UAE-based artificial intelligence company CNTXT AI is expanding into the United States. The company is taking this step after securing USD 300 million in committed revenue, with plans to open its first American office in the San Francisco Bay Area.
Founded in 2024 by Mohammad Abu Sheikh, the company provides data services and AI models for businesses developing and deploying artificial intelligence systems. According to the company, most of its committed revenue comes from US-based hyperscalers.
CNTXT AI has built its business around specialized training data, Arabic-language AI and physical AI training. It currently serves more than 250 government and enterprise customers across these areas.
The company raised USD 60 million in Series A funding in June and acquired enterprise AI company Actualize. Its US expansion will target government agencies, technology firms, AI developers and large businesses seeking expertise in Middle Eastern markets.
“Conducting business successfully in the Middle East requires a deep understanding of Arabian business culture, compliance mandates, enterprise environments, dialects and other requirements that directly affect adoption.
“Our expansion to the Bay Area gives US organizations a direct path into those capabilities, and into markets across the Middle East where sovereignty, regulation, language and local operating requirements shape every major AI deployment,” said Sheikh.
CNTXT AI has developed Munsit, an Arabic voice AI platform that ranks first on Hugging Face’s Arabic speech model leaderboard, according to the company. Its speech recognition and speech generation models use more than 110,000 hours of Arabic audio covering over 25 dialects.
Munsit has more than 300,000 users. US companies, including Boston AI, SPSoft and Clarity, use the technology for customer service, voice agents and transcription.
The company also provides specialized datasets for AI training and evaluation, including scientific, technical and industrial material. Its physical AI division collects real-world training data across more than 300 use cases for robotics development.
CNTXT AI has filed a US patent application for a method designed to prevent ‘diversity collapse’ in robot training, where systems struggle to adapt beyond a single demonstration.
“What comes next is bigger: whoever can capture real work, legally and at scale, decides which AI makes it out of the lab and which stays a demo. We intend to be the reason it happens,” Sheikh said.
Also Read: AI Boom and Startup Funding Surge Across MENA Drives Investor Activity