UAE restructures artificial intelligence by merging regulatory authorities into a unified body and a $49 billion sovereign fund that links models to infrastructure

The United Arab Emirates reorganized its artificial intelligence and data ecosystem within a central government and investment structure during the first half of 2026, after Sheikh Mohammed bin Rashid Al Maktoum approved on June 14 the creation of the Federal Authority for Artificial Intelligence and Data. This new authority merges the AI Office and the digital government arm of the Telecommunications and Digital Government Regulatory Authority, as well as the National Data Office, into a single federal entity reporting directly to the Cabinet, and oversees the national strategy, data protection rules, and digital services under one umbrella led by the Ministry of Artificial Intelligence, headed by Omar Sultan Al Olama since 2017.

This merger ends years of regulatory overlap among three federal entities that issued overlapping guidelines to companies operating in the sector.Under this shift, oversight of the Personal Data Protection Law issued by Federal Decree No. 45 of 2021 transfers to the new authority, at a time when the law entered a one-year transition period that began in early January 2026, with full compliance required by early 2027. The system imposes administrative fines ranging from AED 50,000 to AED 5 million, directing inspection teams to enforce a 72-hour reporting deadline for any security breach and to produce documented, ongoing data-protection impact assessments, while the separate data frameworks for the Dubai International Financial Centre and the Abu Dhabi Global Market remain unchanged.

On the financial front, the Abu Dhabi-based technology investment fund MGX, owned by Mubadala Investment Company and the G42 Group chaired by Sheikh Tahnoon bin Zayed, closed the region’s first sovereign fund dedicated to artificial intelligence on July 1 2026 with a size of US$49 billion, surpassing its initial target. The fund directed its capital directly into stakes in major foundational model labs, participating in the Anthropic financing round at a valuation of US$65 billion, the OpenAI round in March at US$122 billion, and the XAI round in January at US$20 billion, alongside the acquisition of the data-center company Al-Land in a US$40 billion deal and a Paris computing complex with 3 gigawatts capacity.

This equation relies on simultaneous acquisition of Western foundational models and ownership of the physical infrastructure that runs them, rather than choosing between the two.In this context, the first phase of the UAE StarGate computing complex in Abu Dhabi, with a total capacity of one gigawatt, includes a 200-megawatt segment implemented by Khazna, a subsidiary of G42, and is scheduled to become operational in 2026 in partnership with OpenAI, Oracle, Nvidia, Cisco and SoftBank. Microsoft and G42 also announced an additional 200-megawatt expansion at Khazna facilities, while Core 42 increased its New York site from 18 to 60 megawatts, bringing its operating portfolio to ten locations worldwide.

At the level of language model development, the open-source Falcon H1 Arabic model, created by the Technology Innovation Institute with a hybrid architecture that combines Mamba and transformer designs and that tops the Arabic model evaluation leaderboard, together with the Gies series from the Inception unit of G42, trained on more than 126 billion Arabic tokens, constitute the two open-source options ready for production use in the Gulf. Microsoft has integrated Core 42’s Compass infrastructure suite into the Microsoft 365 Copilot application, and Brisight has begun deploying intelligent-agent tools to over 102 thousand small and medium enterprises registered with the Abu Dhabi Chamber of Commerce and Industry.

This central shift imposes a new practical reality on organizations, developers and systems-engineering teams across the Gulf and the Middle East, ending the dilemma of choosing between translated Western models and native Arabic models by providing production-ready, standards-based options such as Falcon and Gies, which lower local model-training costs and enable the creation of Arabic business tools with reduced inference costs. It also requires any technology company offering services in the UAE market or handling data of residents there to restructure its record-keeping and data-governance policies to comply with the strict fine caps and the 72-hour notification deadline, while the unified licensing and investment authority gives technology leaders a single regulatory partner with both legislative influence and financing capability.