Updated August 2026.
- The UAE National Data Strategy, overseen by the National Data Office (NDO), grants ML/data science companies structured access to government datasets under data-sharing agreements.
- Dubai Pulse, the emirate’s unified data platform, provides anonymised smart city datasets usable for urban mobility, utilities, and public health ML models.
- MBZUAI (Mohamed Bin Zayed University of Artificial Intelligence) is the UAE’s primary ML research partner for startups, offering TTO-managed academic collaboration agreements.
- DIC/TECOM free zone licenses ML and data science under “Data Analytics” and “Machine Learning Solutions” activity codes, with first-year setup costs of AED 50,000–AED 300,000.
- Healthcare AI opportunities are concentrated around Cleveland Clinic Abu Dhabi and Mubadala Health, both of which maintain vendor programmes for ML diagnostic tools.
- UAE data localisation laws require certain categories of health, financial, and government data to remain on UAE-sovereign infrastructure.
The UAE National Data Strategy and the National Data Office
The UAE National Data Strategy, launched under the mandate of the UAE Data Office (rebranded as the National Data Office — NDO), establishes a federated framework for how government data is classified, shared, and monetised. For ML and data science companies, the NDO is the primary gateway to government dataset access — which is commercially invaluable given the richness of UAE administrative data on trade, logistics, health, and urban mobility accumulated over decades of digital government investment.
The NDO operates a data-sharing agreement (DSA) framework under which private companies can apply for licensed access to specific federal datasets. Eligibility requires: UAE company registration (free zone or mainland), a documented data governance policy aligned with UAE Data Law (Federal Decree-Law No. 44 of 2021 on the Organisation and Protection of Personal Data — commonly referred to as the UAE PDPL), a data security certification (ISO 27001 or NESA-equivalent), and a defined use-case with demonstrable public benefit. Access is typically approved within 90–180 days of application, with datasets provisioned via a secure NDO data exchange portal rather than direct download.
Strategically, NDO-approved companies gain a significant market positioning advantage: they can credibly represent to enterprise and government clients that their ML models are trained on official UAE government data, a claim that carries substantial weight in public sector procurement processes. Companies should factor NDO engagement into their go-to-market strategy from incorporation, not as an afterthought.
Dubai Pulse: Smart City Data for Urban ML Models
Dubai Pulse is the Smart Dubai Office’s unified data platform, consolidating real-time and historical datasets from across Dubai’s city infrastructure. Categories available to registered developers and research partners include: road network and traffic (from RTA), utility consumption patterns (from DEWA), weather and environmental data (from Dubai Municipality), economic activity indices, and population mobility proxies derived from anonymised telecoms data (via du and e&).
ML companies can access Dubai Pulse through two tiers: the public API (free, limited resolution, anonymised at emirate level) and the premium data partnership programme (paid, higher resolution, borough-level, requires a formal agreement with the Smart Dubai Office). Premium partnership costs are negotiated case-by-case but typically range from AED 30,000 to AED 200,000 annually depending on data volume and exclusivity terms. Companies building urban mobility, demand forecasting, or ESG analytics products should prioritise Dubai Pulse premium access as a core data asset.
A parallel platform, Abu Dhabi Data (operated by ADDA — Abu Dhabi Digital Authority), provides equivalent smart city datasets for the Abu Dhabi emirate. ADDA’s data sharing framework operates under the Abu Dhabi Data Law (Law No. 45 of 2021) and is similarly accessible via a formal data partnership application. Combined Dubai Pulse and ADDA data access effectively covers over 85% of the UAE’s total economic and population activity.
MBZUAI Research Partnerships for ML Startups
The Technology Transfer Office at Mohamed Bin Zayed University of Artificial Intelligence manages all commercial research collaborations with private sector entities. For ML and data science companies, MBZUAI’s most strategically relevant partnerships are in: natural language processing (particularly Arabic NLP, where MBZUAI’s AraT5 and CAMeL Lab tools are globally recognised), computer vision for industrial and medical imaging, federated learning for privacy-preserving healthcare analytics, and time-series forecasting for energy and financial applications.
Sponsored research agreements with MBZUAI typically cost AED 200,000–AED 800,000 per year and include: co-development rights over resulting IP (negotiated on a case-by-case basis, typically 30–70% to the company), access to MBZUAI’s GPU compute cluster (currently over 10,000 NVIDIA A100-equivalent GPUs via the Condor Galaxy partnership with G42 and Cerebras), and co-authorship on publications. For early-stage companies, MBZUAI’s Incubation Programme offers lighter-touch engagement — mentoring, co-working space in Masdar City, and introductions to Abu Dhabi’s investment ecosystem — for an equity stake of 5–10%.
Healthcare AI: Cleveland Clinic Abu Dhabi and Mubadala Health
The UAE healthcare sector is the most data-intensive and ML-ready vertical in the country, driven by two anchor institutions: Cleveland Clinic Abu Dhabi (CCAD) and Mubadala Health. CCAD, a joint venture between the Cleveland Clinic (USA) and Mubadala Investment Company, maintains a dedicated health informatics team and an open vendor programme for ML diagnostic tools. Accepted solutions must comply with the Dubai Health Authority (DHA) digital health framework and Abu Dhabi DoH (Department of Health) medical device regulations, which classify AI diagnostic software as a regulated medical device requiring pre-market approval.
Mubadala Health’s portfolio includes Imperial College London Diabetes Centre, Healthpoint Hospital, and Corniche Hospital, representing a network of over 1,500 beds and millions of patient interactions annually. Mubadala Health’s Data Science and Analytics team actively sources ML solutions for patient flow optimisation, chronic disease prediction, and imaging analysis. Entry into the Mubadala Health vendor ecosystem typically requires: ISO 13485 certification (medical devices quality management), UAE DoH or DHA AI medical device approval, and a formal pilot proposal submitted through Mubadala’s innovation portal.
For ML companies not yet certified for clinical deployment, the Malaffi health information exchange (HIE) platform in Abu Dhabi and the Nabidh HIE in Dubai provide controlled environments for developing and validating clinical NLP and predictive models without requiring full medical device approval — subject to research use limitations and patient data anonymisation protocols.
DIC/TECOM Licensing for ML and Data Science Firms
Dubai Internet City (DIC) and the broader TECOM Group free zone network are the preferred incorporation destinations for ML and data science companies in the UAE. TECOM’s trade licence activities relevant to data science firms include: Data Analytics Services, Machine Learning Solutions Development, Business Intelligence Consulting, Data Engineering and Architecture, and AI Model Training Services. Multiple activities can be listed on a single licence at marginal additional cost (typically AED 1,000–AED 2,000 per additional activity).
Visa allocation under a DIC licence follows a floor-space formula: a flexi-desk provides a quota of 1–2 visas; a 22–50 sqm office provides 4–6 visas; a 50+ sqm office provides 8–15 visas. Given that ML teams are typically small (3–8 people) but may require several specialised hires (data engineers, ML engineers, domain scientists), a mid-tier office (30–40 sqm, approximately AED 80,000–AED 100,000 per year) is typically optimal for balancing cost and visa capacity.
Total first-year costs for a DIC-based ML company with a 4-person team: trade licence AED 20,000, office space AED 90,000, visa processing AED 20,000, legal and registration fees AED 15,000, total approximately AED 145,000. Costs scale to approximately AED 300,000 for a 10-person team with dedicated server room and ISO 27001 certification overheads.
Data Localisation and UAE Data Sovereignty Requirements
UAE data sovereignty law affects ML companies that handle personal data belonging to UAE residents or data generated within UAE critical infrastructure. The UAE PDPL (Federal Decree-Law No. 44 of 2021) restricts the cross-border transfer of personal data without explicit consent, a contractual adequacy framework, or approval from the UAE Data Office. This means ML models trained on UAE personal data (health records, financial transactions, biometric data) must be trained on infrastructure physically located in the UAE or within an approved data transfer arrangement.
Approved UAE-sovereign cloud infrastructure includes: Microsoft Azure UAE North (Abu Dhabi), Microsoft Azure UAE Central (Dubai), Google Cloud Platform (UAE-hosted via e& partnership), G42 Cloud, and Khazna Data Centres. AWS currently routes UAE-region workloads through its Middle East (UAE) region (me-central-1, located in Dubai). All five meet UAE data localisation requirements for most PDPL categories. Selecting a compliant cloud provider from day one avoids costly data migration exercises as your ML pipeline scales.
| Data Category | Localisation Requirement | Governing Authority | Penalty (AED) |
|---|---|---|---|
| Health Records | UAE-only (no exceptions) | DHA / DoH | Up to 5,000,000 |
| Financial Data | UAE or approved jurisdiction | CBUAE / SCA | Up to 10,000,000 |
| Government Datasets | UAE sovereign infrastructure | National Data Office | Criminal liability |
| General Personal Data | Consent or adequacy required | UAE Data Office | Up to 1,000,000 |
| Biometric Data | UAE-only (sensitive category) | UAE Data Office | Up to 5,000,000 |
FAQ: ML and Data Science Companies in the UAE
How do I access UAE government datasets for ML model training?
Apply to the National Data Office (NDO) through their official data-sharing agreement portal. You will need a UAE company registration, an ISO 27001 or NESA-equivalent security certification, a PDPL-compliant data governance policy, and a defined use-case with demonstrable public benefit. Approval typically takes 90–180 days. Dubai Pulse (Smart Dubai Office) and Abu Dhabi Data (ADDA) provide separate portals for emirate-level smart city datasets.
What makes DIC/TECOM better than ADGM or DIFC for an ML company?
DIC/TECOM offers lower setup costs (AED 50,000–AED 150,000 first year vs AED 100,000–AED 280,000 for ADGM), a larger existing tech community for talent networking, and proximity to major Dubai enterprise clients including Emirates Group, Emaar, and Majid Al Futtaim. ADGM is preferable if your ML products target Abu Dhabi’s financial services sector or if you seek ADGM’s regulatory sandbox for financial data applications.
Do I need a special licence to deploy ML models in UAE healthcare?
Yes. ML models used for clinical diagnosis, prognosis, or treatment recommendation in UAE healthcare facilities are classified as Software as a Medical Device (SaMD) and require pre-market approval from either the Dubai Health Authority (for Dubai) or the Abu Dhabi Department of Health (for Abu Dhabi). This process typically takes 6–18 months and requires IEC 62304 (medical software lifecycle) and ISO 14971 (risk management) documentation.
Can a foreign ML company own its UAE entity without a local partner?
Yes. The UAE Companies Law (Federal Decree-Law No. 32 of 2021) allows 100% foreign ownership in all major technology free zones (DIC, ADGM, DIFC, DMCC) and in UAE mainland for most technology sector activities since the 2021 Commercial Companies Law reforms. A local partner (UAE national holding 51%) is no longer required for most business categories.
What GPU compute infrastructure is available in the UAE for ML training?
G42’s partnership with Cerebras (Condor Galaxy cluster) and its internal GPU infrastructure provides the UAE’s largest domestic compute capacity, accessible commercially via G42 Cloud. Microsoft Azure UAE North, AWS me-central-1, and Google Cloud UAE provide global-standard GPU instances (A100, H100) within UAE data centres. For sovereign compute with government data access, the UAE’s national AI compute programme targets 100,000+ GPUs by 2026 under the NATI (National AI Technology Initiative).