There are two ways to steer a phased-array beam: shift phases in analog hardware (cheap, power-efficient, rigid) or do it digitally (flexible, capable, power-hungry and expensive). US11652525B2, granted to SpaceX on May 16, 2023, claims doing both, covering "beamforming in hybrid beamforming antennas."
The CPC is concise and pure RF: H04B 7/0617 (transmit beamforming using multiple antennas), H04B 7/0686 (feedback/adaptive beamforming), and H04B 17/11 (transmitter calibration). The calibration code is telling — hybrid architectures live or die on keeping their analog and digital halves coherent.
The mechanism is partition. Analog beamforming forms coarse beams in the RF front end at low cost and power; digital beamforming then refines and multiplies them in software for flexibility — multiple simultaneous beams, fast reconfiguration, interference nulling. By splitting the work, a hybrid array gets much of digital's capability without paying the full digital power-and-cost bill. For SpaceX, multiplying that decision across millions of user terminals and thousands of satellites is exactly where the economics are decided.
Beamforming is the battleground, and the hybrid split is how you fight it affordably at scale. The dependent claims — how the analog and digital stages are coordinated and calibrated — are where the practical moat sits.
Claim-scope honesty: hybrid beamforming is an established class of architecture with broad prior art, so this protects SpaceX's specific systems and methods, not the concept. SpaceX's antenna portfolio (US10770790B1, US11695222B2, US11563270B1, among others) is dense and consistent, and this grant is one well-placed brick in that wall — a clear read on how the company engineers its arrays to be both capable and cheap enough to deploy by the millions.