RC / LOADING
DIRECT-TO-DEVICE · U.S. WIRELESS
T-Mobile helped SpaceX turn satellite-to-phone connectivity into a commercial product. SpaceX's spectrum purchases and broader mobile ambitions are now changing the balance of that relationship.
On August 25, 2022, Elon Musk and T-Mobile CEO Mike Sievert stood on a stage at Starbase and announced "Coverage Above and Beyond," a plan to end cellular dead zones by connecting ordinary phones to Starlink satellites broadcasting on T-Mobile spectrum. The pitch was simple: if you can see the sky, you have coverage.
Four years later, T-Mobile still sells that product. It is called T-Satellite, it costs $10 a month, and it works.
SpaceX, meanwhile, now talks about the carrier business differently. On August 4, 2026, on its first earnings call as a public company, President and COO Gwynne Shotwell noted that AT&T, Verizon and T-Mobile generate roughly $600 billion a year between them, and said she anticipates SpaceX "being able to acquire quite a few of their customers" because, in her words, "our service will be better."
This is not a breakup; both companies still make money together. What has changed is the character of the relationship. A straightforward partnership has become coopetition: each side still needs the other today, while working to depend on it less tomorrow.
01 · THE ORIGINAL DEAL
In 2022, SpaceX had almost everything required to put a cell tower in orbit: satellites, launch capacity, phased-array antennas, a global network. What it lacked was the one ingredient it could not build itself: licensed terrestrial cellular spectrum that ordinary phones already listen to. A smartphone only tunes to the bands it shipped with, and every usable band in the U.S. was licensed to someone else.
T-Mobile supplied that ingredient: a nationwide slice of its 1.9 GHz PCS mid-band spectrum, plus the customer relationships, devices, and network core that turn a radio link into a phone service. The architecture was, and still is:
ORDINARY PHONE → STARLINK SATELLITE (broadcasting as a T-Mobile cell) → STARLINK NETWORK → T-MOBILE CORE
Under the FCC's "supplemental coverage from space" framework, the satellite operator borrows a terrestrial carrier's license. T-Mobile's spectrum was load-bearing: without it, there was no legal way for a Starlink satellite to talk to an unmodified American phone.
02 · THE RECORD
The record matters, because this is not a story about a failed alliance. The first Direct-to-Cell satellites launched in January 2024 and were texting within days; the FCC authorized commercial service that November. T-Mobile ran a public beta from January 2025 and launched commercial service, branded T-Satellite with Starlink, on July 23, 2025, pointedly open to AT&T and Verizon customers via eSIM. App data over satellite followed in October.
As of this week, T-Satellite is live on T-Mobile's product pages, running on more than 650 Direct-to-Cell satellites and covering the continental U.S., Puerto Rico, Hawaii, parts of southern Alaska, Canada, New Zealand and Japan. It is the largest satellite-to-phone service in operation. That success frames what came next.
03 · THE TURN
On September 8, 2025, SpaceX agreed to buy roughly 50 MHz of paired spectrum from EchoStar, the AWS-4 (2 GHz) and PCS H-block licenses, for about $17 billion split evenly between cash and SpaceX stock, plus roughly $2 billion of EchoStar's debt interest through late 2027. On November 6 the companies amended the deal to add 15 MHz of nationwide unpaired AWS-3 uplink (Band 70n) for about $2.6 billion in stock. The total came to roughly $19.6 billion for 65 MHz of nationwide spectrum, in licenses that carry both satellite and terrestrial rights. The FCC approved the transfer in May 2026, deferring a few of SpaceX's more ambitious technical requests.
The important distinction is that SpaceX no longer has to imagine a future in which it owns spectrum. It now owns a lot of it. Before these deals, any Starlink phone service anywhere required a carrier willing to contribute its licenses. After them, SpaceX can design a network around frequencies it controls, in orbit and, if it chooses, on the ground. The company says its next-generation Direct-to-Cell satellites are designed around the new bands, with roughly twenty times the throughput of a first-generation satellite. The transaction also came with a commercial agreement putting Boost Mobile subscribers onto that next-generation service, a second U.S. retail channel that does not run through T-Mobile.
FIG. 1 · Four years from joint announcement to open competitive intent. The relationship did not end at any point on this line; its terms changed. RC SPACE INTELLIGENCE
04 · THE DISCLOSURE
The strategy became explicit on the August 4 call. Shotwell told analysts SpaceX would not stop at satellites: "We definitely intend to build out the terrestrial component… you will have not only the capacity from the satellites themselves, but you will have a build-out of the hardware and systems necessary to make a true mobile service." The next-generation Direct-to-Cell satellites are slated to fly in 2027, with upgraded service, which Shotwell said would be "100 times better" than today's, targeted for the end of that year. Then came the line about acquiring "quite a few" of the big three's customers. Musk suggested the network could eventually offer more bandwidth than traditional cellular networks, a claim that will take years to test. Shares of all three carriers fell in the trading that followed, Bloomberg reported.
A company that owns spectrum, satellites, launch, backhaul, terrestrial radios and the customer billing relationship is no longer a satellite infrastructure supplier to telecom operators. It is a vertically integrated telecom operator. That is a different company from the one Mike Sievert shared a stage with in 2022.
05 · THE PHYSICS
One misconception is worth addressing directly: this is not a plan to serve Manhattan from orbit. A satellite beam covers an enormous footprint, and everyone inside it shares a limited slice of spectrum. Dense cities generate more traffic per square kilometer than any orbital cell can carry, which is why terrestrial networks use hundreds of thousands of small, tightly packed cells.
SpaceX's own pitch acknowledges this. The plan described on the earnings call is a hybrid network: small terrestrial cells where demand is dense, and satellites everywhere coverage runs out. SpaceX has suggested mounting those cells on the same rooftop hardware that already hosts millions of Starlink dishes, with the dish providing backhaul. In a city, your phone would talk to a local SpaceX radio; on a highway in Wyoming, to a satellite. None of that terrestrial layer is deployed today, and analysts are skeptical of the timeline. The major carriers have signaled they will not lease SpaceX wholesale network access, and Craig Moffett of MoffettNathanson called a competitive direct-to-consumer offering within five years "extraordinarily challenging" to imagine. Bloomberg has reported that SpaceX discussed carrying some phone traffic over Charter's terrestrial network, a sign the company is looking for ground capacity wherever it can find it.
FIG. 2 · In the operating architecture, T-Mobile owns the spectrum and the customer. In the architecture SpaceX described on August 4, it owns every layer. The lower path is stated intent, not a deployed network. RC SPACE INTELLIGENCE
06 · THE BALANCE
In a stronger position than the competitive framing suggests, at least for now. T-Mobile gets dead-zone coverage no rival can match today, disaster resilience it markets heavily, and a two-year head start over the AST SpaceMobile-based services AT&T and Verizon are bringing up. SpaceX gets things it cannot quickly replicate: a proven 1.9 GHz band that phones already support, tens of millions of subscribers at zero acquisition cost, immediate revenue, and a partner that handles billing, devices and decades of regulatory experience. Neither side has an obvious incentive to unwind the arrangement.
T-Mobile remains a valuable partner. The uncomfortable part is that SpaceX is simultaneously building the option not to need it. T-Mobile has noticed. Srini Gopalan, Sievert's successor as CEO, told the Financial Times this month that the Starlink threat is "exaggerated," asking "what is the gap in the market that they're looking to serve?" and what a new entrant's differentiation would be. The argument has force, since T-Satellite already fills coverage gaps for T-Mobile customers. That coverage is delivered over SpaceX's own satellites.
07 · THE SYMMETRY
This is where the relationship becomes genuinely two-sided. On May 14, 2026, AT&T, T-Mobile and Verizon announced an agreement in principle to form a joint venture that would pool spectrum and build a unified platform for satellite direct-to-device service. Its status matters: it is an agreement in principle rather than an operating company, and definitive contracts and regulatory clearances are still pending. The carriers were explicit that existing satellite deals, including T-Mobile's with SpaceX, continue untouched, and that the platform is meant to host multiple satellite providers.
The strategic logic is straightforward. If the carriers pool their spectrum and standardize access, no single satellite operator can become an indispensable gatekeeper. In practice that means SpaceX, the only provider with hundreds of Direct-to-Cell satellites already overhead. The hedging now runs in both directions. SpaceX bought spectrum so it no longer depends on T-Mobile's licenses; T-Mobile is organizing with its rivals so it does not depend on SpaceX's constellation. Each is spending real money to reduce its exposure to the other.
FIG. 3 · The solid boxes show the working relationship; the dashed boxes show each company's insurance against it. Both hedges were announced within a year of each other. RC SPACE INTELLIGENCE
08 · THREE SCENARIOS
These are analytical scenarios, not forecasts with probabilities attached.
SpaceX concludes the terrestrial economics do not close and sells satellite capacity through carriers: T-Mobile, Boost, the joint venture, international operators. The 2022 arrangement scales up rather than out.
Arguably the most plausible middle path. SpaceX keeps supplying T-Mobile while selling through Boost, signing foreign carriers, and gradually standing up its own retail mobile offering. The companies cooperate in some layers and compete in others, indefinitely.
SpaceX combines spectrum, satellites, small cells, backhaul and billing into a full competitor. The barriers are formidable: only 65 MHz of spectrum, no terrestrial footprint, handset support for its bands, capital, customer service at scale, and regulators. Possible, but not imminent.
None of these paths is inevitable, and the third faces the steepest barriers. What has changed is narrower but real: a year ago, SpaceX lacked direct ownership of the spectrum such a strategy requires. The EchoStar transaction removed that constraint. The terrestrial, regulatory and commercial barriers remain substantial.
CONCLUSION
T-Mobile helped SpaceX prove that a satellite could function as part of an ordinary cellular network. SpaceX now controls spectrum, and is planning infrastructure, that could let it own more of that network itself. None of this ends the partnership: the product works, the money flows, and each company remains useful to the other.
What has changed is the logic underneath. SpaceX is assembling more of the pieces it would need to compete for the same customer, and T-Mobile is organizing with its rivals to make sure no satellite provider becomes indispensable. The partnership remains intact. Its strategic logic is simply less comfortable than it was on that stage in 2022.