU’s space policy. None of this is unusual for a WRC cycle but what is unusual is the scope of equity at stake and the inadequacy of traditional sharing solutions. The conventional approach of geographic exclusion zones around incumbent earth stations does not scale when the incumbent is a low-Earth-orbit constellation.
Where filters fall apart
Whatever WRC-27 produces, the result will not be a clean contiguous block. It will be a set of channelized allocations with hard adjacent-band rejection requirements driven by the need to protect specific incumbent services. That is a filter problem, and at FR3 frequencies it is a problem the existing component ecosystem is not ready for.
Acoustic filters have been the workhorses of mobile front ends below 6 GHz for decades. Surface acoustic wave filters dominate below roughly 3 GHz. Bulk acoustic wave filters, particularly film bulk acoustic resonators, extend the range upward but encounter problems above 10 GHz. Although FBARs are commercially successful, their ultrathin piezoelectric layers, heavy metallization loading, and intrinsic mechanical losses hinder scaling beyond 10 GHz.
An FBAR or film bulk acoustic resonator, is a piezoelectric device in which a thickness-mode bulk acoustic wave resonates within a thin film, typically aluminum nitride, sandwiched between top and bottom electrodes. The resonant frequency is set by the film thickness, so higher frequencies require thinner films. An acoustic mirror or an air cavity beneath the stack confines the energy and sustains a high quality factor, yielding low insertion loss and good power handling below roughly 6 GHz. Above about 10 GHz the film becomes impractically thin, and insertion loss, quality factor, and power handling degrade together.
In contrast, an XBAR, or laterally excited bulk acoustic resonator, is a piezoelectric resonator in which interdigitated electrodes patterned on a single surface of a suspended thin-film membrane apply a lateral electric field that excites a bulk acoustic mode propagating through the membrane thickness. The membrane is typically thin-film lithium niobate. The excited mode is the first-order antisymmetric (A1) Lamb mode.
The configuration decouples resonant frequency from electrode pitch. In a SAW resonator, frequency is governed by interdigital transducer pitch, which constrains operation above approximately 3 GHz. In an FBAR, frequency is set by film thickness. In an XBAR, resonant frequency is determined primarily by membrane thickness and mode order, permitting operation from 10 to 25 GHz and, using higher-order antisymmetric modes, beyond 100 GHz, while retaining a membrane thickness compatible with available fabrication processes.
Lithium niobate provides high electromechanical coupling, yielding wide fractional bandwidth. The suspended membrane, released with free surfaces on both faces, confines acoustic energy and sustains high quality factor, holding insertion loss low. Periodically poled piezoelectric film, in which alternating poled layers are stacked, extends operation to higher frequencies and higher-order modes. Demonstrated tri-layer periodically poled lithium niobate filters have achieved 2.2 dB insertion loss, 8.5 percent fractional bandwidth, and 49 dB close-in rejection at 19.3 GHz.
DARPA’s Compact Front-End Filters at the Element-Level program has funded substantial work on XBARs in suspended thin-film lithium niobate, which can scale beyond 10 GHz and even above 100 GHz while maintaining high electromechanical coupling and quality factor. Recent results from the University of Texas at Austin demonstrate a tri-layer periodically poled lithium niobate filter at 19.3 GHz with 2.2 dB of insertion loss, 8.5% fractional bandwidth, and 49 dB of close-in rejection. An earlier prototype achieved 1.79 dB of insertion loss at 20.5 GHz with out-of-band rejection greater than 14.9 dB across the FR3 band, in a footprint of 0.9 by 0.74 mm². These are laboratory results, not production parts, but they suggest that the underlying physics permits acoustic filtering through at least the lower half of FR3.
What is not yet clear is whether acoustic technology alone will be the predominant solution. Electromagnetic alternatives such as ceramic, cavity, and dielectric resonator filters work at these frequencies but bring size and integration penalties that mobile handsets will not tolerate. Tunable filters using BST, MEMS, or YIG technologies offer a partial answer for base stations but introduce their own loss, linearity, and reliability constraints. The likely outcome is a mix: integrated acoustic filters where possible, cavity or dielectric filters at base station level, and aggressive use of digital filtering to clean up what passes through.
The power amplifier question
Power amplifier efficiency degrades with frequency as parasitic capacitance, transit time, and losses in matching networks all scale unfavorably. At sub-6 GHz, gallium arsenide and laterally diffused MOS handle most of the cellular workload. Above 7 GHz, gallium nitride becomes hard to avoid for any application requiring meaningful output power. Power amplifiers have degraded power efficiency at these frequencies, which becomes a design constraint that the rest of the system has to accommodate.
This is pretty good news for the GaN supply chain as MACOM, Wolfspeed, Qorvo, and the smaller pure-play GaN houses have been positioning for this transition for several years. GaN-on-silicon for high-volume base station applications and GaN-on-SiC for higher-performance defense and infrastructure roles cover most of the FR3 PA design space.
At MWC 2026, MediaTek demonstrated a reference design using the Skyworks SKYR60002, a 6G FR3 LNA and power amplifier module with integrated filters designed to support the 6.425 GHz to greater than 7 GHz spectrum in the latest 3GPP standard. The module is specified for high linearity, wide bandwidth, and robust thermal performance.
Antennas and the array problem
Extracting useful capacity from FR3 will require large antenna arrays at both the base station and, increasingly, the user equipment. Samsung’s coverage analysis assumes that X-MIMO with extremely large-scale massive MIMO can offer more than twice the average user spectral efficiency of mid-band massive MIMO, but the additional path loss means EIRP at FR3 needs to be at least 6 dB higher than FR1 to match coverage. At 13 GHz, an antenna element is roughly 1.15 cm across, which makes 256-element and 512-element base station arrays practical.
The handset side is harder. A 6G handset using FR3 must accommodate multiple antenna elements at frequencies where the wavelength is comparable to the device thickness. Antenna-in-package solutions developed for 5G millimeter-wave frequencies provide a template, but lower frequency means larger elements, which means fewer elements can fit, which means less beamforming gain to compensate for path loss.
The path that ends in deployment
Assuming WRC-27 produces an IMT identification for at least some FR3 bands, the deployment path runs through dynamic spectrum sharing, large antenna arrays at base stations, spatial nulling toward incumbent satellite earth stations and overhead satellites, and front-end components that do not yet exist in production volume. Each is a serious engineering problem. The collective challenge is whether they can be solved on a schedule compatible with the IMT-2030 timeline, which anticipates initial 6G deployments around 2030.
The honest answer is that nobody knows. The filter work is moving fast but has not reached commercial maturity. The PA work is incremental from a well-understood baseline. The coexistence work is partly technical and partly political, and the political part is harder. WRC-27 will resolve some of the regulatory uncertainties. It will not resolve the question of whether terrestrial 6G can share Ku-band downlinks with LEO constellations without degrading either service to the point of uselessness, and that question may take another decade to settle. In the meantime, FR3 is the band the industry is committed to and the component ecosystem will have to catch up.