SpaceX filed an FCC application for 100,000 Gen3 Starlink satellites with increased capacity to solve orbital data center downlink bottlenecks.

SpaceX has filed with the FCC for a 100,000-satellite Gen3 NGSO (non-geostationary orbit) system. The satellites will weigh 2,000–2,500 kg each with a deployed area of 300–400 square meters, operating in orbits between 320–480 km at various inclinations. According to the FCC application, the Gen3 constellation will deliver extremely low-latency, multi-gigabit symmetrical throughput, with the sheer number of satellites ensuring multiple spacecraft are always visible from any point on Earth.

The filing positions Starlink as AI-connectivity infrastructure rather than simply a satellite broadband network. The system builds upon SpaceX's current fleet of over 10,000 V1.5 and V2 Starlink satellites already in orbit.

The Gen3 system will use spectrum in Ku-, Ka-, V-, and E-bands that the FCC previously granted for SpaceX's Gen2 system. Critically, SpaceX is also requesting to operate in greenfield W- and D-band frequencies between 92 and 275 GHz—spectrum not yet widely allocated for satellite use. Each satellite payload will employ advanced phased-array beamforming and digital processing technologies to make highly efficient use of spectrum resources and share spectrum flexibly with other space-based and terrestrial licensed users. User terminals will feature highly directive, electronically steered antenna beams that track the constellation. Gateway earth stations will generate high-gain steered beams to communicate with multiple satellites from a single site. SpaceX will implement advanced optical inter-satellite links on the Gen3 system to provide seamless network management and service continuity while minimizing the spectrum footprint and facilitating spectrum sharing with other systems. Broadband service will be available to residential, commercial, institutional, governmental, and other users worldwide.

The core workhorse bands—already largely authorized for Gen2 and carried forward or augmented—include Ku-band (approximately 10.7–13.4 GHz class downlink ranges with associated uplinks), primarily for user links; Ka-band (approximately 17.3–21.2 GHz and higher gateway/user ranges) for higher capacity; V-band (approximately 37.5–42.5 GHz and related) for significant capacity expansion; and E-band (previously authorized approximately 71–76 GHz space-to-Earth and 81–86 GHz Earth-to-space for Gen2 gateways and feeder links).

The new W- and D-band spectrum (92–275 GHz) primarily unlocks the next leap: reliable multi-gigabit symmetrical service to far more customers simultaneously, plus the massive uplink and backhaul needed for AI devices. Current V3 satellites deliver approximately 1 Tbps downlink and 160–200 Gbps uplink per satellite. A single Starship flight of approximately 60 V3 satellites can add roughly 60 Tbps of network capacity. With dense VLEO (very low Earth orbit) shells, advanced digital beamforming, and optical inter-satellite links, this supports multi-gigabit peaks and gigabit sustained service for many users under good conditions, especially with upgraded terminals. However, rain fade, beam sharing, and backhaul constraints still limit how many customers achieve true multi-gigabit speeds simultaneously and how symmetrical the service can be, with uplink remaining the more challenging direction for high rates.

The W- and D-band frequencies will primarily serve gateway feeder links and backhaul, removing the bottleneck that currently prevents satellites' full user-link capacity from being delivered to many users simultaneously. This also enables higher-order modulation, more beams, and better symmetry, as uplink capacity can scale more freely. The filing explicitly ties the new bands and optical inter-satellite links to "multi-gigabit symmetrical throughput" and the "massive uplink capacity" required for AI applications involving high-definition spatial and auditory data, industrial automation, and edge devices.

SpaceX typically iterates incrementally, starting with spectrum bands for which it already has authorization. Immediate V3 early flights in 2026–2027, Starship-enabled, will rely heavily on improved core workhorse bands already largely authorized for Gen2: Ku-band for volume user links, Ka-band for higher-capacity user and gateway services, V-band for significant expansion of user and feeder capacity, and E-band for high-capacity feeder links. These will feature bigger phased arrays, better digital beamforming, higher-order modems, and more efficient power amplifiers with denser frequency reuse. Optical inter-satellite links will handle most inter-satellite traffic. This phase alone will deliver multi-gigabit peaks and sustained gigabit-plus service for many users. New user terminals with larger, more advanced phased arrays will be required for full performance.

Mid-term Gen3 and early V4 iterations will see progressive introduction of W- and D-band on gateways and then satellites as hardware matures and spectrum access is granted through waivers, non-conforming operations, and coordination agreements. This rollout will start on high-capacity gateways—where power and thermal constraints are more manageable—before migrating to space hardware. This phase unlocks true multi-gigabit symmetrical service at scale and accommodates the AI uplink tsunami. Full multi-gigabit symmetrical capability plus AI-scale deployment will come from the complete 100,000-satellite Gen3 constellation density, mature W/D hardware, refined sharing frameworks, and widespread upgraded terminals, with an optical mesh further reducing RF spectrum pressure.

SpaceX will likely ramp to 300–400 Starship launches per year for communication satellites to achieve a three-year deployment and then ongoing replacement and maintenance of the constellation. Additional launches will carry AI-dedicated satellites.

V3 satellites are already approximately 2,000 kg class. Adding high-band arrays, more solar capacity, better thermal management, and higher-capacity digital payloads will push mass, power, and complexity higher, though Starship's payload capacity can accommodate this growth. Expect larger solar arrays, more efficient argon or next-generation thrusters, advanced heat rejection systems, and denser electronics.

For V4 and mature Gen3 versions, further modifications will include even larger, higher-performance phased arrays with multi-band shared-aperture or stacked configurations; more sophisticated hybrid digital and analog beamforming with onboard AI for dynamic spectrum and interference management; higher-capacity optical inter-satellite links already on a trajectory toward multi-hundred-Gbps and Tbps-class performance; improved propulsion and station-keeping for denser VLEO shells; better space-to-ground optical links as secondary high-capacity connections; user terminals with multi-band phased arrays supporting higher frequencies, higher effective isotropic radiated power and gain-to-noise-temperature ratios, and better power efficiency; gateway evolution into dense W/D-capable sites with optical fiber backhaul; and software incorporating more aggressive dynamic beam and power control, spectrum sharing algorithms, and AI-driven traffic prediction.

Moving from E-band (approximately 70–90 GHz) into W-band (roughly 75–110 GHz) and especially D-band (110–170+ GHz up to 275 GHz) represents a significant technical step. Phased-array element size scales with wavelength; at 100+ GHz, elements become millimeter-scale or smaller, allowing arrays to pack enormous numbers of elements into the same physical aperture, yielding higher gain, narrower beams, and better spatial reuse. However, manufacturing tolerances become extreme—requiring micron-level accuracy—and mutual coupling, surface waves, and fabrication defects become harder to control. Active electronics including millimeter-wave monolithic integrated circuits and beamformers require high-performance III-V semiconductors such as indium phosphide or gallium arsenide, or advanced silicon germanium and CMOS silicon-on-insulator technologies. Power amplifiers suffer rapidly declining efficiency and saturated output power as frequency increases, while low-noise amplifiers degrade similarly. Hybrid beamforming combining analog sub-arrays with digital processing becomes nearly mandatory to keep power consumption and complexity manageable.

Power and thermal constraints are severe on space platforms. High-frequency power amplifiers are inefficient, generating excess heat and increasing solar array and battery demands. Cooling via radiators, heat pipes, and advanced materials becomes critical. On the ground, user terminal power consumption and heat also rise significantly with higher frequencies.