Theses›Telecoms & wireless›RF front-end silicon

RF front-end silicon

RF front-end - the qualification moat, not the process, wins sovereign and defence demand

  • Bear
  • Early
  • $126B by 2035

The call6G and rearmament need GaN at new frequencies, but Qorvo, Skyworks, Broadcom and Qualcomm hold the sockets behind patent walls, and no European company holds either of the two narrow openings.

D1D2R1R2C1foragainst D1D2R1R2C1foragainst
Fig. 1The balance. Drivers and catalysts against restraints, each block sized by its weight; catalysts hang dashed because they have not happened yet, sized by their odds. It tips toward the restraints.
  1. D1Radio moves up to 6G's 7-24 GHz FR3 bands and out to satellites talking to phones, and today's handset front-ends are tuned for sub-6 GHz.
  2. D2NATO's 3.5% defence target and European sovereignty create buyers who need radar and secure-link GaN from a source they can qualify.
  3. R1Patent walls and handset lock-in hold, the market grows 4.5%, and the generic out-amplify-Qorvo play breaks every time.
  4. R2The two wedges exist on paper with no European company holding either, and defence GaN at 10% a year is a niche.
  5. C1WRC-27 identifies 7.125-8.4 GHz for mobile, starting a new FR3 power-amplifier generation on GaN foundry capacity.Nov 2027
Rests onthe beliefs underneath
What would prove itdated tests, open and resolved
202720282029today123watch 202720282029today123watch
Fig. 2Each ring is a dated test: dashed while open, struck through when overdue, ticked or crossed once resolved.
  1. 1WRC-27 outcome on the 7.125-8.4 GHz band, the dated trigger for a new FR3 amplifier generation.Dec 2027 · open
  2. 2imec's FR3 GaN-on-silicon result reaches pilot-line yield at FAMES, and the Skyworks-Qorvo close.Dec 2027 · open
  3. 3A European defence GaN pilot goes to a non-incumbent, or a GaN-on-Si or 6G-filter spin-out wins a design-in on GlobalFoundries capacity.Dec 2028 · open
Expressionslisted, private, technologies
  • 2379.TWRealtek Semiconductor
    • fair
  • AVGOBroadcom
    • cheap
    Custom ASICs for Google/Meta + networking silicon.
  • MXLMaxLinear
    • cheap
  • QRVOQorvo
    • fair
  • SWKSSkyworks Solutions
    • full
  • QCOMQualcomm
    • not screened
    Integrated RF front-end modules, PAs, filters (ex-RF360) and antenna tuners sold with its modems to Android handset makers.
  • 6981.TMurata Manufacturing
    • not screened
    SAW/XBAR filters, multiplexers and RF modules sold to handset and connectivity-module makers.
  • NXPINXP Semiconductors
    • cheap
    RF power transistors (LDMOS, GaN) for cellular base stations plus automotive and industrial RF transceivers.
Private · 7 private companies, 4 of them in the US.

SignalNova · Univity · Arralis · Movandi · MixComm · Falcomm · Soctera

3 more private companies tracked, not shown

Who has already bet on itprivate rounds, from public sources

$63M raised by 4 companies, $55M of it in the last three years. Wave building: Money is arriving: recent rounds, but no runaway leader yet.

Backed by a strategic investor: Movandi.

What changedthe record, newest first
  • 23 Jun 2026The Skyworks-Qorvo merger that thins the RF supply base is now tracking to a possible late-2026 close, and imec just posted hard GaN-on-silicon numbers for the 6G upper bands. Neither changes the call (the market grows too slowly to chase broadly), but the close date is the moment a credible second-source challenger could get design-win oxygen.
  • 1 Jun 2026RF front-end is a $50B+ market growing at only ~4.5% a year, dominated by four incumbents with deep filter-and-epitaxy patent walls. The interesting seed question is not a me-too amplifier; it is whether a materials edge (gallium-nitride-on-silicon for the 6G upper bands) or a sovereign defence/satellite niche can open a door the incumbents leave ajar.