In-Region GPU Infrastructure Comes to the UAE

Telnyx has launched sovereign GPU infrastructure in Dubai. Inference compute is live in the UAE on GPUs that Telnyx owns and operates, with data stored and processed inside the country under a single API and a single contract. (Read the release notes.)

For most enterprises in the Middle East, the question is no longer whether to adopt AI. It is where the AI runs.

Regulated buyers across banking, healthcare, insurance, and enterprise SaaS in the UAE and wider GCC now face a hard requirement: the data their AI touches must stay in-region. Not just at rest, but end to end. And that requirement has been getting sharper, not softer.

The stitched-vendor problem

In-region compute has technically been available in the UAE for a while. Hyperscalers and sovereign cloud providers offer it. But reaching it has meant stitching together separate vendors for connectivity, voice, and inference. Each brings its own contract, its own integration surface, its own compliance evidence, its own procurement cycle.

For an enterprise inference project, that stitched stack is the thing that stalls the launch. Every vendor added is another compliance review to clear, another SLA to reconcile, another audit trail to defend. Projects that clear technical review often die in the legal or procurement one.

Sovereignty by configuration only works if you have the time and legal capacity to configure and defend it.

Sovereignty by architecture

Telnyx has been building a private global communications network for over a decade, on infrastructure it owns end to end, from fiber to GPU. The Dubai launch extends that owned stack into the UAE, with the me-central-1 region now live for inference.

That distinction matters. Most inference providers rent their compute. Telnyx owns the GPUs, the network they sit on, and the fiber connecting them to customers. When Telnyx says data stays in the UAE, it is not a configuration setting on a partner's infrastructure. It is architectural.

For teams that have to defend AI systems in front of a regulator, that difference shows up in the review.

What runs on the Dubai cluster

The Dubai cluster is general-purpose inference infrastructure. While Telnyx is best known for powering real-time voice AI agents, the same GPUs serve any inference workload:

• Chatbots and copilots
• Document processing and retrieval-augmented generation (RAG)
• Fraud detection and analytics
• Real-time voice agents
• Any other workload that benefits from in-region inference

Customers access open-weight models through an OpenAI-compatible API by targeting the me-central-1 region with a single endpoint change. Kimi K2.6 is the first model available in the region, with additional models rolling out soon. No new SDK, no rearchitecting. See the developer docs for integration details.

Real-time voice and in-region latency

Voice AI is where in-region infrastructure earns its keep most visibly.

When speech-to-text, inference, and text-to-speech all run in the same Dubai facility, round-trip latency drops significantly. Co-locating inference with regional infrastructure eliminates the 200-400ms of cross-border routing overhead that MENA users would otherwise absorb on every inference call. That is fast enough for a real conversation rather than the stilted, scripted exchange that customers hang up on.

For the contact centers, banks, and insurers running real-time customer operations at scale, in-region inference is what turns voice AI from a proof of concept into a production system.

Compliance built into the platform

For regulated buyers, technical fit is only half the decision. The other half is what the platform can prove.

Telnyx operates against SOC 2 Type II, HIPAA, PCI, and GDPR standards. These are the certifications that banking, healthcare, and insurance procurement teams require before AI touches sensitive data. Combined with in-region processing, they close the gap between an inference project that clears compliance and one that does not.

Who this is built for

The Dubai launch is built for organizations across the UAE and wider GCC where in-region inference is not optional:

• Banking and financial services running fraud detection, customer authentication, and real-time advisory
• Healthcare running clinical documentation, patient intake, and diagnostic assistants
• Insurance running claims processing, underwriting, and customer service automation
• Enterprise SaaS serving regulated MENA customers
• Contact centers running real-time voice operations at scale

For these teams, the choice was previously between accepting the stitched-vendor overhead or delaying the AI roadmap until the compliance path cleared. The Dubai cluster gives them a third option: production AI, in-region, on infrastructure their compliance team can actually sign off on.

Availability

Inference in Dubai is available through the Telnx Chat Completions API, targeting the me-central-1 region. Sign up in Mission Control to start testing, or contact your account team for early access. UAE data storage is available to all Telnyx customers through the Mission Control Portal.

About Telnyx

Telnyx is real-time AI infrastructure. Edge compute, voice AI platform, and global communications in one system. Carrier-owned network, local numbers in 140+ countries, 18 global points of presence with colocated GPU infrastructure. Built for bots that talk to humans. Learn more at telnyx.com.