Abu Dhabi-based AIREV is pushing its autonomous AI platform deeper into physical computing infrastructure through a new strategic agreement with Qualcomm Technologies.
The deal will bring AIREV’s OnDemand platform to Qualcomm Dragonwing products, including the Dragonwing AI On-Prem Appliance. That means AI agents built and managed through OnDemand can operate closer to where an organisation’s data actually lives — inside its own infrastructure, at the edge, or potentially directly on devices.
No mandatory trip to the cloud.
That detail matters.
AIREV and Qualcomm Are Taking Agentic AI to the Edge
OnDemand is AIREV’s no-code and low-code platform for creating, deploying and managing autonomous AI agents. Under the Qualcomm agreement, that software layer is being integrated with Dragonwing hardware. OnDemand bundles are also expected to be factory-preloaded into Dragonwing chipsets for global distribution, making deployment easier for customers using Qualcomm-powered systems.
It is a different proposition from simply putting another AI application on a server.
AIREV gets access to Qualcomm’s hardware footprint. Qualcomm gets an established agentic AI platform that can sit on top of its edge and on-premise computing products. The two pieces fit rather neatly.
Four Million Users and 6.8 Trillion Tokens
AIREV isn’t arriving at the partnership with an experimental platform. OnDemand already serves more than four million users globally and provides more than 300 specialised AI agents across over 50 languages. AIREV says its products processed 6.8 trillion tokens during the second quarter of 2026 — roughly six times the volume recorded a year earlier.
Those numbers make the Qualcomm agreement more interesting. This isn’t only about proving that autonomous agents can run outside massive cloud data centres. AIREV is trying to put an already-used AI platform onto hardware that businesses and governments can control themselves.
Why Running AI Without the Cloud Matters
Cloud AI is convenient. It also creates some awkward questions when the information being processed is confidential, regulated or simply shouldn’t leave an organisation’s infrastructure. Running inference locally changes that equation.
An enterprise could keep sensitive information on-premises while still using AI agents for document analysis, internal workflows, automation or other specialised jobs. Moving computation closer to the data can also reduce the latency created by repeatedly sending information to remote cloud infrastructure.
Qualcomm has been steadily building around this idea. Its Snapdragon platforms, for example, already support generative and agentic AI experiences that execute locally, while its Dragonwing portfolio extends edge computing into enterprise and industrial environments. AIREV now has another route into that ecosystem.
Dragonwing Gives OnDemand a Physical Home
One of the more notable pieces of the agreement is the Qualcomm Dragonwing AI On-Prem Appliance. The system is designed to give organisations local generative AI infrastructure without requiring them to build everything themselves. Configurations range from systems aimed at models up to 10 billion parameters to higher-end deployments capable of handling models up to 70 billion parameters.
Put OnDemand on top and the hardware gains a ready-made environment for building and operating AI agents. AIREV, meanwhile, gets something software companies spend years chasing: a path into enterprise hardware deployments.
And potentially much more than one appliance. Qualcomm says the collaboration establishes groundwork for bringing AIREV’s capabilities to a broader range of Qualcomm technologies over time.
Sovereign AI Is Quietly Becoming a Bigger Part of the Story
There is another angle here that shouldn’t get buried beneath the hardware specifications.
Sovereign AI.
Governments and large organisations increasingly want AI systems that don’t automatically require their data to be shipped into infrastructure controlled somewhere else. A platform that can run inside a sovereign data centre, corporate server or edge device gives buyers another option.
For the UAE, there is a strategic angle too.
AIREV was founded in Abu Dhabi in 2024. Two years later, its technology is moving into the ecosystem of a major US semiconductor and computing company. The partnership also includes a pathway for AIREV into the US market.
UAE Minister for Foreign Trade and AIREV Chairman Dr. Thani bin Ahmed Al Zeyoudi framed the agreement as evidence that the country is moving beyond importing technology and increasingly exporting it.
That’s probably the bigger story. Not simply another AI partnership announcement, but a UAE-built agent platform finding its way into the hardware layer underneath enterprise AI.
Qualcomm Is Already Betting on AI That Lives on Your Devices
The AIREV agreement also lands during a broader Qualcomm push toward local and agentic AI.
In August, Qualcomm highlighted five agentic AI applications — AnythingLLM, Pokee AI, LLMWare, Deepgram and Memories.ai — running natively on Snapdragon X Series PCs. The pitch revolves around sustained local processing, privacy, lower cloud dependence and AI agents capable of handling multi-step work directly on PCs.
Qualcomm has been pushing the same direction elsewhere, from smartphones and PCs to wearables, industrial systems and personal AI devices. AIREV slots neatly into that strategy.
AI doesn’t necessarily have to live in a browser tab connected to a distant GPU cluster anymore. Increasingly, parts of it are moving onto the machine sitting beside you — or into the server rack down the hall.
What Happens Next Could Be More Important Than the Initial Deal
For now, Dragonwing is the headline. The longer-term language around the partnership is broader. Qualcomm and AIREV have left the door open to expanding OnDemand across additional Qualcomm technologies. Exactly how far that goes hasn’t been detailed yet. Still, the direction is hard to miss.
AI agents are moving away from being purely cloud-hosted software. Chipmakers want AI workloads running locally. Enterprises want more control over their data. Governments want sovereign infrastructure.
AIREV wants its agents sitting in the middle of all three. That could make this partnership much more interesting than another logo-meets-logo AI announcement.
