Loft Orbital and Marlan Space have announced a programme valued at $1bn to expand their constellation of AI-equipped satellites from 10 spacecraft to 50. Under the plan, the French AI company Mistral will supply the models that run on board. The two companies presented the initiative in a press release dated 9 September, and the announcement was made publicly by French President Emmanuel Macron during an international space event.
What happened
The core of the announcement is a significant scaling of an existing satellite fleet. Rather than launching a single demonstration craft, the partners intend to grow an operational constellation more than fivefold, moving from 10 satellites to 50. Each of these spacecraft is described as carrying AI capabilities, with Mistral providing the models that operate in orbit.
The programme carries a stated value of $1bn, which signals that this is not a research experiment but an industrial-scale commitment. The involvement of the French president in the announcement also indicates that the project is being positioned as a strategic, nationally relevant initiative rather than a purely commercial venture between private firms.
The press release did not detail the specific tasks the onboard AI models will perform, nor the timeline for launching the additional satellites. What is clear from the announcement is the structure of the partnership: Loft Orbital and Marlan Space handle the satellite side, while Mistral contributes the artificial intelligence layer that runs on the hardware.
What it means for business
The move reflects a broader trend in which AI is no longer confined to data centres on the ground. Placing models directly on satellites means that some processing and decision-making can happen in orbit rather than depending on a round trip to Earth-based infrastructure. For industries that rely on satellite imagery and connectivity, such as logistics, agriculture, insurance and telecommunications, this can translate into faster insights and less dependence on ground stations.
The partnership also shows how AI vendors are seeking new deployment environments beyond conventional cloud and enterprise settings. For companies that are already integrating AI agents into their operations, the lesson is that the underlying models are becoming more portable and more embedded in physical systems. That shift gradually changes expectations: customers begin to assume that intelligence is available wherever data is generated, not only in a central platform.
For businesses evaluating their own AI roadmaps, the announcement is a reminder that infrastructure choices matter as much as model selection. The scale of the commitment, from 10 satellites to 50 and a $1bn budget, suggests that serious players are willing to invest heavily in specialised AI hardware and deployment. Companies that adopt AI today should keep an eye on how these orbital systems mature, because they may eventually feed new data services and operational tools into everyday commercial workflows.
