UAE-based CNTXT AI opens first US office in Milpitas

Company says $300 million in committed revenues, mostly US clients

CNTXT AI, a United Arab Emirates-based data and artificial-intelligence startup, is expanding into the United States with an eye toward building partnerships that bridge the U.S. and the Middle East. The company is opening its first U.S. office in Milpitas, California, and plans to hire a senior vice president of revenue, a vice president of marketing, and a commercial team of 15 to 25 people in its first year.

The company said it has $300 million in “committed revenues,” most of it coming from U.S. hyperscalers, AI developers, and large enterprises. The expansion follows a $60 million Series A financing round closed in June and co-led by AI71 and BlueFive Capital. In the same month, CNTXT AI acquired Actualize, which builds dialect-aware Arabic voice agents. Abu Sheikh, who is 25, co-founded the company and serves as its chief executive.

CNTXT AI spent two years developing Arabic-first AI models, and its Middle East and North Africa data help global AI leaders build models that work in underrepresented languages and real-world settings, according to Abu Sheikh. The company’s Munsit Arabic voice AI includes speech-to-text and text-to-speech models trained on more than 110,000 hours of Arabic-language audio across more than 25 dialects. The company said Munsit has more than 300,000 users. U.S.-based companies use it to serve Arabic-speaking customers in their native language and dialect, according to the company.

Beyond language models, CNTXT AI supplies data that hyperscalers, foundation model developers, and enterprise AI teams use for training and evaluation. The company said it also has robotics training facilities and a data collection network for training physical AI models.

Abu Sheikh framed the company’s robotics data work around the gap between laboratory demonstrations and real-world performance. “Basically the problem statement that we’re solving for is this: You find robots right now that are doing back flips and you see them…doing like Kung Fu moves and breakdancing. But you ask a robot to fold a T-shirt, it just breaks. Ask them to pick up a water bottle for you, it just breaks,” he said. The training of robots is overly reliant on lab data rather than data based on observation of people performing tasks in the real world, he added. “There aren’t enough human data that’s outside of the lab.”

Basing the data collection effort in the Middle East is an advantage because the region has a large blue-collar labor force that can be observed and recorded for AI and robotics training, Abu Sheikh said. The region also has world-class hotels, data centers, restaurants, and warehouses on par with the U.S. or Europe, and an “extremely supportive regulatory environment,” he said. He anticipates a rapid acceleration of product development and deployment in robotics over the coming 12 to 24 months.

The Silicon Valley presence is a necessity because 90% of the company’s clients are from the region, mostly hyperscale cloud-computing companies and robotics labs, Abu Sheikh said. Having a presence in the same time zone is important.

Abu Sheikh said he also wants to help companies navigate a rapidly evolving relationship between the U.S. and the Middle East and North Africa. In previous tech cycles, there was more of a buyer-seller relationship, he said; today, it is a partnership. “The relationship that we have with some of our clients here, they’re buying data off of us. So now they are the buyer. But then we are deploying the robots in the same locations that we collected the data from, right?” he said. “So, it’s a complete partnership cycle that we are both working towards together. Without each other, it’s going to be extremely difficult to achieve the same results.”