This article explores the evolving landscape of RF power suppliers and the critical transitions seen as NXP concludes its Radio Power product line.
The first article in this series, 6G Has a Date Now – Your RF Front End Has a Deadline, shows that new spectrum near 7 GHz will call for new RF front ends. The RF power supplier base these designs will depend on is changing. NXP is winding down its Radio Power product line and closing its ECHO gallium nitride (GaN) fab in Chandler, Arizona.
Designs that still use NXP RF power parts now need qualified alternates from other suppliers. This article maps who still makes RF power devices, where that supply is moving, and how to keep an RF BOM resilient through the transition.
Both NXP Radio Power last-time-buy windows have now closed. The first discontinuation notice, as updated, set a June 10, 2026, last-time-buy date, with deliveries ending December 10. The second notice closed ordering on September 30, 2026, with final deliveries extending through September 30, 2027.
There is no universal list of NXP replacements. The viable supplier set shifts with frequency, output power, package, power-amplifier (PA) architecture, qualification requirements, and whether the design serves telecom, industrial, or defense markets.
For sub-6 GHz wireless infrastructure, several suppliers have active GaN portfolios. Sumitomo Electric Device Innovations (SEDI) manufactures GaN high-electron-mobility transistors (HEMTs) for mobile base stations, including 5G applications. RFHIC lists production GaN-on-silicon carbide (SiC) transistors for 4G and 5G macro infrastructure, including products across the 3.4 to 5.0 GHz range.
Mitsubishi Electric has shipped samples of three GaN power-amplifier modules for 5G massive MIMO base stations since 2023, most recently in March 2025. Qorvo lists production infrastructure GaN power amplifiers and broader RF front-end devices. Ampleon covers laterally diffused metal-oxide-semiconductor (LDMOS) and GaN, while MACOM offers GaN-on-SiC products, monolithic microwave integrated circuits (MMICs), and in-house foundry capacity.
Below 6 GHz, the supplier map is well established. Near 7 GHz, production parts for cellular infrastructure are scarce. Mitsubishi Electric has demonstrated a prototype 7 GHz GaN power-amplifier module for 5G-Advanced. RFHIC offers a 5.8 to 7.2 GHz GaN amplifier at the sample stage, and Qorvo lists a production 6 to 12 GHz GaN power amplifier aimed at broader RF applications. GaN works at these frequencies, but a production supplier pool for Frequency Range 3 (FR3) cellular infrastructure has not yet fully formed.
Spun out of NXP’s RF Power business in 2015, Ampleon has the clearest historical link to its former parent. Through spring and summer 2026, it added several wireless infrastructure devices to its catalog, including a 70 W GaN Doherty transistor for 3.4 to 4.0 GHz, a 55 W GaN device for 4.8 to 5.0 GHz, and a fully integrated 70 W GaN massive-MIMO power-amplifier module.
On its May 2026 earnings call, MACOM management said any new revenue from a competitor’s RF-power exit would probably not arrive until the second half of 2027. Customers are still working through last-time buys, and moving an RF power stage to a new supplier takes redesign and qualification work before any volume ships.
In July 2025, MACOM also assumed full control of the former Wolfspeed RF wafer fab in Research Triangle Park, roughly six months ahead of schedule. The fab, an accredited U.S. Department of Defense (DoD) Trusted Foundry, runs GaN-on-SiC processes for RF power devices and MMICs used in telecom and defense.
Another part of the RF supplier base has consolidated. Skyworks announced its agreement to acquire Qorvo in October 2025, received the last regulatory clearances on September 30, 2026, and closed the transaction in early October.
Large semiconductor mergers can lead to portfolio rationalization where product lines overlap. Until the combined roadmap is clear, lifecycle review should precede any new design-in that depends on overlapping product families from Skyworks or Qorvo. The consolidation risk extends beyond high-power base-station PAs, as both companies operate across multiple RF front-end tiers.
China is building more of the RF GaN stack at home. SICC supplies semi-insulating SiC substrates for GaN RF devices, while Sanan IC operates a 4-inch GaN-on-SiC foundry platform aimed at massive-MIMO base stations and future FR3 applications. Sanan says its 5G PA process has reached stable customer mass production, although whether a given program can use that capacity depends on qualification, sourcing, and trade-control constraints.
Lifecycle resilience depends on seeing changes early enough to act and having a process ready to respond. These four practices shorten the time between a supplier change and the engineering response.
1. Check lifecycle status at design-in. Identify single-source or aging parts while the schematic and qualification plan are still open, so you have time to evaluate alternates without forcing a late redesign.
2. Qualify alternates while the primary device is still in production. For RF power, qualification can include matching, bias, thermal behavior, layout, linearity, predistortion, and production testing. Complete that work before relying on an alternate as a second source.
3. Give product change notifications (PCNs) a named owner and a defined process. A named owner logs each notice, identifies the affected BOMs and programs, assigns a disposition (alternate qualification, last-time buy, redesign, or no action) with a due date, and tracks it to closure. Without a defined process, a PCN lands in a single inbox and gets forwarded with no record of who decided what.
4. Watch inventory and lifecycle signals between formal notices. Distributor stock trends and lifecycle status can move before or alongside a PCN. Octopart’s BOM Tool shows distributor availability and lifecycle status for the matched parts on a BOM, so you can see falling inventory on an RF power part before a notice arrives.
Five dates mark the NXP wind-down and the Chandler handover, from December 2026 through early 2029.
Choosing an RF power device in 2027 starts with three supplier questions: who can build it at the target frequency, where is fab capacity shifting, and which sources are likely to remain viable throughout the product’s service life.
One question remains: can a part that exists, qualifies, and has stock ship to where the product is built? The concluding article in this series, The Gallium Problem in Your Radio: Export Controls and Single-Source Front Ends, answers it.