Abel Avellan, 5 August, on X after BlueBirds 11-13 reached orbit: “Successful launch… again,” “full Nominal Operations,” “~20,000 sq ft of combined aperture hardware in Space,” “Beta service around the corner.” Garry Tan, same day: “Direct-to-phone broadband from orbit, no ground towers required.” The 10 August 8-K is colder. Thirteen spacecraft. Peak rates on Block 2 “approaching 200 Mbps.” Block 1 already demonstrated “nearly 100 Mbps.” Sixty-plus MNO partners, three billion subscribers on paper. Backlog $1.30 billion. FY26 guide $150-200 million. Beta in 2026. Continuous coverage still wants about 45 satellites, and that date slipped to early 2027 in the 15 July filing. Starlink Direct to Cell, the product T-Mobile markets as T-Satellite, has been commercially live for text and a short list of apps since July 2025, with more than 650 dedicated D2C satellites, in 22 countries. Those are not the same product. X cashtags them as the same war.
AST BlueBirds in orbit after 5 Aug (EX-99.1, 10 Aug) versus Starlink dedicated D2C sats (TechTimes, 5 Aug). AST continuous coverage ≈ 45 sats, slipped to early 2027 in the 15 July filing.
Array versus flock
Disclosed / tracked · as of 22 Aug 2026
Two D2D products, mid-August 2026
Array versus flock
AST flies a few enormous arrays, about 2,400 square feet per Block 2 bird, 20,000 square feet combined after 13 sats, and speaks ordinary cellular to an unmodified phone on AT&T, Verizon, Vodafone, Orange, Telefónica, Deutsche Telekom, Rakuten, Bell, stc. Starlink flies many small D2C birds, antennas on the order of 65 square feet, and speaks a narrowband service to T-Mobile and other partners. TechTimes, 5 August: T-Satellite is text, location, and data to about 34 apps; voice and full broadband are “in development,” with a late-2027 language still attached to the bigger Starlink mobile story. AST’s beta will have hour-scale gaps until the 45-sat plane exists. Starlink’s D2C already works as a gap-filler because 650 small birds are overhead more often for a thin message.
Avellan’s sales pitch, on the 10 August call and in Light Reading’s write-up of it, is that AST is “in partnership with, not in competition with” MNOs. SpaceX’s inaugural earnings week, same window, said the opposite about itself: terrestrial small cells, EchoStar spectrum, a path that competes with US carriers. That is why MNOs still take AST meetings. It is also why the 45-sat slip matters. A partner who is late while the competitor is already texting 400 million people in 22 countries is a partner who gets a smaller SIM menu.
What the 10 August print actually is
The EX-99.1 is a manufacturing and backlog letter, not a usage letter. BlueBirds 14-16 ready to ship. 17-46 in the plant. Six spacecraft in 50 days (BB 8-13). 3,000 of about 5,600 US digital cells lit. Japan notified as operational with Rakuten on 4 August, the first market AST told the FCC it had started. J-LEO with Rakuten, up to about $1 billion of non-dilutive Japanese government capital, still “preliminary selection.” US government awards over $125 million. Government.” Light Reading had Q2 revenue at $31.5 million and a $231 million net loss, capex $610 million. Those are the numbers of a company that is still the customer.
The FCC, 14 August: a 30-day STA, at most 100 commercial handsets, 817-824 / 862-869 MHz, non-commercial, coordinate with AT&T, Verizon, FirstNet. April’s licence: 248 sats, 124 by August 2030, 248 by August 2033. That is the regulator treating AST as a test article. Grain’s $2.9 billion T-Mobile 800 MHz purchase, with AST in support, is spectrum chess, not ARPU.
New Glenn was supposed to take eight BlueBirds at a time. A pad explosion in 2026 took that vehicle out of the near-term book. Falcon 9 takes three. The 15 July 8-K that pushed 45 sats into early 2027 and talked about raising about $1 billion, convert, vertical integration, “secure additional access to orbit,” is the launch-constraint sentence in SEC English. X posted the 5 August splash. The 8-K posted the slip.
Lynk, Amazon, the rest of the plane
Lynk and Omnispace completed a merger as Elveo Mobile on 17 August, with an SES collaboration the same day. That is a third architecture: a messaging-first MSS story trying to become a company with a new name. Amazon Leo, after agreeing to buy Globalstar for about $11 billion, filed on 24 July for a Leo D2D system of up to 5,105 satellites, deploy starting 2028, using Globalstar 1.6 / 2.4 GHz, the MSS band the FCC had already kept away from SpaceX. Apple’s continuity on that spectrum is part of the deal narrative. So the plane is: Starlink text now, AST broadband beta later this year if the STA and the MNOs hold, Amazon-Globalstar D2D in 2028, Elveo as a rebuilt messaging name. Four products. One cashtag war.
Forty-five operational BlueBirds by 30 June 2027 is the continuous-coverage date AST wrote for itself. Commercial, not beta, not government, D2D service revenue as a named line in a 2026 10-Q or 10-K is the usage tell. A world in which Starlink V2 mobile broadband is live for T-Mobile customers at useful speeds in 2027, while AST is still in beta, ends the only-broadband-D2D sentence. Thirteen sats and a Guinness array are not that print. Neither is a Stocktwits $100 target. The 5 August X posts are true: the birds are up, the arrays are large.
Cell tower on a green hill.
Launch-slot tags on an aluminum bus.