Why Musk's Starlink Mobile Plan Gets Costly in Cities

SpaceX President Gwynne Shotwell told investors on a recent earnings call that Starlink's satellite mobile service would provide better service than AT&T, Verizon and T-Mobile by the end of 2027. Benedict Evans, a former Andreessen Horowitz partner who has studied mobile technology for decades, is skeptical the plan survives a dense city.

"The hard part is having coverage in a city with steel and concrete blocking every signal past 50 yards," Evans wrote on Threads on Thursday. "That's the cost."

Starlink Mobile is, for now, a satellite safety net for phone carriers. The company's satellites act as cell towers in space, connecting handsets when terrestrial networks drop out, which mostly means remote areas and emergencies. The service pulled in about $1.66 billion in the second quarter and ranks as the most profitable revenue stream across Elon Musk's companies. Shotwell's 2027 claim flips that relationship: the gap-filler becomes the service that outperforms the carriers it works with today.

The expansion plan is where the economics get contested. Musk said on the earnings call that Starlink would deploy "a large number of small stations" instead of full-size cellular base stations to bring coverage into built-up areas. Small stations are cheaper to put up and can be mounted on street furniture and building rooftops. The open question is whether the savings survive real city density.

Physics works against satellite here. A direct-to-cell signal needs a reasonably clear line of sight to the handset, and a handset in a city is usually indoors, behind glass and reinforced concrete. Low Earth orbit satellites also pass overhead in minutes, so the connection has to be finished during a short window.

Terrestrial carriers faced the same problem years ago and answered it with dense layers of small cells and in-building systems, because outdoor macro towers could not reach indoor users. A satellite signal must clear the same obstruction from a much longer distance.

Evans's question cuts to the one place the economics could still work. "Can you save enough money from using satellite for rural and for backhaul on some (how many?) of your base stations to deliver a meaningful cost advantage," he said. "What's the algebra?"

Backhaul, the data link that carries traffic from a local station back to the core network, is where running fiber is often the biggest real cost. The whole question comes down to how many stations actually need that fiber. Replace it with satellite on enough of them and per-station costs fall below what a terrestrial build pays. If the station count needed to reach indoor urban users stays high, the advantage gets diluted until it disappears.

Separately on Friday, Evans posted a chart showing urban and suburban users generate the bulk of monthly internet traffic. Starlink would have to reach those users by building its own small-station network or renting access from the very companies it hopes to displace. A countryside-only service keeps the addressable market thin. Reaching downtown users means building the same dense small-cell network the carriers already paid for.

Shotwell's target also has to account for the incumbents. The three carriers have spent much of the past decade densifying their own networks, so the service-quality benchmark Starlink is chasing will keep moving. A ground build of the scale the small-station plan implies, for its part, usually runs through years of permitting and siting disputes before the first station goes live.

Starlink did not immediately respond to a request for comment.