Starlink is not a dish

In short

Starlink began as a dish on the roof. Since January 2024 SpaceX has also flown a separate class of about 650 satellites built to talk to an ordinary 4G phone. They transmit on a mobile operator's own frequencies, so to the phone each one looks like a cell tower in orbit.

Ukraine's Kyivstar switched the service on for every 4G phone on its network on 24 November 2025: no app, no settings, texting only, outdoors. Three million had registered by January and nearly five million by March 2026. T-Mobile in the United States has sold the same service since July 2025. Ten operators on five continents have signed for it, each under its own brand.

On 5 November 2026 the Deutsche Telekom CEO, who chairs T-Mobile US, said Musk needs at least 200 billion euro to become a real American operator: 76 megahertz of spectrum against T-Mobile's 400, plus 20 to 30 billion dollars for towers. He is right about cities and indoors. The post's point: Musk already does mobile where the towers stop, under other operators' brands, and the 200 billion is the price of the part he lacks.

Position: the dish was the product; the satellite that pretends to be a tower is the business. Next read: end of 2027, SpaceX's own date for its US service on its own frequencies.

Core claim

Text as posted. Corrections are in the log at the bottom, not in this text.

Starlink is not a dish. Five million Ukrainians use it with no dish, on the 4G phone they already had.

Three things most people believe about satellite phones, against the record.

One. Starlink means the white dish on the roof. No, SpaceX has also flown lots of satellites to talk to a normal phone.

Two. A normal phone cannot talk to a satellite. Kyivstar switched it on for every 4G phone on its network to speak with starkink. No app, no settings (but texting only), works outdoors. Nearly five million users. T-Mobile in the United States has sold the same (when phone shows "T-Mobile SAT" - this is Starlink).

Three. Musk cannot do mobile because he has no towers. That actually triggered me to write this... On 5 November the Deutsche Telekom CEO, who chairs the T-Mobile (THAT RESELLS MUSK'S SATTELLITES communication), said Musk needs at least 200 billion euro to become a real American operator. He is right about cities and indoors but he forgot the contract with Musk he has. Musk already does mobile everywhere outside buildings, under T-mobile brand! If I would be product manager in Starlink I would give for free femto/nano/microcell (tiny "tower" that makes your home connected to 4/5G) for each starlink dish and the deal is done (ok regulators would have something to say here, of course).

Now the part the telco people will recognise. A virtual operator, in our trade, is a brand that sells a network it does not own. Today T-Mobile is the brand and the network in the sky is Musk's.

I met many people that say "space tech is not important, too far, not practical", I have typicla answer "the space is closer than you drive to your grandma"... it's really close, we live on a tiny "strata" between the rock and nothing

LinkedIn post

Claims and grades

CONFIRMED: two independent sources, both linked. SINGLE-SOURCE: one. PRIMARY DOCUMENT: quoted from the document of record, here a company filing or release. COMPANY COUNT: a company's own figure for its own customers, carried by two outlets but with one origin. DERIVED: author's calculation or inference, inputs in Method. POSITION: the author's view, not a fact claim.

Key numbers: the satellite-to-phone layer

Key numbers: what the Deutsche Telekom CEO said

Method

Counting users. Kyivstar's January figure is "registered users"; its March figure is "users ... using the technology" and is not further defined. The post says "users" and the page says what each count is. Messages delivered are the only usage figure published.

Counting satellites. SpaceX's S-1 counts its dedicated satellite-to-mobile constellation at about 650 on 31 March 2026. CelesTrak's catalogue, read through azmth, lists 639 direct-to-cell satellites launched between 3 January 2024 and 13 June 2025, with none catalogued as launched since. The two are within two percent; the post says "lots", the panel says "about 650".

Where the service works. The satellites transmit on the operator's licensed terrestrial frequencies, so the phone needs no change. The phone attaches to a satellite only when no terrestrial signal is available (hands-on report, row 6) and only with a clear view of the sky (VEON). This is why the page reads "where the towers stop", not "everywhere".

The towers claim. Höttges's figure is his estimate of what it costs to match today's US carriers on the ground; it is not a SpaceX budget. The page keeps his three numbers (200 billion, 76 against 400 MHz, 20 to 30 billion for towers) and his two limits (indoors, bandwidth) as stated.

Limits

• As posted, "everywhere outside buildings" is wider than the record (texting only, outdoors, dead zones). "Forgot" is the author's framing; on the same call Höttges called satellite complementary.
• Everything live today is texting. Voice and data on these satellites are plans with 2026 and 2027 dates, not services.
• The Kyivstar figures are the company's own, carried by VEON and Interfax; there is no independent count and the delivery ratio is a wide band.
• "T-Mobile SAT" rests on one hands-on report from May 2025; T-Mobile's own page was not opened.
• The ten operators are signed partners; which of them had live service on the posting day is not separated here.
• Höttges's call was read through press and transcript carriers; Deutsche Telekom's own transcript was not fetched. The euro-dollar rendering of his figure differs by outlet.
• The femtocell line is the author's position. SpaceX's president described a similar architecture in August 2026, which is one report of one call.
• The virtual-operator analogy is the author's framing from telecom work; it is not a statement about the legal form of the T-Mobile and SpaceX contract, whose terms are not public.
• Nothing on this page is about the European market; the EU band, IRIS² and the four-operator letter are a separate page if that post ships.

Falsifier

The post states a position and no dated call. Stated on this page, what would make the position wrong.

If by 31 December 2028 Starlink Mobile sells no mobile service of its own in the United States on the spectrum SpaceX has bought, the reading that the 200 billion is a price being paid, not a wall, fails.

If T-Mobile ends its satellite service on Starlink before SpaceX's own service launches, the line "under other people's brands while they still can" had the direction wrong.

Next read: end of 2027, SpaceX's own date for Starlink Mobile service in the United States.

Corrections log

(empty; corrections will be dated and appended here, never edited in place)

Related

• Evidence index

Sources

Read date for all items: 6 November 2026.

• SpaceX: SpaceX Form S-1/A, 2026 (archive.org snapshot); Reuters via MacRumors, 5 Aug 2026.
• Catalogue: azmth Starlink tracker, CelesTrak and McDowell counts.
• Kyivstar: VEON, 15 Jan 2026; Interfax-Ukraine, 16 Mar 2026.
• T-Mobile: Telco Magazine; Converge Digest; STL Partners; TechCrunch, 26 Nov 2024; Lon.TV hands-on, 28 May 2025.
• Operators: TechAfrica News, 8 Oct 2025; starlink.com/mobile.
• Deutsche Telekom, 5 Nov 2026: Benzinga via TradingView, 6 Nov 2026; TelecomTV, 6 Nov 2026; Broadband Breakfast, 6 Nov 2026; MLQ, 6 Nov 2026; TeslaNorth.
• Ownership: Deutsche Telekom, 25 Feb 2026; Reuters via PYMNTS, 22 Apr 2026.