AI cloud provider GMI Cloud is partnering with AI transformation company Magna AI on a sovereign AI factory initiative.

Under an agreement revealed on June 11, the two aim to jointly architect and deploy a global network of sovereign AI factories for governments, enterprises, and national AI programs.

Elsewhere, Singtel and WEKA are teaming up on sovereign AI infrastructure across ASEAN and Singapore.

The first deployments are planned for Malaysia, Belgium, and Romania, with the companies then planning to expand across the Middle East and Africa.

The 'AI factories' will be built using Nvidia Vera Rubin NVL72 architecture.

"Governments increasingly recognize AI infrastructure as a strategic national asset," said Alex Yeh, CEO of GMI Cloud. "Our mission is to help nations design, build, and operate sovereign AI platforms that enable innovation, security, and long-term competitiveness. We believe sovereign AI will become a foundational pillar of future economic development."

"Sovereign AI is becoming a foundation for future economic and technological competitiveness," added Dr. Moataz BiAali, CEO of Magna AI. "Together with GMI Cloud, we are enabling countries to accelerate AI adoption through secure, resilient, and scalable sovereign AI infrastructure."

The companies have begun joint planning activities covering site evaluation, infrastructure architecture, compliance requirements, and phased deployment strategies, with groundbreaking activities expected to begin later this year.

Earlier this year, GMI Cloud announced plans for a $12 billion sovereign AI infrastructure initiative in Japan, which the company said will see the launch of an AI Factory for "large-scale physical AI purposes" in Kagoshima, Japan, in partnership with Wistron.

Meanwhile, in Singapore, Singtel Digital InfraCo's AI cloud business RE:AI has announced it will work with WEKA to build and deliver sovereign AI infrastructure solutions across Singapore and the broader ASEAN region.

The two will combine RE:AI's AI cloud infrastructure and network orchestration platform with WEKA’s NeuralMesh, a high-performance, software-defined storage system. The two aim to remove storage bottlenecks and reduce idle GPU time.

Singtel will also integrate NeuralMesh into its GPUaaS, which serves governments and enterprises.

Bill Chang, CEO of Singtel Digital InfraCo, said, “As sovereign AI adoption accelerates across ASEAN, governments and enterprises require infrastructure that is secure, compliant, and built to perform at scale. Modern AI is no longer limited by compute power alone. The real bottleneck is how quickly data can reach GPUs, making high-speed data architecture just as critical as the GPUs themselves.

"By partnering with WEKA, we are addressing this challenge and strengthening our sovereign AI platform by bringing together trusted cloud infrastructure, advanced data capabilities, and network intelligence, so that customers can deploy complex AI workloads with confidence.”

The two will offer customers access via three models: a Dedicated Sovereign Pod for single-tenant environments supporting highly sensitive workloads; the Regulated Multi-Tenant Sovereign Zone, which provides isolated, shared environments for regulated industries; and the Hybrid Sovereign Model, which integrates on-premises systems with sovereign cloud capacity.

RE:AI teamed up with Mistral AI earlier this year, aiming to increase sovereign AI capabilities in Singapore. At the time, RE:AI said it would work with Mistral to develop an "advanced AI infrastructure" using Mistral AI's stack to provide a multi-tenant GPU-as-a-Service and AI-as-a-Service offering.

Singtel's Nxera launched a data center in Singapore - DC Tuas - earlier this year. The data center has a capacity of 58MW and is the company's largest facility in the country.

Although Singtel is closing a number of small legacy data centers in Singapore, it still operates several large facilities in the city-state through its Nxera unit.