SpaceX picks Qualcomm chips for Starlink Mobile's US 5G network

The deal: a multi-year agreement that covers tens of thousands of base station sites. A base station is the equipment on a tower or rooftop that talks to your phone.

The timing: rollout is planned to begin in 2027.

The money: nobody has shared a price or a total value.

Qualcomm's own words were short and upbeat. Durga Malladi, an executive vice president at Qualcomm, said: "Together with SpaceX, we are building world-leading cellular infrastructure in the United States, advancing 5G today." He added that the two companies have a chance to "lay the foundation" for 6G in the coming years.

Two Qualcomm platforms are named in the deal. Each does a different job at a cell site.

• QRU100: a chip platform for the radio unit, the box that sends and receives radio signals at the tower.
• QDU100: a chip platform for the processing behind it, the part that handles fast, time-sensitive work on those signals.

Starlink Mobile will use radio units with 64 transmit and 64 receive paths, usually written 64T64R. This design is called Massive MIMO. Think of it as a radio with many small antennas working together, so it can serve many phones at once and point its signal at each one.

Qualcomm says the two platforms together are built to get more performance from each watt of power. That matters when you have tens of thousands of sites, because many towers have strict power limits and every watt adds up.

Starlink Mobile started as a way to connect phones to satellites. In the US it works with a partner carrier, T-Mobile, to offer limited coverage where there are no towers.

SpaceX now wants more. It has said it wants to combine satellites and ground equipment to become a "major" US carrier, competing with T-Mobile, AT&T and Verizon.

Satellites are great for remote places. They are harder to use for crowded cities, where thousands of people want fast data in the same spot. A tower close to the phone can carry much more traffic than a satellite far above it.

In real life Picture a stadium on game night. A satellite is like one very tall radio mast covering the whole county. Towers and small cells are like having a signal in every section of the stands.

Radio signals travel on slices of the radio spectrum, and carriers pay billions for them. SpaceX has been buying.

• EchoStar spectrum: about 65 megahertz, bought for more than $10 billion, which SpaceX says gives Starlink a "massive increase in capability."
• Low-band spectrum: up to 14 megahertz of paired spectrum in the 800 megahertz band. SpaceX describes it as a new coverage layer that helps signals get through walls. It still needs final FCC approval.
• Satellites: the FCC has approved 15,000 second-generation Starlink Mobile satellites, which SpaceX says will offer 100 times the bandwidth of the current generation.

The catch: the big three carriers usually hold hundreds of megahertz. Analysts have said 65 megahertz may not be enough to serve a large number of customers with fast speeds.

Before the Qualcomm deal, SpaceX executives described a different, lighter plan. They talked about small stations that are really Starlink dishes, the white panels on many roofs, that also send mobile signals.

Elon Musk said the company feels "reasonably confident" it can deploy a large number of such small stations. Analysts have pushed back on that idea, saying it is not yet clear how it would work at scale.

Tens of thousands of Qualcomm-powered base stations sound like a more traditional approach. SpaceX has not explained how the two ideas fit together, or how many sites will be big towers and how many will be small cells.

What's next: SpaceX has not said which cities come first or when the first phone customers could connect to the new network.

Dragonwing is Qualcomm's brand for its cellular infrastructure chips. Reports on the deal say the platform has been deployed in places like Asia, but has been less used in the US.

That makes this a notable win. A nationwide US build with tens of thousands of sites is a showcase customer for a company best known for phone chips.

One detail is still unclear. The QDU100 handles low-level processing, and Qualcomm appears to supply the software for that layer. The software that sits above it seems to come from somewhere else, and the companies have not said who.

Qualcomm has been busy elsewhere this fall. We covered its patent deal with Huawei a few weeks ago.

Not very fast. Building ground equipment takes land, permits, power and crews, which is a big reason most US mobile service still runs through the big three carriers.

That is why some observers have said SpaceX could go one of two ways. It could sign a deal with the major carriers and act as a virtual operator, one that rents their networks, in the short term. Or it could wait much longer for a true Starlink Mobile launch.

The Qualcomm order points to the second path, at least in part. You do not order chips for tens of thousands of sites if you plan to rent everyone else's towers forever.

SpaceX is not starting from nothing. Starlink has more than 10,000 satellites in orbit, and its home internet dishes are already on the roofs of many houses and businesses.

Starlink also already offers a direct-to-phone text service that works through satellites. Together, those pieces show that satellites can handle real everyday uses.

What SpaceX has not shown yet is a full mobile network that can carry calls, video and apps for millions of people. That is the gap the new towers are meant to close.

Maybe. Reports have suggested SpaceX was exploring a purchase of Boost Mobile, a US carrier that already has customers and a network. Neither company has confirmed this.

If it happened, it would give Starlink Mobile subscribers on day one, rather than starting from zero in 2027. For now, treat it as a rumor.

• Not yet: you cannot switch to a Starlink Mobile tower network today. The first sites are planned for 2027.
• More choice later: if SpaceX succeeds, the US could get a new major carrier with satellite backup built in.
• Dead zones: the plan is meant to fix places where towers do not reach, such as highways, farms and parks.
• Watch the prices: nothing has been said about plans or prices, so there is nothing to compare yet.