AI for satellite communications networks: Junction Brief – November 2026

Acentric Research is produced with AI tools under human review. The authors are not financial professionals, and nothing here is investment advice or a recommendation to buy or sell any security. AI analysis can be wrong; check every claim against its cited source.

This Brief is an initial scan, not a full investigation. The method and scores behind it are in AI × Space in the Landscape Report.

Overview

A communications satellite carries a limited supply of radio spectrum, power and antenna beams. This Brief covers artificial intelligence (AI) proposed for part of that work: steering beams, allocating capacity, detecting anomalies and handling faults.

The sources opened show intent and study but little proof. UK Space Agency guidance describes a 2025 European Space Agency (ESA) funding call for AI in satellite communications, and a 2021 review surveys many possible uses. [1][2] The sources opened give no demonstration in orbit and no list of awarded projects. The review is from 2021 and the funding guidance from 2025; the ESA page is undated in the text read. Later developments were not searched.

Current state

A funding call. UK Space Agency guidance describes an “AI for Satcom” call under ESA’s ARTES programme, with a €5,000,000 envelope and no upper or lower limit on a project. It quotes a description of the call: AI is “revolutionizing satellite communications” through resource allocation, dynamic beamforming and real-time anomaly detection. UK applicants needed national approval first; expressions of interest were due on 22 May 2025. [1] That is the call’s own view of the field, not a finding. The guidance gives dates for UK decisions and does not list the outcome, and no list of awarded projects was found. [1]

Why the field expects more autonomy. ESA’s Intelligent System Initiative page says future satellites, and constellations in particular, will increase autonomy to allow more complex operations, lower operating costs and new services, and it sets the aim of distributing intelligence among individual satellites, a fleet or constellation and the ground. [3] The UK guidance describes ARTES as a commercially driven programme to support UK businesses, to which the UK subscribed £190 million in 2022. [1]

ESA’s autonomy work plan. ESA’s Intelligent System Initiative lists tender titles that include an AI-based system for autonomous failure isolation, recovery and resource optimization on a communications constellation, machine learning techniques for data rate reduction, and autonomous health monitoring for platform mechanisms. The page lists titles, not results. [3] It also lists two system studies of an “Intelligent Platform for Constellations”, by Thales and by OHB System, and a study of an “Enhanced AOCS On-Board Autonomy and Adaptability for Satcom Constellations” (AOCS is the system that keeps a satellite pointed and positioned), whose invitation to tender was pending when the page was read. [3] It also lists in-orbit experiments for communications satellites; for one of them it says the planned experiment was cancelled and is to be replaced by industry-proposed experiments. [3]

The research literature. A 2021 review by two authors, read here in its arXiv version, lists applications of AI in satellite communications, including beam hopping (moving a beam from place to place), anti-jamming, traffic forecasting, channel modeling, interference management and energy management. The authors write that these have shown excellent potential. The abstract also says managing resources, networks, security and spectrum is harder for satellites than for ground networks. [2]

Who is involved. In the sources opened: the UK Space Agency, which gives national approval to UK applicants; ESA, which runs the call and the Intelligent System Initiative; Thales and OHB System, named in the titles of two platform studies; and the two review authors, who are academics. [1][3][2] No company’s own description of an AI product for satellite networks was opened.

How far it has come. At the stage of funding calls, studies and reviews in the sources opened. No in-orbit demonstration of AI managing a communications payload was found, and the sources opened do not say whether operators use AI in service. That is not evidence that none exists.

Opportunities

The call’s own text says demand for reliable, low-delay connectivity in remote and mobile settings is growing and that AI-driven systems are becoming essential. [1] If methods like those in the review work on real payloads, they could allocate capacity and handle faults with less ground effort. [2] The test is a flight demonstration with measured results.

Challenges

  • Engineering. The review states that resource management, network control, security and spectrum management are more challenging for satellite networks than for ground networks. [2] Running AI on a satellite needs computers that survive radiation; see the space-qualified computing Junction.
  • Evidence. The review was read as an arXiv abstract and the call is a funding document. Neither reports results from operating satellites. [2][1]
  • Capital and scale-up. ESA’s cancellation of one planned in-orbit experiment shows that flight tests are not guaranteed to proceed as first planned. [3]

Data sources

  • ESA ARTES programme pages and the esa-star portal, for calls and tenders. [3][4]
  • National space agency guidance, such as the UK Space Agency’s, for call terms and dates. [1]
  • Research literature, through arXiv, for methods and simulations. [5]
  • No registry or award list was found for this Junction, and none records which operators use AI.

Changelog

  • 2026-11-18: First published.

Sources

  1. UK Space Agency, “ARTES call: AI SatCom call guidance”, GOV.UK, 2025 (guidance for UK applicants to an ESA call; it quotes the call’s own description of AI in satellite communications). gov.uk (accessed 2026-11-18)
  2. Fares Fourati and Mohamed-Slim Alouini, “Artificial Intelligence for Satellite Communication: A Review”, arXiv:2101.10899, 2021-01-25 (arXiv version; the arXiv record gives DOI 10.23919/ICN.2021.0015, a journal version that was not read). arxiv.org (abstract read through the arXiv API, accessed 2026-11-18; the full text was not read)
  3. European Space Agency, “Intelligent System Initiative”, ARTES Advanced Technology programme page, with its work-plan table (agency statement). resilience.esa.int (accessed 2026-11-18)
  4. European Space Agency, esa-star Publication portal (named on the Intelligent System Initiative page as where tenders are published). esastar-publication.sso.esa.int (listed as a data source, not opened for this Brief)
  5. arXiv, API and search. arxiv.org (listed as a data source; the API was queried for one abstract on 2026-11-18)