Weekly Wrap: Starlink–Amazon Leo clash highlights shortcomings of satellite filing rules

Back in 2020, Amazon got approval for its Gen‑1 satellite constellation, a vast undertaking that would see 3,232 satellites launched into low‑Earth orbit (LEONGSO, or non-geostationary satellite orbit…) to provide high‑speed, low‑latency internet to customers.

In the messy world of satellite filings, something I have written about before in the aftermath of China’s ITUThe International Telecommunication U… filing for two constellations totalling 96,714 satellites, few questions seem to be asked about the deliverability of these programmes.

Whether it be Rwanda’s Cinnamon‑937 constellation of 337,320 satellites or Starlink’s filings to launch 29,998 satellites (made through the Kingdom of Tonga), an air of impossibility hangs over these “paper satellites”. At the ITU level, mechanisms to stop the abuse of the system can be circumvented. This extends to national regulators as well.

Fast forward to today, and Amazon is hurtling towards a problem: one of the approval conditions laid down by the US regulator, the FCC, for the Gen‑1 constellation was that 50 per cent (1,616 satellites) must be launched and operational no later than 30 July 2026.

The situation is not promising. A mere 241 Gen‑1 satellites are currently in orbit, and Amazon is highly unlikely to launch 1,375 satellites in 98 days – this would mean sending up about 14 satellites every day, perhaps more. The recent Globalstar acquisition could add a modest number of extra satellites to Amazon’s tally, but it still only chips away at the shortfall rather than solving it.

Failure to meet the FCC’s conditions could mean partial or full licence revocation. While responsibility for foresight lies with Amazon, which requested a licence extension in January, seeking an extra 24 months to meet the deadline, the FCC is also facing staunch, constant objections from rival satellite operator Starlink.

In a 20 April letter to the FCC, Starlink’s satellite policy director Jameson Dempsey said Amazon’s request “should be denied” and “the Commission could simply treat Amazon’s so‑called extension request for what it is: a modification that would increase interference to competitors”.

While not asking that Amazon stop launching more satellites after the deadline, Dempsey said that satellites it deploys after that date should be “deferred to a subsequent processing round” – delaying the company’s plans.

This is just the most recent letter – the exchange has been going on for months. In March, Amazon hit back at Starlink, calling it “the sole commenter disagreeing” with its extension plans and accusing the company of not engaging with FCC policies. Perhaps in retaliation, Amazon is calling on the FCC to block Starlink’s grandiose filing for a million orbiting data centres, and both companies are now accusing each other of being unable to deliver on their ambitions.

In many ways, both Amazon and Starlink are guilty of playing the system, and as the tit‑for‑tat around paper satellites continues, taking up even more of the FCC’s time, this situation becomes yet another case study in the flawed handling of satellite filings. The current rules are not preventing increasingly unrealistic plans from clogging the system.

Here is what else PolicyTracker reported on this week:

• Returning to Starlink, proposed orbiting data centres could share S-band spectrum as the FCC looks to create a new licensing framework
• The FCC is aiming to make at least 100 MHz of 3.98-4.2 GHz spectrum available for flexible-use licences by 2027
• Norway is working on a new 5G emergency network to be in place by 2031
• NASA’s Artemis II tested S-band, Ka-band and optical communications on its recent trip around the moon and back
• New Zealand’s government has delayed assigning the 26 and lower 28 GHz bands for mobile and FWA until 2028-2030, citing a lack of maturity in potential use cases