12 November 2025
What Starlink’s latest purchase of AWS-3 spectrum tells us about its D2D plans
EchoStar has agreed to hand SpaceX a valuable Direct-to-Device (D2D) ingredient: nationwide unpaired AWS-3 uplink spectrum (1695–1710MHz range) in an all-stock deal of around $2.6bn1. Combined with the AWS-4 and PCS H-block downlink spectrum bought in September, this extra 15MHz potentially allows Starlink to use 3GPP Band n70, which is supported on existing phones, boosting its position in the D2D market.
D2D and where Starlink currently sits
Direct-to-Device (D2D), or Direct-to-Cell (D2C), aims to connect ordinary smartphones to satellites over licensed cellular spectrum. In the US, this is permitted through the FCC’s Supplemental Coverage from Space framework2, which allows satellite operators to use mobile carriers’ frequencies. SpaceX and T-Mobile have FCC approval to use the PCS G-block (1910–1915 / 1990–1995MHz) for Starlink D2C. The service launched commercially in the US in July 2025 initially as a text-only service and has since expanded to support selected apps in terrestrial coverage “dead zones”. Starlink markets it as working with “every LTE phone wherever you can see the sky”3, an eye-catching claim that also highlights the next challenge: moving to support a greater load of mobile traffic – which requires more spectrum.
To expand the spectrum available for D2D, Starlink has recently made two nationwide spectrum acquisitions in the United States, both from EchoStar4:
Given these two back-to-back deals, it is hard not to wonder what Starlink is trying to build with all this mid-band spectrum. Our view is that the picture splits into two scenarios: an alignment with the global mobile satellite S-band based on the first deal (which is currently not supported in existing mobile phones), and alignment with 3GPP Band n70 which is already supported in today’s phones through the latest acquisition of AWS-3 spectrum.
Speculated Scenarios
Scenario 1 – Using AWS-4 to align with the Global Mobile Satellite S-band
When the September transaction was announced, our assumption was that Starlink would try to align the new US holdings with the global mobile satellite S-band, which is currently licensed in Europe to Viasat and EchoStar until 2027. AWS-4 comprises 2000–2020MHz uplink and 2180–2200MHz downlink, while PCS-H fills the small gap around 1995–2000MHz. This would allow Starlink to assemble something close to the global S-band pairing (1980–2010 / 2170–2200MHz). However, whilst this band has been identified as a band for 5G mobile services (3GPP 5G NR Band n256), it is not currently supported in mobile handsets, and it will take 2–3 years before this is the case – and then also take more time for the handsets to be adopted by mobile users.
Scenario 2 – Adding on AWS-3 to align with 3GPP Band n70
The November purchase of the nationwide AWS-3 block at 1695–1710MHz adds another ingredient. With this additional spectrum, Starlink could choose to reframe the same assets around 3GPP 5G NR Band n70. The newly acquired AWS-3 spectrum provides the uplink, whilst the previously acquired PCS-H and lower part of AWS-4 would provide the main downlink. Band n70 is already supported in several handsets and can be incorporated into a wider range of handsets than is the case for Band n256. The remaining AWS-4 spectrum at 2180–2200MHz currently has no handset support, but if it were to receive it in the future, it could then be used to provide supplemental downlink capacity that can be added as demand builds.
Considering both transactions, for a total spectrum acquisition of 65MHz for a total spend of $19.6bn, we arrive at a price of ~USD0.89 per MHz per pop across the two deals, a value comparable to what 3.5–3.8GHz 5G awards in the US have typically cleared around. Acquiring the AWS-3 block allows Starlink to own a band that handset vendors are already building into devices, whilst also keeping the option of Scenario 1 alive if it ever wishes to use it. Furthermore, if D2D fails to generate the forecast returns, the spectrum still has intrinsic underlying value for use on terrestrial mobile networks.
What’s next?
In the near term, Starlink’s own roadmap is pretty clear. D2D is already available in multiple countries, and is extending beyond basic texting, with early support for apps/data, richer messaging (photos, short videos, voice notes) and IoT connectivity, and longer-term plans for fuller voice services. The service has been launched commercially with T-Mobile in the US and One NZ in New Zealand, with additional launches or late-stage preparations under way with operators including Kyivstar (Ukraine), Rogers (Canada), Telstra (Australia), Salt (Switzerland), Entel (Chile/Peru) and Virgin Media O2 (UK), with the latter targeting a 2026 go-live5. As its spectrum platform expands in response, we will soon see how Starlink plans to utilise its new spectrum holdings.
Sources:
[1] EchoStar, ‘EchoStar Agrees to Sell Full Unpaired AWS-3 Spectrum License Portfolio to SpaceX’, 6 November 2025
[2] Federal Communications Commission, ‘Single Network Future: Supplemental Coverage From Space; Space Innovation’, 30 April 2024
[3] Starlink, ‘Business | Direct to Cell’, accessed 12 November 2025
[4] EchoStar, ‘EchoStar Announces Spectrum Sale and Commercial Agreement with SpaceX’, 8 September 2025
[5] Virgin Media O2, ‘Virgin Media O2 signs landmark partnership with Starlink to boost rural UK mobile coverage’, 30 October 2025
