UAE-based data and AI company CNTXT AI announced on 20 October 2026 that it has surpassed AED 1.1 billion, or US$300 million, in committed revenue. The company also revealed its first United States expansion with a new office in the San Francisco Bay Area. Growth has been driven mainly by demand from US hyperscalers, AI developers, and large enterprises for Arabic language models, data products, and robotics training.
Committed Revenue and US Expansion
Founded in 2024, CNTXT AI spent two years building Arabic-first AI models and data operations before raising a US$60 million Series A in June 2026. The round was co-led by ai71 and BlueFive Capital, and it was followed by the acquisition of Actualize, an enterprise AI company focused on dialect-aware Arabic voice agents. CNTXT AI now supports more than 250 government and enterprise customers across three core areas: data products, AI models, and physical AI training.
Data Products
CNTXT AI supplies the data that hyperscalers, foundation model developers, and enterprise AI teams use for training and evaluation, including sourcing, multimodal annotation, expert reinforcement learning from human feedback, model evaluation, and red teaming. The company specializes in hard-to-source scientific, technical, industrial, and physical-world data where accuracy and context determine model performance. All work runs on CNTXT AI's own platform, with automated pipelines, quality controls, model-assisted pre-annotation, and human domain experts on every judgement call.
AI Models
CNTXT AI's best-known model is Munsit, which ranks first on Hugging Face's Arabic speech model leaderboard for accuracy. Its speech-to-text and text-to-speech models are trained on more than 110,000 hours of authentic Arabic-language audio across more than 25 dialects. Munsit has more than 300,000 users, and US-based companies including Boston AI, SPSoft, and Clarity use it to serve Arabic-speaking customers in products built in English.
Physical AI Training
CNTXT AI operates dedicated robotics training facilities and a data collection network that captures real-world data across more than 300 use cases for physical AI model training. It generates training data daily from diverse real-world environments, with consent and chain-of-custody documentation designed to meet rigorous privacy standards. The company has also filed a US patent application for its method to prevent diversity collapse, where robots copy a single demonstration but fail under real-world variation.
Regional Partnerships
CNTXT AI's growth is supported by deployments and partnerships across major regional organizations, including the Technology Innovation Institute, Nationwide Middle East Properties, and Venture One. The company's work includes specialized linguistic datasets, multilingual annotation frameworks, and evaluation methodologies that support AI research and Arabic-centric language model architectures. These partnerships reinforce its role in sovereign AI and regional data infrastructure.
US Expansion and Hiring Plans
The new San Francisco Bay Area office will support US government agencies, technology companies, AI developers, and large enterprises that need sovereign AI infrastructure inside the Middle East. CNTXT AI plans to hire a US-based Senior Vice President of Revenue and a Vice President Marketing, and build a 15 to 25 person commercial team in its first year. The expansion gives US organizations a direct path into Middle Eastern markets where sovereignty, regulation, language, and local requirements shape AI deployments.
CNTXT AI's growth reflects rising global demand for specialized data and language models that can operate in real-world and underrepresented language environments. Its expansion into the United States positions the company to serve a wider base of hyperscalers, AI developers, and enterprises while extending its data and robotics training capabilities. Founder and CEO Mohammad Abu Sheikh said the company intends to be the reason AI moves out of the lab and into real-world deployment.