SpaceX used its first public earnings call to tell investors it intends to build a ground-based mobile network and take customers away from America's largest wireless carriers. President and Chief Operating Officer Gwynne Shotwell sized the opportunity on that call as "the big three in the U.S., AT&T, Verizon, and T-Mobile, roughly between them, $600 billion a year" — while the three companies' own 2025 annual disclosures add to about $352 billion. SpaceX did not say what the larger figure covers.
The discrepancy is worth checking first, because it is the frame everything else on the call hung from. Wireless service, handset sales and every enterprise and wireline dollar those carriers collect already sit inside their reported top lines; there is no hidden pool of subscriber revenue parked outside them. Shotwell may have been reaching for a broader definition of American connectivity spending, or she may have rounded in the direction of ambition. Neither reading changes what her company said it will do — but it changes how much of that ambition anyone can put in a model.
The arithmetic behind $352 billion is public and unglamorous:
- AT&T (NYSE: T): full-year 2025 revenues of $125.6 billion, per its Jan. 28, 2026 earnings release
- Verizon (NYSE: VZ): total operating revenue of $138.2 billion for 2025, per its Jan. 30, 2026 news release
- T-Mobile (NASDAQ: TMUS): total revenues of $88.3 billion for 2025, per its FY2025 Form 10-K
What was actually said
The mechanism is spectrum. "The spectrum that we purchased from EchoStar does have terrestrial components, so we definitely intend to build out terrestrial," Shotwell said, per the earnings-call transcript published by Investing.com. She put the holding at 65 megahertz and called it a "massive increase in capability" when paired with a next-generation satellite fleet.
The distinction between the two businesses matters more than it sounds. Starlink's existing direct-to-cell service, sold to American consumers through T-Mobile under the T-Satellite brand, uses satellites as substitute cell towers to fill gaps where no ground network reaches — a coverage product, not a capacity product. A terrestrial licence is the opposite: it permits transmitters on the ground in dense places, which is where the revenue is and where satellites are least useful. What Shotwell described is SpaceX crossing from the first business into the second.
The build, as sketched, would not resemble a carrier network. Shotwell described cellular base stations bolted onto the same hardware that already holds a Starlink broadband dish — small, self-installed radios rather than leased macro towers. "Small stations essentially... they're really just Starlink dishes that also provide connectivity in the mobile spectrum bands, and have them be all over the place," she said, adding that the approach would let SpaceX "deploy that as you need it... without spending $10 [billion] or many billions of dollars." On SpaceX's own timeline, the supporting satellites fly in 2027 and service begins at the end of that year.
Why the market flinched
The reaction was immediate, and it was not confined to the obvious targets. All three incumbents fell in after-hours trading on Tuesday — and so did the company issuing the threat.
- Verizon (NYSE: VZ): −3.6%, to $45.19
- AT&T (NYSE: T): −2.7%, to $22.75
- T-Mobile (NASDAQ: TMUS): −2.4%, to $172.90
- SpaceX: down roughly 11%
(After-hours moves on Aug. 4, per MarketScreener and Investing.com.)
SpaceX's own decline had nothing to do with telecom. Revenue came in at $7.8 billion, up 92% from a year earlier, and the net loss narrowed to $541 million — both ahead of what analysts had modelled, per Teslarati's and Sci-Tech Today's summaries of the release. What unsettled the tape was capital spending of $18.4 billion in a single quarter.
Run that against the top line and it works out to roughly $2.36 of capital expenditure for every dollar of revenue recognised in the quarter. That ratio appears nowhere in the release, and it sits awkwardly next to a pitch built on how little the new network will cost.
Capital intensity is the axis on which this story gets settled, not rhetoric. Carriers spend heavily and predictably to keep their networks standing; a challenger becomes credible only if it can add coverage without inheriting that cost structure. That is precisely the claim SpaceX made on Tuesday, and one nobody can verify until the hardware exists.
How big Starlink actually is today
Measured against the incumbents, the base is small. Connectivity — the Starlink line that would house any mobile offering — generated $4.3 billion in the quarter, up 66% year over year, per Teslarati's account of the segment disclosure. Annualised, that is roughly $17 billion, under 5% of what the three U.S. carriers reported in revenue last year.
Subscribers say the same thing from the other direction. Starlink closed the quarter with 12 million paid subscribers worldwide after adding 1.7 million in three months, its best quarter of net additions to date. T-Mobile alone ended 2025 with 142.4 million U.S. customers, per its fourth-quarter release.
A gap that size is not by itself an argument about who wins. Insurgents in wireless have never needed to match incumbent subscriber counts to damage incumbent economics; they need only make the marginal customer harder to keep and the marginal price increase harder to push through. That is the channel through which a credible entrant first shows up in carrier results — churn and pricing power early, subscriber share much later. It is also why three large-cap share prices moved on a description of a network that does not yet exist.
The spectrum has been down this road
There is an uncomfortable precedent attached to these exact airwaves. When T-Mobile's acquisition of Sprint closed in April 2020, Dish Network took on an obligation to stand up a fourth facilities-based national network. It built far enough to claim coverage of roughly 80% of the U.S. population — and still could not hold subscribers.
The exit came through the spectrum itself. EchoStar (NASDAQ: SATS) — the Dish TV and Boost Mobile parent that regulators had designated the industry's fourth carrier — agreed to sell more than $40 billion of licences to AT&T and SpaceX. SpaceX's portion runs to about $19.6 billion across two agreements covering AWS-4, H-block and unpaired AWS-3.
EchoStar's own filing on the first tranche describes it as roughly $17 billion, split between cash and SpaceX stock, plus about $2 billion of debt-interest payments funded by the buyer. The FCC released EchoStar from the network buildout requirement.
The lesson buried in that sequence is not that a terrestrial network cannot be built. It is that the licence is the cheap part. Dish had the spectrum, a regulatory mandate, a deadline and a retail brand, and what defeated it was everything that comes after the airwaves — site acquisition, backhaul, handset certification, roaming economics and retail distribution. SpaceX proposes to skip most of that by hanging radios off equipment its own customers install. Whether that substitutes for a national footprint is the whole question, and Tuesday's call did not answer it.
What we wrote six days ago
LineVest covered the incumbent side of this problem on July 30, when Bain Capital and Tillman Global Holdings — a private infrastructure developer — committed $1.5 billion to Eaton Fiber, the wholesale builder that serves as Verizon's fibre partner. The structure existed for one reason: Verizon wanted the footprint without carrying construction capital on its own balance sheet. Five days later, a competitor told the market that the construction capital need not exist at all.
Put the two capital plans side by side and the contrast sharpens. T-Mobile has guided to roughly $10.0 billion of capital spending for 2026 to operate the network serving those 142.4 million customers, per TelecomLead's summary of its second-quarter release. SpaceX spent $18.4 billion in the June quarter alone (an annualised rate of roughly $73.6 billion), while telling investors its mobile build will not require billions. Both statements can be true; they describe very different companies making very different promises about cost.
There is also a relationship problem embedded in the announcement. SpaceX's satellite-to-phone service reaches American consumers today through T-Mobile, which markets it as T-Satellite; that partner is now also the named target. Nothing said on the call terminates the arrangement, and neither side has suggested it will. But a wholesale supplier that publicly announces it wants the same subscribers changes the negotiating table at renewal, whatever the contract currently says.
What would make this reading wrong
The straightforward reading of Tuesday night is that owning spectrum and flying a satellite fleet are not the same thing as running a terrestrial mobile network, and that the after-hours move repriced an intention rather than a build. If SpaceX ships a working dish-mounted base station and turns on commercial service, that reading fails — because the point of the small-cell design is that it never requires a tower lease, and the capital-intensity objection disappears with it. The counter-case does not rest on SpaceX being late; it rests on whether self-installed radios deliver coverage where customers actually stand.
What to watch
Two checkpoints carry dates. The first is T-Mobile's third-quarter report, where a satellite-driven change in customer behaviour would surface before it surfaces anywhere else, and the line to read is churn: postpaid account churn ran at 0.99% in the second quarter of 2026, per TelecomLead. A move there without a promotional explanation would be the earliest hard evidence either way.
The second is regulatory, and the third is hardware — whether a commercial base station ships on the 2027 timetable SpaceX cited. Together, those checkpoints will settle what Tuesday's after-hours tape could only reprice.
