The announcement from Ooredoo came with a number that has become the new unit of ambition in the global AI race: one gigawatt. The Qatari telecommunications group said it will build an AI platform of that scale in Indonesia, in partnership with Nvidia and Nokia, bringing the kind of computing capacity once reserved for the world’s largest data-center operators to the largest market in Southeast Asia. The project is the latest in a string of sovereign AI builds that have stretched from the Gulf to the Pacific.
A gigawatt of AI compute is a statement of intent. Data centers of that scale require dedicated power plants, extensive cooling, and supply chains that span continents, and few organizations in the world have built even one. The fact that national governments and regional operators are now measuring their ambitions in gigawatts says everything about how the AI buildout has changed the scale of the computing industry. Two years ago, such a project would have been unthinkable outside the largest hyperscalers; now it is a chapter in a regional arms race.
For Nvidia, the deal is another brick in a global distribution strategy. The company has spent the past year selling not just chips but ecosystems, pairing its GPUs with the networking, software and expertise needed to stand up a sovereign AI stack. Indonesia, with more than 270 million people and a rapidly digitizing economy, is the most populous market in Southeast Asia, and the project gives Nvidia a flagship anchor in a region where the competition for AI infrastructure is intensifying. Every gigawatt platform built with Nvidia’s kit deepens the moat around its market position.
The sovereign AI framing matters as much as the hardware. Governments across the Middle East and Asia have concluded that AI capacity is strategic infrastructure, like ports or power grids, and they are spending accordingly. The term has become a shorthand for a simple argument: nations that own their compute will own their AI future, and nations that rent it will be dependent. Nvidia has embraced the framing enthusiastically, arguing that its platforms let countries build capacity without ceding control. The Indonesia project is the latest proof that the pitch is working.
Ooredoo’s role in the project shows how the telecom industry is positioning itself. Carriers have watched their traditional businesses mature, and many have concluded that data centers and AI infrastructure are the growth market. Ooredoo, which operates across the Middle East, North Africa and Southeast Asia, has been among the most aggressive, and the Indonesia platform gives it a flagship beyond its home region. Nokia’s involvement, providing the networking layer, completes the picture: the project is a full-stack build, with each partner supplying the piece it knows best.
The economics of gigawatt-scale AI are still being written. The capital costs are enormous, the power requirements strain grids, and the revenue models depend on demand forecasts that few have validated at this scale. Analysts who follow the sector caution that announced capacity and operating capacity are different things, and that some projects in the sovereign AI wave will stall at the permitting or power-procurement stage. The winners, they say, will be the regions with reliable electricity, stable regulation and the workforce to run the machines.
The competitive implications extend beyond Nvidia. The spread of AI platforms to Southeast Asia shifts the map of global compute, which has historically been concentrated in the United States, Europe and parts of China. Countries that build their own capacity gain bargaining power in their dealings with the technology giants, and the resulting infrastructure becomes an asset that outlasts any single hardware cycle. The Indonesia project, if it reaches full operation, would make the region a more significant node in the global network of AI infrastructure, with consequences for everyone from chip suppliers to cloud providers.
The scale of the Indonesia project invites comparison with the largest builds elsewhere. In the United States, the biggest AI campuses are measured in the hundreds of megawatts, and the gigawatt threshold has been crossed only by a handful of hyperscalers with dedicated power contracts. A project of that size in Indonesia will face its own arithmetic: the country’s grid is expanding, its power sector is being opened to private investment, and its government has made clear it wants to host the region’s AI capacity rather than import it. Those conditions, investors said, are what made the partnership credible enough to announce.
The competitive map of Southeast Asia is taking shape around such projects. Singapore has long been the region’s data-center hub, constrained by land and power; Malaysia has courted hyperscale builds with incentives; and now Indonesia is entering with a gigawatt-scale plan of its own. Each country’s pitch to AI investors differs, but the underlying contest is the same: who will host the compute, capture the jobs and control the data? The Ooredoo-Nvidia-Nokia project gives Indonesia a strong opening bid in that contest.
For now, the announcement is a plan with partners, a scale and a location. The gigawatt figure sets the ambition, the partnership sets the execution path, and Indonesia sets the stakes. Nvidia’s strategy of spreading its ecosystem into every market with the means and the will to build has found another convert, and the map of where AI compute lives has another marker. The race to build gigawatts has moved to Southeast Asia, and the companies that build the machines are already there.
