Abu Dhabi funds $1B expansion into a 50-satellite AI constel

Abu Dhabi is moving from buying satellite capacity to owning the technology chain behind it, and the clearest evidence is a $1 billion plan to expand the Altair fleet into a 50-satellite AI constellation. Orbitworks, the joint venture of UAE-based Marlan Space and US-French satellite builder Loft Orbital, announced the program on the opening day of the Paris International Space Summit this week, adding 40 spacecraft to the ten originally planned and equipping them with artificial intelligence supplied by Mistral AI. Investment in the expansion is anchored by Abu Dhabi's Royal Group, and the satellites will draw on a mostly European supply chain.

The expansion follows a consolidation step announced a day earlier in Abu Dhabi. International Holding Company said its International Tech Group unit is acquiring an 80 percent stake in UAE-based Marlan Holding, the group behind Marlan Space, in a deal that still requires regulatory approval. IHC chief executive Syed Basar Shueb has described the purchase as a way to give the space unit access to the conglomerate's capital and wider capabilities. Read together, the ownership move and the billion-dollar commitment show where Abu Dhabi wants its money working: in space infrastructure it builds and controls at home rather than in foreign imaging services.

IHC has said the acquisition extends its footprint into space technology and deep tech, covering satellite infrastructure, advanced manufacturing and adjacent fields, while supporting commercially scalable infrastructure with international reach. Marlan keeps its focus on developing space systems, autonomous operations and advanced computing from Abu Dhabi.

Orbitworks began in 2025 as the Middle East's first private space infrastructure venture, launched with initial backing of more than $100 million. Its Abu Dhabi manufacturing site has absorbed about $200 million and is designed to produce ten satellites a year, with headroom to scale toward fifty. The first Altair spacecraft is already complete and further units are being assembled on site, meaning the enlarged constellation rides on hardware in production rather than on paper.

The build-up runs through two partners at once. Marlan runs the Abu Dhabi factory and coordinates the program, drawing on a wider portfolio that spans space systems, autonomous operations and advanced computing. Loft Orbital contributes the standardized satellite architecture it has used to shorten the cost and lead time of bespoke missions, and it acts as the mission-operations partner for the fleet.

Altair itself is a multi-sensor Earth observation system. It combines optical, hyperspectral, thermal and infrared imaging with passive radio-frequency collection, and it performs processing on board the satellite rather than at a ground station. The imaging products are aimed at customers in Europe and the UAE. France's CNES secured dedicated capacity under a multi-year agreement in August, becoming the first institutional anchor customer before the fleet was expanded fivefold.

Inside the 50-satellite AI constellation

Keeping Mistral AI's models in orbit instead of on the ground is the architectural choice that makes a fleet this size workable. The venture is designing Altair so that onboard intelligence does the analysis where data is collected: software filters what the sensors record, flags meaningful changes and decides what earns scarce downlink capacity, shortening the time between a satellite passing over a target and a finished product reaching a customer.

The economics drive the design. Imaging satellites generate far more data than operators can afford to stream to Earth, and on-orbit inference turns that raw feed into finished answers, cutting both bandwidth cost and delivery time. For a 50-satellite AI constellation, the effect compounds: as the fleet grows, the share of traffic that can be processed, compressed or discarded in space rises with it. For security and infrastructure buyers in Europe and the Gulf, the practical difference is speed, since a change spotted during one pass can be checked on the next without a ground team in the loop.

The alternative architecture would be brute force: larger ground stations, more frequent downlink passes and heavier data pipelines on Earth. That approach works for a handful of spacecraft but gets expensive as the fleet multiplies, because every added satellite increases the volume of raw imagery waiting to be moved and stored. Processing in orbit inverts that equation, and choosing an established French AI supplier over an in-house model team lets the venture adopt proven technology without building a parallel terrestrial machine-learning operation.

What each side brings, and what it costs

The arrangement reads as a division of labor rather than a purchase order. Europe supplies the intellectual core: Loft Orbital's standardized satellite platform and its mission operations, Mistral AI's models and a predominantly European component chain. Abu Dhabi supplies the capital, the assembly hall and, increasingly, the corporate control. France receives guaranteed imaging capacity through the CNES agreement and an export channel for its new-space suppliers. The Emirates receive a domestic production line and a data product they own outright, built faster than organic development would allow.

The trade-offs are just as concrete. Feeding a UAE assembly line with European parts and AI keeps sensitive technology inside an export-control regime that will tighten as the fleet scales. IHC's 80 percent stake in Marlan Holding is still pending regulatory sign-off; once it closes, commercial direction will sit in Abu Dhabi while mission-critical software and operations stay with Western partners, a split that can strain governance. And the business case for the expanded 50-satellite AI constellation depends on demand catching up with production: CNES anchors the first contract, but the added spacecraft need more institutional buyers before factory capacity becomes revenue.

Underlying the deal is a structural shift in the space economy that predates this announcement. Launch costs have fallen far enough to make large fleets affordable, demand for satellite-derived data is growing across government and industry, and sovereign investors are building national capabilities rather than relying on foreign operators. IHC's move into Marlan is a direct expression of that trend, giving an Abu Dhabi conglomerate a controlled position in a market it previously watched from the outside. The same logic explains the venture's two home markets: Europe brings an institutional anchor and established buyers, while the Gulf brings capital and fast-growing demand for sovereign data.

The verdict

For decision-makers, this week's news shows sovereign capital industrializing AI in orbit rather than only funding terrestrial data centers. The concrete milestones to follow are regulatory approval of IHC's stake, the build rate at the Abu Dhabi factory as production scales toward fifty satellites a year, and new capacity deals beyond CNES. The Mistral partnership also places a European AI vendor inside a flagship Gulf-backed space program, a position that will matter as constellations become larger buyers of model technology.

Why this matters

These announcements shift Abu Dhabi from customer to owner of orbital data infrastructure, with capital now flowing into the factory floor, the launch manifest and the onboard AI rather than into foreign capacity leases. For European suppliers the program offers a scale that domestic budgets have not matched, and for the wider industry it signals that the next AI infrastructure build-out may happen above the Earth, financed by sovereign balance sheets.

✔Human Verified

Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.

Inside the 50-satellite AI constellation

What each side brings, and what it costs

The verdict

Why this matters

Related Articles