SpaceX now owns about 65 MHz of nationwide US spectrum bought from EchoStar for roughly $19.6 billion, and the FCC approved the transfer on May 12, 2026 with permission to use it from space, on the ground, or both. On its first earnings call as a public company (August 4, 2026), SpaceX said it intends to build terrestrial components and take customers from AT&T, Verizon and T-Mobile, with upgraded service targeted for the end of 2027.
A Starlink-only phone plan is now plausible. But the three incumbents have refused to host it as an MVNO and have pooled their own spectrum in a joint venture, so the first version will be a rural-first, outdoor-first product rather than a drop-in replacement for your current carrier.
What SpaceX actually bought from EchoStar
Two deals, announced two months apart, give SpaceX its first exclusive, nationwide, contiguous spectrum for mobile service.
The first, announced September 8, 2025, covers EchoStar’s full portfolio of AWS-4 and H-Block licenses for approximately $17 billion: up to $8.5 billion in cash and up to $8.5 billion in SpaceX stock, plus roughly $2 billion of cash interest that SpaceX funds on EchoStar debt through November 2027 (EchoStar press release). Boost Mobile subscribers also get access to SpaceX’s next-generation Starlink Direct to Cell service through EchoStar’s 5G core.
The second, announced November 6, 2025, adds EchoStar’s unpaired AWS-3 licenses (1695 to 1710 MHz, 15 MHz, part of 3GPP Band 70n) for approximately $2.6 billion in SpaceX stock. Combined: about $19.6 billion before the interest subsidy.
Trade coverage of the FCC approval summarized the holdings as 40 MHz of AWS-4, 10 MHz of H-Block and 15 MHz of unpaired AWS-3, about 65 MHz in total (Telecoms.com, Mobile World Live). For context, today’s direct-to-cell service with T-Mobile runs on roughly 5 MHz borrowed from the carrier, while the three national carriers together hold about 1,020 MHz (New Street Research, cited by IEEE ComSoc).
Why this spectrum matters more than its size suggests
Spectrum is the raw material of any wireless network. More megahertz means more simultaneous users and higher speeds per cell, and exclusive use means no one else transmits in the same band in the same place. Until this deal, every Starlink Mobile connection depended on a carrier lending part of its own band. That works for emergency texting. It is not a foundation for a retail business, because the partner can withdraw the spectrum or cap the service.
SpaceX described the purchase as 50 MHz of US S-band plus global mobile satellite service licenses (Light Reading, September 8, 2025). Gwynne Shotwell, SpaceX’s president and COO, said it enables next-generation satellites with a step change in performance (EchoStar press release).
When the FCC approved the deal, and with what conditions
The FCC’s Wireless Telecommunications Bureau and Space Bureau approved both EchoStar sales (65 MHz to SpaceX, 50 MHz to AT&T) on May 12, 2026, with a combined value above $40 billion (FCC press release via Benton; SDxCentral).
Three conditions matter:
- Architecture flexibility. The FCC granted waivers allowing SpaceX to deploy the spectrum in terrestrial, space-based or hybrid architectures (Mobile World Live, Telecoms.com). This is the legal basis for the ground-station plan discussed below.
- Buildout obligation. SpaceX must use the spectrum intensively and provide connectivity to the public, direct-to-device, terrestrial or both, within a set number of years (SDxCentral). The exact deadline was not spelled out in coverage we could verify, so we do not quote a year.
- Escrow on the seller. EchoStar must fund an escrow account of about $2.4 billion for possible claims tied to its past license obligations, a condition it called unprecedented and involuntary (Advanced Television).
FCC Chairman Brendan Carr framed the approvals as part of a plan to release approximately 300 MHz of low- and mid-band spectrum by the end of 2027 through secondary-market deals and auctions (SDxCentral).
What Shotwell and Musk have actually said about a Starlink phone plan
Separate the reporting from the on-the-record statements.
Reported, not confirmed (June 2026). During the June IPO roadshow, the Financial Times reported (via Reuters) that Shotwell told investors SpaceX was weighing a direct-to-consumer Starlink mobile offering in the US and possibly its own terrestrial network. Reuters could not verify the remarks and SpaceX did not comment (IEEE ComSoc, June 27, 2026). A credible leak, not a plan.
On the record (August 4, 2026). SpaceX’s first earnings call as a public company put the ambition in its own words:
- Shotwell said the company “definitely intend[s] to build out terrestrial components”, noting the EchoStar licenses carry terrestrial rights (Broadband Breakfast; Wireless Estimator).
- Shotwell said SpaceX expects to acquire “quite a few” customers from AT&T, T-Mobile and Verizon because its service will be better (Wireless Estimator).
- Musk said SpaceX could deploy “a large number of small stations”, essentially Starlink dishes that also provide mobile-band connectivity, since large base stations are expensive and hard to site (Broadband Breakfast; Wireless Estimator).
- Shotwell called these “little femtocells” housed with a customer’s Starlink dish, deployed incrementally so SpaceX avoids spending “10 or many billions of dollars” upfront (Wireless Estimator).
- Next-generation Starlink Mobile satellites launch in 2027, with upgraded service targeted for the end of that year (Advanced Television; Wireless Estimator).
Earlier context. At MWC in March 2026, SpaceX renamed the service Starlink Mobile and said about 650 direct-to-cell satellites were active, with 10 million monthly active users (Urgent Communications). What that service does for your phone today is covered in can you make phone calls with Starlink.
Why AT&T, Verizon and T-Mobile formed a direct-to-device joint venture
On May 14, 2026, two days after the FCC approval, the three national carriers announced an agreement in principle to form a joint venture for satellite direct-to-device service (AT&T press release). Stated goals: pool limited spectrum, cut dead zones, provide backup connectivity during disasters, and set common technical specifications with OS providers, app developers and device makers. The release names no satellite operator and says existing satellite deals stay in place.
On October 1, 2026, the three signed the definitive agreement and named Paul Roth, a former AT&T, Cingular Wireless and Ameritech executive, as interim CEO. A board with representatives from each founder manages the entity; ownership percentages were not disclosed (AT&T press release, October 1, 2026).
The unspoken logic: a united front
The press releases talk about dead zones. Analysts read it differently. Tim Farrar of TMF Associates said the joint venture was clearly designed to present a united front in negotiations over Starlink’s next-generation system. Walter Piecyk and Joe Galone of LightShed Partners said the carriers appeared nervous, noted the timing weeks before the SpaceX roadshow, and doubted SpaceX would get access to the joint venture’s spectrum (Light Reading, May 14, 2026).
The mechanism behind that fear: a satellite operator that borrows carrier spectrum is a supplier. One with its own nationwide spectrum, 12 million broadband subscribers (as of June 30, 2026) and a consumer brand is a competitor that could sell under its own name and leave the carrier as a wholesale “dumb pipe”. Pooling spectrum into a neutral platform makes satellite providers “more interchangeable”, in the words of Roger Entner of Recon Analytics (Light Reading), and weakens the leverage any single one, including SpaceX, would hold.
Why none of the big three will give SpaceX an MVNO deal
An MVNO (mobile virtual network operator) buys wholesale access to a carrier’s network and resells it under its own brand. For SpaceX, that would be the fastest route to a Starlink plan that works indoors and in cities, with satellites covering the gaps. All three carriers have said no, in public.
T-Mobile. CEO Srini Gopalan, on the April 28, 2026 Q1 earnings call, said MVNOs make sense for T-Mobile only when they expand the total addressable market, and it was “not obvious” how a SpaceX or other LEO operator would pass that test (Fierce Network). T-Mobile remains SpaceX’s primary US direct-to-cell partner, so this is a boundary, not a breakup.
AT&T. On the Q2 2026 earnings call, CEO John Stankey said he was “not prone to look for a wholesale agreement” to solve a problem he does not have. AT&T says its network handles about 98% of the traffic its converged customers need; satellite is for the remaining 2%, via a partnership with AST SpaceMobile expected in 2027 (Light Reading, July 23, 2026).
Verizon. On its Q2 2026 call, CEO Dan Schulman said: “There is no backdoor to our MVNOs in any structure”, and that Verizon sees no reason to extend an MVNO to any satellite provider. He prefers to deal with satellite operators through the three-carrier joint venture (SDxCentral, July 30, 2026).
Bloomberg reported that Starlink had held talks with Charter about offloading mobile traffic to Charter’s network (via IEEE ComSoc). Charter is itself a Verizon MVNO, which is the route Schulman’s “no backdoor” line closes.
The four ways SpaceX could reach your phone
Satellite-only service on own 65 MHz
- What it requires
- Next-gen satellites (launches planned 2027), new handset band support
- Who controls the gate
- SpaceX, FCC, chipmakers
- Status (October 2026)
- Confirmed plan; upgraded service targeted end of 2027
Own terrestrial network of small cells on Starlink dishes
- What it requires
- Host rooftops, power, permits, backhaul, core network
- Who controls the gate
- SpaceX, local permitting
- Status (October 2026)
- Stated intent on August 4, 2026 call; no hardware details yet
MVNO on AT&T, Verizon or T-Mobile
- What it requires
- A willing host carrier
- Who controls the gate
- The three carriers
- Status (October 2026)
- All three publicly declined (April to July 2026)
Acquire a carrier
- What it requires
- Capital, antitrust and FCC approval
- Who controls the gate
- Regulators, shareholders
- Status (October 2026)
- Musk said buying a company like Verizon is "not out of the question" (2025 podcast, via Light Reading); no deal announced
What a Starlink-only phone plan would realistically look like
Work through the physics, the handset supply chain and the published performance data.
Coverage: outdoors first, indoors later
Today’s direct-to-cell service is an outdoor, sky-view product (see how direct-to-cell works). Ookla’s RootMetrics drive tests of T-Mobile’s Starlink-powered service in rural New York in the second half of 2025 found that 143 of 238 text attempts succeeded (60%), with an average round-trip time of 1 minute 17 seconds. Typical signal strength was about -108 to -126 dBm, versus -80 to -120 dBm on terrestrial networks (Mobile World Live summarizing Ookla, April 22, 2026). That is weak by cellular standards, and walls, roofs and vehicles cut it further.
That is why the August 2026 terrestrial plan matters. A femtocell on a rooftop next to a Starlink dish is a conventional ground radio on the AWS-4, H-Block and AWS-3 bands. It propagates like any 2 GHz cell site and can reach indoors across a neighborhood in a way a satellite hundreds of kilometers up cannot. Shotwell’s line, “we will eliminate dead zones” (Advanced Television), is a rural promise. Indoor coverage in a dense suburb depends on host rooftop density per square mile, a number SpaceX has not published.
Capacity: the carriers’ physics argument
Verizon’s Schulman made the capacity case on the Q2 2026 call: a satellite beam covers hundreds of square miles, so at 100 Mb/s down and 20 Mb/s up it can serve only 10 to 40 homes per square mile, versus 500 to 2,000 for Verizon’s fixed wireless. “Not because of execution, but because of physics”, he said (SDxCentral). AT&T’s Stankey added that each of the big three has roughly 300 MHz per cell site, more than triple what SpaceX’s constellation offers, and an 80 MHz spot beam about 20 miles across is not comparable to a cell site with a 2 to 2.5 mile radius (SDxCentral).
SpaceX’s answer is scale from orbit plus ground infill. Shotwell has called the new constellation on the EchoStar spectrum “100 times better” than today (IEEE ComSoc summary). SpaceX has filed with the FCC for up to 15,000 direct-to-cell satellites, and an April 30, 2026 FCC rule change allows up to eight Starlink satellites per area and frequency cell (Light Reading, May 19, 2026). Even so, SpaceX cautioned in September 2025 that the service would likely top out around LTE Advanced speeds of roughly 100 Mbps (Notebookcheck summarizing Bloomberg).
Both sides are right about different geographies. Satellites plus rooftop cells can beat a thin rural macro network. They cannot match a dense urban grid with 300 MHz per site.
Handsets: the chipset bottleneck
This is the constraint most coverage skips. A phone’s modem only transmits on bands its radio front end was built for. Today’s T-Satellite works on “most smartphones from the last four years” (T-Mobile) precisely because it uses T-Mobile’s own spectrum that US phones already support. The EchoStar bands are different: AWS-4 and H-Block sit around 2 GHz, and unpaired AWS-3 is Band 70n uplink. Few US phones ship with full support for them because no US network has used them at scale.
SpaceX knows this. Shotwell told Bloomberg in September 2025 that SpaceX was developing a custom Starlink modem for integration by chipmakers of the Qualcomm, Apple or MediaTek type, with first test phones in 2026 (Notebookcheck). At MWC 2026, Starlink showed a MediaTek test device receiving alerts on band n66 from existing satellites, and MediaTek confirmed the collaboration (Urgent Communications). A Starlink-only plan needs phones with the right bands on shelves before it is worth buying, which points to 2027 at the earliest for a meaningful installed base, in line with SpaceX’s own end-of-2027 target.
Pricing: no data, but one benchmark
SpaceX has not priced a retail plan. The only market benchmark is T-Mobile’s T-Satellite: $10 per month per line as an add-on (a limited-time price that saves $5), included on Experience Beyond plans, supporting texting, location sharing, WhatsApp voice chat and select satellite-optimized apps (T-Mobile). Ookla noted that US direct-to-device connections declined after T-Mobile began charging around $10 when free trials ended, and that only 0.46% of US Speedtest users had a D2D connection in March 2026 (Mobile World Live). The current product is a niche add-on, and consumers are price sensitive even at $10. A Starlink-only primary plan would have to clear a far higher bar.
What to expect in 2027 and beyond
A sober sequence built only from dated, sourced commitments:
Late 2026. T-Mobile remains the US distribution channel for Starlink Mobile. AT&T and Verizon continue with AST SpaceMobile (Verizon committed $100 million in prepayments and convertible notes in May 2024 on 850 MHz), and Verizon also uses Skylo for messaging. Apple runs its own path through Globalstar, backed by a $1.5 billion commitment announced in November 2024 that includes $400 million for a 20% stake and rights to 85% of Globalstar’s capacity (Verdict). Google’s Pixel Satellite SOS runs on Skylo. The carrier joint venture hires a permanent CEO.
2027. SpaceX launches next-generation Starlink Mobile satellites and targets upgraded service by year end. AT&T expects direct-to-device service with AST SpaceMobile in 2027 (Light Reading). The 3GPP 5G non-terrestrial standard is expected to reach consumer-grade operation (Light Reading, May 19, 2026). First phones with the EchoStar bands arrive if the chipmaker work stays on schedule. Rooftop femtocells will go where Starlink has dish density, which today skews rural.
2028 and beyond. This is where a Starlink-only plan either becomes a real primary line for rural households and outdoor workers, or stays a complement. The deciding variables are invisible from outside SpaceX: rooftop cell density, handset penetration of the new bands, and whether any carrier changes its mind on wholesale. Musk’s remark that acquiring a company like Verizon is “not out of the question” is the wild card, and nothing has been filed.
Sources
- EchoStar Announces Spectrum Sale and Commercial Agreement with SpaceX (EchoStar IR, September 8, 2025)
- EchoStar Agrees to Sell Full Unpaired AWS-3 Spectrum License Portfolio to SpaceX (EchoStar IR, November 6, 2025)
- SpaceX snares some EchoStar spectrum for $17B (Light Reading, September 8, 2025)
- FCC Secures Win for America’s Leadership in Next-Gen Connectivity (FCC release via Benton Institute, May 12, 2026)
- FCC approves $40B EchoStar spectrum sales to SpaceX and AT&T (SDxCentral, May 13, 2026)
- FCC signs off on EchoStar deals to SpaceX, AT&T (Mobile World Live, May 13, 2026)
- FCC gives its blessing to EchoStar spectrum deals (Telecoms.com, May 13, 2026)
- FCC approves EchoStar’s spectrum sale to SpaceX/AT&T (Advanced Television, May 13, 2026)
- AT&T, T-Mobile and Verizon to Form Joint Venture (AT&T newsroom, May 14, 2026)
- AT&T, T-Mobile and Verizon form D2D JV for rural coverage (Light Reading, May 14, 2026)
- AT&T, T-Mobile and Verizon Launch Joint Venture, Name Paul Roth Interim CEO (AT&T newsroom, October 1, 2026)
- Starlink’s path to mobile is a checklist (Light Reading, Roger Entner, May 19, 2026)
- T-Mobile CEO reaffirms hard no on SpaceX MVNO (Fierce Network, April 28, 2026)
- AT&T CEO: A satellite MVNO pact would solve a problem I don’t have (Light Reading, July 23, 2026)
- Verizon CEO plies physics to baffle satellite broadband, declares no MVNO backdoor (SDxCentral, July 30, 2026)
- FT: SpaceX considering Starlink direct-to-consumer mobile service (IEEE ComSoc Technology Blog, June 27, 2026)
- SpaceX Planning Terrestrial Deployments to Support Direct-to-Device (Broadband Breakfast, August 2026)
- SpaceX targets US cellular entry (Advanced Television, August 5, 2026)
- SpaceX wants to be a fourth major network, but it’s building around towers, not on them (Wireless Estimator, August 6, 2026)
- At MWC, SpaceX execs tout Starlink V2, key carrier partner for it (Urgent Communications, March 8, 2026)
- Starlink 5G in every phone as SpaceX develops mobile satellite chip (Notebookcheck summarizing Bloomberg, September 17, 2025)
- Ookla finds D2D scale is elusive despite increased traction (Mobile World Live, April 22, 2026)
- T-Satellite with Starlink (T-Mobile)
- AST SpaceMobile and Verizon Announce Plans to Target 100 Percent Geographical Coverage of the Continental United States (Business Wire, May 29, 2024)
- Skylo Connectivity Enables New Satellite SOS Feature on Google Pixel 9 Series (Business Wire, August 13, 2024)
- Apple to invest $1.5bn in Globalstar (Verdict, November 4, 2024)
All sources accessed October 19, 2026.
