Starlink in Operation: Investigation

An independent investigation into how Starlink works, who controls it, its military use, and the risks of depending on a single privately controlled network. Commissioned discussion paper: draft for review, October 2026.

Read the study findings Discuss the findings
Nineteenth-century engraving of the Trojan horse being brought into Troy, with a satellite terminal added on its back and a satellite in the sky
Gift or Trojan horse? A nineteenth-century engraving of the Trojan horse, modified to add a satellite terminal and a satellite.

About the investigation

Starlink, run by SpaceX, has become the largest satellite network ever built, and is now central to Ukraine's defence, to US military communications and to connectivity in places that cables and phone masts do not reach. This investigation, carried out as part of our work in technology and innovation assessment, asks what that means in practice.

It looks at what Starlink is, the technology behind it, how it is controlled and switched on and off by area, its military and government uses, its links to the AI company xAI and its chatbot Grok, and its history in conflicts. It then ranks the potential dangers, sets out actions to reduce them, and offers two old stories, the Trojan horse and Dr Strangelove, for discussion. Every factual claim is referenced and rated as Confirmed, Reported or Unverified.

Main findings so far

  • Scale: about 11,100 satellites in orbit and more than 12 million customers in over 160 countries (October 2026).
  • Not an independent “internet of the sky”: all internet traffic still passes through SpaceX-controlled ground stations.
  • Central control: SpaceX can switch service on or off for a single terminal, a region or a whole country, and service has been restricted during conflicts.
  • Concentrated ownership: one person holds about 82% of the voting power in SpaceX, and the safeguards in government contracts are not public.
  • Military role: SpaceX's military business, Starshield, is becoming part of US missile-defence plans. SpaceX also owns xAI, the company behind Grok.
  • The main danger is structural: dependence on a single point of control and failure, rather than any one person's intentions.

Documents

Documents open in a viewer on this page; close the viewer to return here.

Study findings: full paper (47 pages) Press and internet update, 15 October 2026 (4 pages) Investigation outline (1 page)

Latest update (15 October 2026): Ukraine says its forces lost full use of Starshield, the military version of Starlink, when a speed limit was applied without warning to terminals on drones and aircraft. The press and internet update sets out what is known, and who controls the satellites.

For discussion

Read about the Trojan horse? Starlink arrived in Ukraine as a gift, became indispensable, and the keys stayed with its owner. Seen Dr Strangelove? Kubrick's film warned of systems where one decision can set off events that nobody can call back. The paper asks how far either story fits, and what should be done.

Two views

To stimulate debate, two companion articles each deliberately argue only one side. Read them together, and decide for yourself.

The case against: Is Starlink a Trojan horse? (4 pages) The case for: So many new opportunities (4 pages)

Prepared by Mike Anstey · Research assistant: Claude (AI, Anthropic)

🤝 Comments and Collaboration

The findings are a draft for review. We welcome comments, corrections and new evidence, and would be glad to hear from anyone interested in collaborating on technology and innovation assessment.

Get In Touch