Asia is becoming a hub for telco-powered ‘AI factories’
Operators in Asia are retrofitting existing fiber infrastructure and data centers to develop AI sovereign capabilities that empower national security and government technology initiatives.
Asia is becoming a global hub for telco-powered AI infrastructure. Operators in the region are rapidly building “AI factories” designed to promote data sovereignty, embed AI into the telecoms’ infrastructure with a longer-term goal of better monetizing edge applications (and with an eye on incorporating AI into future 6G networks), and strengthen national security.
Operators are doing this by retrofitting local existing fiber and data centers, and partnering with hyperscaler providers such as AWS and Azure, to create a combined infrastructure that reduces cost volatility and helps organizations to meet compliance obligations, in addition to the benefits mentioned above.
The concept of AI sovereignty is born out of a growing need for national security. Where countries adopt proprietary AI, this is, in part, to limit dependence on other countries’ AI models.
AI is becoming a hotbed for telecom-powered AI factories
Asia is a hotbed for this kind of thinking. Some of the pioneers of this rapid build-out of AI factories in the region include operators, such as Singapore’s Singtel (RE:AI – the company’s cloud-based AI infrastructure arm), Indonesia’s Indosat Ooredoo Hutchison, Vietnam’s FPT, Malaysia’s YTL, India’s Tata Communications, and SK Telecom in South Korea.
Such companies are building AI platforms tailored to local needs with many being driven by national data sovereignty policies. These moves are transforming operators in the region from connectivity providers into vital AI sovereignty enablers.
According to a report from DSP Leaders from TelecomTV, entitled “Trends in Telco AI Infrastructure,” an AI factory is “a datacentre facility optimised for AI workloads, from data ingestion to training, fine-tuning and AI inference.” This requires specialized infrastructure, particularly graphics processing units (GPUs).
California-based NVIDIA, a leading provider of AI GPUs, is highly active in the region. It claims 18 telco AI factory partnerships, including India, Indonesia, Japan, Malaysia, Singapore, Vietnam, and Kazakhstan.
Singtel also launched a specific arm for its data center business, named Nxera, in 2024. Singtel’s Digital InfraCo unit is collaborating with NVIDIA to meet Singapore’s National AI Strategy 2.0. It allows customers to access Singtel’s extensive local fixed broadband, submarine, and 5G network capabilities.
Asia’s regional operators are reinventing themselves as AI enablers
South Korea’s SK Telecom has also set up a new data center division to help develop its AI infrastructure superhighway. The plan is based on three core pillars: AI data centers (AIDCs); a cloud-based GPU service (GPU-as-a-Service, GPUaaS); and Edge AI. The operator is working with NVIDIA, but also adopting SK Hynix’s high-bandwidth memory chips into its infrastructure, with the aim of extending its infrastructure globally.
Japan’s SoftBank and KDDI have both recently – separately – announced plans to create large data centers and to create Japanese-speaking large language models (LLMs) to ensure AI sovereignty for the country.
Indonesia’s Indosat Ooredoo Hutchison, meanwhile, has established the country’s first sovereign AI factory, while YTL has established its AI Cloud, which is aimed at developing a sovereign telco AI factory to support its government’s National AI Strategy. YTL’s AI Cloud supports Malaysia’s digital economy ambitions, while FPT has launched AI factories in Vietnam and Japan. The list goes on.
Telecoms-led AI sovereignty is becoming a significant topic, and Asia seems to be ahead of the game.
For more information on AI data centers, download our paper here, or contact us to find out how we can help you make the opportunities available in China and Asia, while minimizing the risk.
