SpaceX has completed design of a miniature rooftop cell tower intended to eliminate indoor dead zones for Starlink satellite coverage, with production expected to begin in 2027, CEO Elon Musk said at the DealBook Summit this week.

The device addresses a fundamental limitation of satellite internet: signals struggle to penetrate buildings, particularly those with metal roofs. The company's approach pairs this terrestrial layer with its Direct-to-Cell V2 satellite constellation, launching next year with a targeted 40-fold bandwidth increase.

SpaceX holds $25 billion in spectrum—a $17 billion deal with EchoStar in 2025 and an $8 billion low-band acquisition this year—that underpins the economics of this hybrid model. SpaceX received FCC approval for direct-to-phone Starlink service in 2024 and operates under a partnership with T-Mobile.

Musk characterized the rooftop unit as a simple installation, similar to a Starlink dish. SpaceX President Gwynne Shotwell indicated in August earnings guidance that the network would use small, low-power cellular base stations rather than conventional towers. The company is still determining bundling strategy with Starlink terminals.

The critical distinction in SpaceX's framing: this is not a frontal assault on AT&T or Verizon's consumer voice-and-messaging business. Instead, Musk positioned the hybrid network as a play on a new demand curve driven by artificial intelligence and machine-to-machine traffic. He predicted "tremendous" future demand for wireless bandwidth from AI and robotics, suggesting telecom incumbents might experience revenue loss but face no existential threat.

The business model hinges on capturing bandwidth consumption from AI workloads and autonomous systems—data-intensive use cases that operate at different price points and service characteristics than consumer mobile. This reframes SpaceX's $25 billion spectrum spend not as a competitor to legacy wireless, but as infrastructure for a category of demand that did not exist five years ago.