What happens next

One of six models is live

The six models below are ordered by how real each one is, which differs from how much each is discussed. The ordering is our judgement. For the licensed rows the status comes from FCC orders and the FCC licence file; ET Docket 26-169 supplies the proposed rulemaking, and company statements as reported supply the rest (Exhibit 1).

The MVNO row is the one most often assumed and least supported. Each of the three incumbents has said in public, separately, that it is not offering one to a satellite operator. Those statements come from trade reporting rather than transcripts, and they are three separate refusals rather than one industry position; the effect is the same.

An MVNO is the cheap way into mobile, and none of the three carriers is offering one. The expensive way is the one now on the table.

The remarks attributed to named executives below were made at investor conferences and on earnings calls during 2026 and are taken from Light Reading, Fierce Network, RCR Wireless and telecoms.com. No transcripts have been obtained, so each remark is marked as reported rather than verified.

Reported Executive remarks: SpaceX Q2 update, 5 Aug 2026, as reported by RCR Wireless; T-Mobile’s CFO, Citi Global TMT Conference, September 2026, as reported by Light Reading; AT&T’s CEO, Goldman Sachs Communacopia + Technology Conference, September 2026, as reported by telecoms.com; Verizon’s CEO, Q2 2026 earnings call, 24 July 2026, as reported by Fierce Network. No transcripts obtained; paraphrased where marked.

What each scenario requires

In Scenario A, SpaceX becomes a facilities-based carrier. It has the regulatory permission and, from late 2027, the spectrum. It still needs cell sites and backhaul, handset support for its bands, retail distribution, billing and care, and indoor coverage, the hardest of them. The small-cell idea exists because of that last item. At the SpaceX Q2 update on 5 August 2026 its executives talked about putting “a cellular base station on the gear that holds a Starlink broadband dish” and deploying “little femtocells around the country”, with service “end of next year”; the words are reported by RCR Wireless rather than taken from a transcript. Two months later SpaceX contracted for a second answer to the same item. On 8 October 2026 it agreed to acquire Grain Management’s nationwide 800 MHz portfolio, up to 14 MHz of paired spectrum, and said it will deploy “an advanced terrestrial deployment” on it as a coverage layer that “penetrates through obstacles, such as walls”, with the 2 GHz spectrum carrying capacity and most existing phones already supporting the band (SpaceX update). The agreement covers the low-band position the incumbents said it lacked, with a network of ordinary cell sites to put it on: the expensive route rather than the clever one. The cost of that network is not on any citable record; TD Cowen’s analyst put a national build at upwards of $80 billion in a note of 8 October, as Fierce Network reported, and that figure is neither adopted nor adjusted here. Part 4 has the price of the spectrum, which is a report rather than a filing, and the block’s history.

The small-cell idea is an architectural argument: a subscriber’s existing dish becomes a site, and distributed capex stands in for a low-band spectrum position. It is also the claim the incumbents attack hardest. T-Mobile’s CFO, speaking at the Citi Global TMT Conference in September 2026, said the approach would need on the order of a billion femtocells to match existing outdoor coverage, and that a count like that brings interference and backhaul problems with it. Light Reading reported the remark as a paraphrase rather than a quotation.

In Scenario B, SpaceX stays supplemental. The scenario is coherent and cheaper. The economics from part 2 still hold: the carrier lends spectrum, gains coverage it could not economically build, and bills its own subscriber. It is also the only one of the two that Exhibit 1 records as live and licensed. The scenario’s weakness is its ceiling: the business is capped at a wholesale margin on someone else’s customer, metered through the 5 to 6 MHz service channel that part 3 sets out, and it would leave $19.6 billion of nationwide spectrum running a supplement, an odd use of the asset (Exhibit 2).

Three hinges decide between the scenarios

Three things decide the outcome, and none is settled by an opinion about anyone’s intentions: whether the service works indoors, whether the spectrum position is deep enough to carry a national network, and what the capital and the regulatory clock require.

• Indoor coverage. Every incumbent argument comes back to this. AT&T’s CEO, speaking at the Goldman Sachs Communacopia + Technology Conference in September 2026, framed satellite as solving the last 2% of coverage rather than offering a new way to serve the other 98% (telecoms.com, as reported). Verizon’s CEO, on the Q2 2026 earnings call of 24 July 2026, called terrestrial networks orders of magnitude more efficient than satellite and put the addressable market at single-digit millions of rural homes (Fierce Network, as reported). Both are paraphrases of spoken remarks. SpaceX’s reply came on 8 October in the form of spectrum: low band, with a terrestrial deployment to put it on.
• Spectrum depth. 65 MHz nationwide is a real position, on the order of what a single large carrier deploys in one mid-band layer, and the 800 MHz agreement adds a 7 MHz pair beneath it, a coverage layer rather than a capacity one. Whether 79 MHz is enough depends on the location: the same 65 MHz behaves very differently in a rural county and in Manhattan. Neither has been modelled here, and the question belongs in a capacity model.
• Capital and the clock. Part 6 showed the money arriving from outside the industry, which settles the capital half and leaves the clock. The FCC’s news release of 12 May 2026 describes performance obligations requiring “meaningful, reliable connectivity to the public — whether D2D, terrestrial, or both — within years” and publishes no number. The order itself sets the numbers at paragraph 57. From 30 November 2027 SpaceX has two years to show, in each licence area, a downlink signal-to-noise ratio of 5 dB and service 70% of the time over 90% of the area, four years for 80%, and nine years for 10 dB and 90%, with uplink throughput and spectral-efficiency floors alongside; a licence that misses the final test terminates automatically. The 800 MHz order of 1 July 2026 writes the same three dates and metrics into that band’s satellite pathway. The first test date is 30 November 2029.

The propagation difference behind the indoor argument is real, and it ran the incumbents’ way for as long as everything SpaceX owned lay at 1.7 to 2.2 GHz, which penetrates walls less well than the 700 MHz AST leases (part 3). Link budgets are not modelled here, so the claim is not adjudicated; the frequencies are public and the physical argument is consistent with them. What changed on 8 October is the frequencies. SpaceX is still buying width where walls are a problem, and it has now also contracted for a narrow block where they are not, the kind of band AST rents, and has said that block goes to work on the ground first (Exhibit 3). If SpaceX solves indoor coverage, it will be with terrestrial radios on low band, as the incumbents did.

The three national carriers answered with a joint venture

On 14 May 2026, two days after the FCC order, AT&T, T-Mobile and Verizon announced a joint venture for satellite direct-to-device (joint release). The release described pooled spectrum, joint investment and common technical specifications, with each company’s existing satellite agreements left in place, and called the venture an agreement in principle. On 1 October 2026 the three said they had entered into a joint venture agreement, named Paul Roth as interim chief executive while a search runs for a permanent one, and put a board of founding-member representatives over it (joint release). The second release repeats the May commitments and adds a promise to work with rural operators. It is still a company announcement rather than a regulatory filing (Exhibit 4). A week later SpaceX agreed to buy the low band.

Taken at face value, the release describes three carriers pooling the one asset class that part 4 identified as what an entrant cannot assemble quickly. Whether it changes the market depends on terms that have not been published and on a regulatory review that, on the public record, has not begun. No reading of motive is offered beyond the releases. One thing the second release does not say: between the May announcement and the October signature, the low-band block T-Mobile sold to Grain went to the satellite operator the venture was formed in answer to.

Each scenario would show in public filings before any announcement

None of what follows needs inside information. The indicators are ours, but each is a public filing, registration, certification or disclosure that would have to appear before a strategy could be announced (Exhibit 5).

Neither list is a forecast or a scorecard. Each sets out only what would have to become visible in public before either path could be executed.

Two dates bound every negotiation in this market

The EchoStar to SpaceX second closing is expected on or about 30 November 2027, and AST’s 1,197 leases expire on 9 May 2028. Every commercial negotiation in this market falls between those two dates. October added two more. Satellite applications for the 800 MHz band are due by 4 December 2026 under the order that approved Grain’s purchase, and SpaceX’s first performance test on its EchoStar spectrum, the 5 dB and 70% test above, falls on 30 November 2029, with the 800 MHz band’s first test written for the same day. The other dated events in the series lead up to them (Exhibit 6).

The calendar is the one thing to keep from these seven parts. Everything commercial in this market must be negotiated between the second closing and the lease expiry, the first performance test comes two years after that, and the filings that would reveal either path are public, so an announcement is not needed to know which way it is going. The first of those filings arrived on 8 October 2026.

Terms used

• SCS, supplemental coverage from space. The FCC framework under which a satellite operator serves ordinary phones on spectrum a terrestrial carrier is licensed to use. The live T-Mobile and Starlink service runs under it.
• Facilities-based carrier. An operator that owns its spectrum and network and bills its own subscribers, as distinct from a wholesaler or an MVNO.
• MVNO. A mobile virtual network operator: a brand that sells service on another operator’s network under a wholesale agreement.
• Femtocell. A small, low-power cellular base station for a home or office, connected to the operator through the premises’ own broadband.
• Special temporary authority. Short-term FCC permission to operate a radio before or outside a full licence, and one of the filings a terrestrial build would generate.
• Antenna structure registration. An entry in the FCC’s register of antenna structures. A new registration is a public sign that a site is being built.
• 800 MHz, band 26. The 817 to 824 and 862 to 869 MHz pair SpaceX agreed to buy from Grain Management on 8 October 2026: low band, 7 MHz each way, already supported by most phones. Part 4 has its history and its price.
• Performance test. The metrics the FCC wrote into the EchoStar and Grain orders in place of a terrestrial build-out: downlink signal-to-noise ratio, availability over each licence area, uplink throughput per MHz and spectral efficiency per beam, tested at two, four and nine years.