The one-way traffic of AI talent — from everywhere into Silicon Valley — just got a little less one-way. CNTXT AI, a data and artificial-intelligence startup based in the United Arab Emirates, is opening its first American office in Milpitas, California, and says it wants to build partnerships that bridge the United States and the Middle East.
The company spent two years building Arabic-first AI models and assembling Middle East and North Africa data that it says helps global AI leaders make their models work in underrepresented languages and real-world settings. That is the pitch: frontier labs can build the best model in the world, but it still stumbles on Arabic dialects, Gulf compliance rules, and robotics training in desert conditions.
Models trained on 110,000 hours of Arabic
CNTXT AI's headline product is Munsit, an Arabic voice AI platform whose speech-to-text and text-to-speech models are trained on more than 110,000 hours of Arabic-language audio across more than 25 dialects. The company says Munsit has more than 300,000 users, and that US-based companies already use it to serve Arabic-speaking customers in their native dialects.
The business has three other legs. It supplies labeled training and evaluation data that hyperscalers and foundation-model developers use to build and test models. It sells custom AI solutions designed to run inside an enterprise's own systems — the "sovereign AI" play, where data never leaves national borders. And it operates robotics training facilities and a data collection network for training physical AI models, the unglamorous real-world data that labs need to make robots useful.
"You find robots right now that are doing back flips and you see them doing like Kung Fu moves and breakdancing," co-founder and chief executive Mohammad Abu Sheikh told the Wall Street Journal. "But you ask a robot to fold a T-shirt, it just breaks."
$300 million in committed revenues, a $60M raise behind it
The company says it has more than $300 million in "committed revenues," most of it coming from US hyperscalers, AI developers, and large enterprises. In June, it raised a $60 million Series A co-led by AI71 — the Abu Dhabi applied-AI company that acquired Abu Sheikh's previous venture, LocAI — and BlueFive Capital. In the same month, CNTXT AI acquired Actualize, which builds dialect-aware Arabic voice agents.
The new Milpitas office will start with a commercial team of 15 to 25 people, including a senior vice president of revenue and a vice president of marketing — a classic beachhead formation for a company that wants to sell to American labs and enterprises without asking them to fly to Abu Dhabi first.
The Gulf's AI ambitions, going west
CNTXT AI's move lands inside a much bigger story: the Gulf states are pouring sovereign wealth into AI — Abu Dhabi's MGX has assembled one of the largest AI funds ever, with stakes in frontier labs — and the next step is turning capital into exports. CNTXT AI is one of the first Gulf AI companies to go west instead of waiting for Silicon Valley to come east.
Whether Arabic-first models and sovereign-data tooling can become a durable American business is the open question. "Committed revenues" is a company claim, not audited revenue, and selling to US hyperscalers means competing with the hyperscalers' own data pipelines. But the trend line is clear: the AI race is no longer just about who trains the best model in San Francisco. It's about whose data — and whose dialects — that model actually understands.
By Hamza Faisal
