SpaceX used its first earnings call as a public company this week to say out loud what the industry had been speculating about for months: it intends to build a terrestrial mobile network and go head-to-head with AT&T, T-Mobile, and Verizon. SpaceX President Gwynne Shotwell told investors during the Q&A that the EchoStar spectrum “does have terrestrial components, so we definitely intend to build out the terrestrial component.” The company is picking up 65 MHz of bandwidth through its EchoStar acquisition, a jump from the roughly 5 megahertz it currently accesses through carrier partnerships for direct-to-device service.
That’s the headline. The detail that matters more for this industry is how SpaceX says it plans to build it — and it’s not the tower-leasing story it might sound like at first read.
A network built around rooftops, not tower sites
Shotwell described the buildout in terms of housing “these little femtocells” inside the same hardware that already holds a customer’s Starlink broadband dish, deployed incrementally as needed rather than as a large upfront capital commitment. Musk was more direct about the intent behind that design choice: he said the alternative to very expensive and difficult-to-locate large cellular base stations is deploying a large number of small stations instead. He described the approach as installing Starlink dishes that also provide mobile-band connectivity, sited on the roofs of houses and businesses that already have a clear view for connecting to phones on the ground.
This isn’t a model where every Starlink Mobile customer needs hardware on their own roof. The rooftop units are network infrastructure, not customer premises equipment for cellular service — a phone connects to the nearest small cell over the air, the same way it connects to a macro tower today. The rooftop hardware rides on existing Starlink broadband dish locations, a subset of SpaceX’s existing subscriber base and other host sites, used as a low-cost way to distribute small cells across a coverage area. SpaceX only needs enough host rooftops to build sufficient density in a given market, not a dish at every mobile subscriber’s home — a meaningfully smaller buildout problem than either a macro tower network or a dish-per-customer model.
Read plainly, that’s SpaceX describing a strategy to avoid conventional macro tower infrastructure, not lease into it. Shotwell framed the approach as capital efficiency, saying it means SpaceX won’t need to spend “10 or many billions of dollars” upfront before deploying service. A rooftop-mounted femtocell riding on hardware a customer already owns is about as far from a leased slot on a lattice tower or monopole as a buildout strategy gets — and it’s worth stating plainly, because the premise going around industry circles, that a SpaceX terrestrial network necessarily means new lease revenue for the major tower companies, isn’t what the company itself described this week.
Shotwell framed the addressable opportunity around the roughly $600 billion the Big Three carriers collect annually in the U.S., saying she expects Starlink to “acquire quite a few” of their customers because the service will be better, pointing to eliminated dead zones and stronger performance during natural disasters. Commercial service is targeted for the end of next year. Even with the spectrum increase, analysts note SpaceX still holds far less total wireless spectrum than any of the three major carriers, meaning it will likely need additional spectrum or unusually efficient infrastructure to compete on a nationwide footing.
Tower REITs aren’t acting worried — and the market agrees
If this announcement was meant to rattle the tower industry, the market didn’t get the memo. Crown Castle, American Tower, and SBA Communications all reported quarterly results in line with or ahead of expectations around the same period, and each stock traded higher, not lower, in the days after. That’s not the reaction you’d expect if Wall Street believed a fourth facilities-based carrier built on rooftop femtocells was coming for tower company revenue anytime soon.
SBA Communications CEO Brendan Cavanagh offered the more grounded read among tower executives, saying SBA has already been in talks with more than one satellite company about deploying terrestrial gear — radios or ground stations — to complement direct-to-device service, and that tower companies stand to benefit because satellite operators will eventually need physical terrestrial infrastructure to truly mimic a mobile carrier. New Street Research analyst David Barden was less convinced, writing bluntly that he’s “highly skeptical this is even a real thing.” Crown Castle, for its part, said it hasn’t seen any change in radio deployment activity tied to satellite competition so far — which, set against Musk and Shotwell’s comments about rooftop-mounted small cells rather than macro site leases, tracks with what SpaceX itself described on the earnings call.
The physics argument tower investors keep coming back to
The more substantive pushback isn’t about SpaceX’s capital strategy — it’s about geometry. One investment note circulating among tower-focused investors called renewed satellite fears “overstated,” arguing Starlink is a real business worth watching but not a serious replacement threat to terrestrial infrastructure in cities and suburbs, where towers sit far closer to users than any satellite constellation can. That framing lines up with how the Big Three carriers have positioned satellite-to-device service themselves: as a dead-zone and emergency-coverage supplement, not a replacement for dense macro networks — which is also, notably, close to how Shotwell herself framed Starlink Mobile’s advantage.
A Possible Small-Cell Classification — With Real Permitting Stakes
The added radio unit is where this could shift from a consumer install to genuine contractor work. Today’s Starlink dish is explicitly self-install — low-voltage cable, plug into a standard outlet, no electrical or building permit typically required. A persistent cellular transmitter mounted alongside it is a different animal: if municipalities classify it as small-cell wireless equipment rather than consumer satellite gear, it would likely need its own dedicated electrical circuit pulled by a licensed electrician, plus a separate small-cell antenna permit of the kind many jurisdictions already require apart from routine residential dish installations, and possibly a structural/building permit, depending upon the size of the unit, given the added weight and wind load on the roof mount. That would open a real, if narrow, lane of work for small-cell and DAS-type contractors — distinct from tower crews and from today’s satellite installers alike.
None of this is confirmed, though; SpaceX hasn’t detailed the hardware, power requirements, or installation process for the terrestrial component, so how it gets classified and permitted remains an open question until the company says more.
Reading it all together
Put the earnings-call comments next to the tower industry’s own reaction and a fairly coherent picture emerges: SpaceX’s stated buildout plan is deliberately structured to avoid conventional tower leasing, and the tower industry’s own executives and covering analysts don’t appear to be pricing in a threat from it. Cavanagh’s more optimistic framing, that satellite operators will eventually need towers too, points to ground stations and backhaul infrastructure rather than the femtocell-on-a-rooftop model SpaceX actually described. That’s a meaningfully different, and much smaller, opportunity than a wave of new colocation leases across tens of thousands of tower company sites.
There’s also a tenant-concentration reality sitting underneath all of this that’s easy to lose in the SpaceX headlines: SBA drew 66% of its 2024 U.S. revenue from AT&T, T-Mobile, and Verizon combined; Crown Castle’s rental revenue from the same three carriers has run around 75% in recent years; American Tower’s U.S. and Canada property segment revenue was roughly 86% carrier-concentrated. A genuine fourth facilities-based competitor would matter enormously to that concentration risk, for better or worse. A rooftop small-cell network that explicitly avoids leasing tower space doesn’t move that number in either direction — it’s simply a different business, built on different real estate, competing for the same customers’ phone bills rather than the same physical infrastructure.
A beat that didn’t matter
More than 911 million SpaceX shares were unlocked this morning, more than doubling the publicly available stock in Elon Musk’s space, connectivity, and AI company, which has struggled since its record-breaking IPO just weeks ago that opened at $135 and quickly peaked at $225.64 on June 16.
In early morning trading it fell to a low of $105.11, despite the company beating Wall Street’s expectations in its first earnings report as a public company Tuesday — posting $7.81 billion in second-quarter revenue against a consensus estimate of $6.93 billion, a 92% jump from a year earlier, while narrowing its net loss to $541 million from roughly $1 billion in the same period last year.
The earnings beat wasn’t enough to hold the stock: capital expenditures jumped sixfold to $18.4 billion, the bulk of it AI-related spending, spooking investors already bracing for today’s lockup expiration and pushing shares down more than 13% in the session that followed.
