SpaceX Is Turning Satellite Coverage into Carrier Leverage

Who actually wins when a carrier rents its network from space?

In August 2022, T-Mobile announced that it would let SpaceX use a slice of its cellular spectrum to beam coverage down from Starlink satellites. The pitch was narrow: fill the dead zones. Three years later, the service works. T-Satellite now reaches more than 500,000 square miles of the United States that no carrier’s towers cover, and it went from beta to a commercial product faster than most people in the industry expected.

That speed is the problem: it let SpaceX accumulate the capability to go it alone faster than the exclusivity structure was built to contain. The exclusivity, spectrum access, and revenue T-Mobile gave SpaceX to build this service are the same assets that could equip SpaceX to stop being T-Mobile’s supplier and become its competitor. T-Mobile’s exclusivity window is reportedly set to expire this summer. SpaceX completed the largest IPO in history on June 12, 2026. And in May, the three carriers that spend billions a year fighting each other did something they had never done: they agreed to build a joint satellite platform together. Taken together, the sequence reads less like coincidence and more like three parties independently responding to the same underlying shift.

SpaceX does not need to replace T-Mobile, AT&T, or Verizon to win here. It needs sufficient credible ability to go it alone that it can reset the exclusivity terms, the revenue split, and eventually the customer relationship itself in its favor the next time a contract comes up for renewal. That is a lower bar than disruption, and it is the bar the evidence in this piece actually supports.

To test how far this has gone, Santiago & Company built two datasets from public records that nobody else has published in this form: a count of how often satellite-to-phone technology now appears in mandatory SEC disclosures, and a full read of every FCC filing in the three dockets that govern SpaceX’s applications. Both point the same direction, and the picture they draw is more specific than anything in the trade press.

‍

Before getting into what SpaceX has done, it’s worth settling two questions: how much is actually at stake, and whether the underlying shift is real or just a good story.

On the first question, the honest answer is that nobody agrees. MarketsandMarkets sizes the direct-to-device satellite market at $2.64 billion by 2030. Omdia, using a narrower smartphone-specific definition, gets to $11.99 billion in the same year. SpaceX’s own IPO filing states a total addressable market of $740 billion for “Starlink mobile services” specifically, inside a broader $28.5 trillion claim that also includes $22.7 trillion for AI enterprise applications. These numbers describe different scopes and different time horizons, so they should not be read as competing estimates of the same thing. But the spread is itself informative: a market this early in its life does not have an agreed price tag, and any answer that sounds too precise is worth distrusting.

‍

On the second question, the record is not ambiguous. Santiago & Company queried the SEC’s full-text search system for four terms tied to this technology (direct-to-cell, direct-to-device, supplemental coverage from space, non-terrestrial network) across every 10-K, 10-Q, 8-K, and S-1 filed since 2022, not just the companies at the center of this story. The count of filings mentioning at least one of those terms went from 7 in all of 2022 to 76 in 2025. The first half of 2026 alone already equals the full-year 2024 total. This is a rising presence in the formal disclosure record across satellite operators, chipmakers, and carriers industry-wide. One company’s announcement cycle does not explain a pattern spread across that many unrelated filers, though a mention count like this shows the topic entering routine disclosure and strategy narratives more than it proves any single filer faced a new legal obligation to report it.

Everything SpaceX bought since September points the same direction

Start with what SpaceX has actually done. Statements about its intentions can wait. In September 2025, EchoStar agreed to sell SpaceX its AWS-4 and H-block spectrum licenses for roughly $17 billion, split between cash, stock, and debt-service payments. Two months later, SpaceX paid EchoStar another $2.6 billion in stock for EchoStar’s entire portfolio of unpaired AWS-3 uplink spectrum. In October, SpaceX filed for two trademarks with the US Patent and Trademark Office: “Starlink Mobile” and “Powered by Starlink,” covering cellular personal communication services. The filing remains pending. It received a non-final office action and was granted an extension as recently as May 28, 2026, which means SpaceX is still working through ordinary trademark prosecution rather than holding a registered mark. That detail matters, because most coverage of the filing treated it as a settled fact rather than an application still under review.

None of this required speculation to document. It required reading the S-1 SpaceX filed on May 20, 2026, for its IPO. Buried in the market-sizing section, SpaceX states its own estimate that “Starlink mobile services” represent a $740 billion opportunity, distinct from the $870 billion it separately claims for Starlink’s existing broadband business. A company preparing to sell shares has every incentive to describe the largest defensible market it can, so the number should be read as an aspiration for investors rather than a forecast of what will happen. It carries one kind of weight that rumor never does: it is filed under securities law, with the liability that comes with it.

Six weeks after the S-1, on June 26, SpaceX President Gwynne Shotwell reportedly told investors at the company’s IPO roadshow that SpaceX is weighing a retail mobile service under the Starlink brand and, potentially, its own terrestrial network in the United States. No publicly available transcript of the roadshow exists. The Financial Times broke the story, and roughly ten other outlets have since repeated the same account. None of the follow-on coverage appears to add independent sourcing of its own, including Reuters, which reported the story without confirming it. That is wide repetition of a single scoop, not corroboration, and the distinction matters: treat this as what SpaceX’s president reportedly told investors, sourced to one outlet, not as an independently confirmed fact. SpaceX has not publicly confirmed any product, price, or launch date.

What ties these pieces together is capability, not intent. Spectrum, a pending brand, and a disclosed target market are the specific ingredients a company needs to stop leasing its network to a carrier and start selling to that carrier’s customers directly. SpaceX did not have to say it was planning to become independent. The spectrum purchases and the S-1 filing are concrete transactions with legal consequences; investor comments are not, and the two should not be weighted equally. But the transactions alone are enough to make the capability argument without relying on what anyone said about intent.

Eight parties have already told the FCC they don’t believe this is just a supply relationship.

If the moves above were only visible to analysts parsing SEC filings, they might be over-read. They are not analyst-only. Santiago & Company analysis from the three FCC dockets that actually govern this fight, using the Commission’s public records system: the docket covering SpaceX and T-Mobile’s specific coverage application, the broader rulemaking that set the industry framework for satellite-to-phone service, and the docket reviewing the EchoStar-to-SpaceX spectrum transfer. Together, those three dockets hold 489 filings.

Twenty-one of those filings were formally styled as an opposition or a petition, the FCC’s own procedural categories for a party actively contesting an application. Reading the metadata alone would have overstated the finding: two of the twenty-one turned out, once we opened and read them, to be misclassified. One was Globalstar objecting to a different company in an unrelated dispute that happened to share a docket number. Another was EchoStar, defending its own spectrum sale against other parties’ objections, arguing on SpaceX’s side of that docket. After correcting for both, the verified count is thirteen filings from eight distinct parties: Omnispace, the National Radio Astronomy Observatory, DISH Network, Verizon, EchoStar, AT&T, Frequency Forward, and DQE Communications. One of the eight, the National Radio Astronomy Observatory, is a research facility rather than a company, and its stake in the outcome (protecting radio astronomy from satellite interference) is scientific, not commercial.

‍

Eleven of the thirteen filings raise the kind of technical interference argument you would expect from a competitor protecting its own spectrum. The other two are a different species entirely. Frequency Forward’s petition against the EchoStar transfer rests on an unproven allegation about foreign control of SpaceX’s majority shareholder. DQE Communications’ petition is a contract dispute over vendor obligations tied to EchoStar’s restructuring. Both are genuine, on-the-record filings, and neither contains any evidence that the market itself sees SpaceX as an existential threat. Folding either into the same bucket as AT&T’s or Verizon’s technical objections would overstate what the filings actually show.

The most useful finding in this set is a reversal, buried inside the count. EchoStar formally petitioned to deny SpaceX’s coverage application in September 2024, arguing SpaceX’s proposed emissions would interfere with EchoStar’s own network. By December 2025, EchoStar was filing in a different docket to defend SpaceX’s spectrum purchase against other companies’ objections to the deal. The same company went from opponent to seller to advocate in just 14 months, and the filings that document each stage are all part of the public record.

The industry’s own defensive move followed a similar logic, on a longer fuse. On May 14, 2026, AT&T, T-Mobile, and Verizon announced an agreement in principle to build a joint venture that pools spectrum and satellite capacity, explicitly open to multiple satellite providers rather than tied to any single one. That announcement predates Shotwell’s retail-mobile disclosure by six weeks, so it cannot be read as a direct reaction to it. What it can be read as is a response to the pattern already visible by May: the spectrum purchases, the trademark filings, and a satellite supplier whose ambitions were expanding faster than any single bilateral deal could contain. One month earlier, Amazon had reached its own $11.6 billion agreement to acquire Globalstar, giving Amazon Leo the spectrum and technology to build a third vertically integrated satellite platform alongside SpaceX and AST SpaceMobile. Three different companies, watching the same signal, chose to reduce their dependence on any single supplier rather than wait to find out whether that supplier’s ambitions were real.

The case against overreacting is also real.

None of the above settles the outcome, and a fair reading of the evidence has to say so plainly. The clearest limit on SpaceX’s independence is spectrum. According to New Street Research analysis cited in the Financial Times, AT&T, T-Mobile, and Verizon together hold roughly 1,020 megahertz of licensed spectrum. SpaceX, even after its EchoStar purchases, holds about 65. New Street’s David Barden has called the idea of SpaceX building a competitive standalone network in a market this saturated “incredibly hard,” and he is not wrong on the numbers. A wireless network is built on licensed spectrum before it is built on satellites or software, and on that measure the gap SpaceX would need to close is still an order of magnitude. One more transaction the size of the EchoStar deal would not close it.

‍

The second limit is that multi-sourcing, the hedge carriers are visibly reaching for, is not available on demand. AST SpaceMobile is the most credible alternative to Starlink for direct-to-device coverage, and it is materially behind its own schedule. The company had roughly nine satellites in orbit in mid-2026 against a stated year-end target of forty-five, and one of its most recent launches, BlueBird 7, suffered a thrust anomaly in April that left it in the wrong orbit. Starlink, by comparison, already operates around 650 direct-to-cell satellites. A carrier that wanted to reduce its dependence on SpaceX today would be choosing a partner that is still years from matching Starlink’s current scale.

‍

Put the two limits together and the honest conclusion is narrower than “SpaceX is about to disrupt the wireless industry.” Satellites cannot yet match dense urban terrestrial networks in capacity, and they won't for years, which is exactly why SpaceX would need a retail play to matter beyond the rural coverage niche it already occupies. The technology gap and the spectrum gap are both real constraints, and either one alone would be reason for caution about a fast takeover. What neither constraint does is undo the opening argument. Winning outright and shifting the balance of a negotiation are different achievements with different thresholds, and SpaceX has cleared the second even where it is nowhere close to the first.

What the 2026 renewal should actually protect

T-Mobile’s exclusivity negotiation with SpaceX, whenever it concludes, is not a pricing exercise, and treating it as one would be the most expensive mistake available to either side. The asset SpaceX is building compounds in value the longer a single-source arrangement continues: every quarter T-Mobile renews on the current terms is another quarter of spectrum, brand-building, and disclosed ambition SpaceX gets to accumulate before it has to compete for the customer relationship directly. The terms worth fighting for in this renewal are structural, not just financial: rights over incremental spectrum SpaceX acquires during the contract term, a most-favored-nation clause tied to whatever terms SpaceX offers any future retail product, or a standing option to convert the commercial relationship into something closer to a joint venture if SpaceX does launch under its own brand.

For AT&T and Verizon, the real lesson in the FCC filings is about single-sourcing itself, independent of whether AST SpaceMobile ultimately succeeds: betting the coverage strategy on one satellite partner, as T-Mobile did in 2022, is now a clearly recognized risk rather than an unexamined convenience. The May 2026 joint venture is the industry’s own admission of that, arriving from three companies that do not agree on much else. Sitting outside that kind of platform, rather than inside it, is itself a strategic choice with consequences, whether or not it is framed as one.

For infrastructure investors, the order-of-magnitude disagreement in market-sizing estimates is a signal about where the real due diligence question sits: who ends up owning the customer relationship once the technology stops being the constraint. A comparison of antenna size against constellation count changes every time a launch succeeds or fails, which makes it a weak basis for a long-term bet either way. The more durable test is the one this piece has tried to apply throughout: does a given claim rest on a document you can open, or on a headline you have to trust? Spectrum auctions, contract renewals, and FCC dockets will answer the ownership question over the next eighteen months, in increments too small to generate a headline on their own. Watch those, not the announcement.