Asia's AI race doesn't have a single playbook. One of its stranger developments is that telecoms, historically the pipes rather than the brains of the digital economy, are increasingly the ones being asked to carry national AI ambitions.

In Indonesia, that conversation involves government agencies, technology partners, industry leaders, universities, and a growing ecosystem of startups. Among other leaders, Vikram Sinha, CEO of Indonesia's Indosat Ooredoo Hutchison, has been the most vocal about what's actually at stake. At MWC 2026 he said it plainly: "Our purpose is to empower Indonesia." That framing, a telco defining itself as a nation-builder, would have sounded odd five years ago.  

Across Asia today, it's becoming the norm. Walk across the region and you'll find three distinct models for how countries are getting this done. The differences aren't just structural. They reflect something deeper about how each nation decides who is responsible for its technological future.

Singapore answers that question with the state. IMDA has been running AI governance since 2019, when it became one of the first governments in the world to publish a formal AI governance framework. Since then, the architecture has only grown. Singapore committed over $1 billion to AI compute, talent, and industry development under its National AI Strategy 2.0. The government runs its own AI safety red-teaming exercises, hosts multilateral governance forums, and co-develops sector-specific AI tools with professional bodies, from legal LLMs for lawyers to GenAI playbooks for SMEs. For a city-state with no natural resources and a population smaller than many Southeast Asian provinces, the logic is almost existential: the government has to be the anchor, because no single company is large enough to carry the weight.

Malaysia has taken a different path, and it's worth watching closely. The state sets the direction. PM Anwar has made "AI nation by 2030" a recurring commitment, and Budget 2026 allocated RM5.9 billion for AI and data center infrastructure. But the actual building is happening in the private sector. YTL Power's partnership with Nvidia to develop Malaysia's first operational Blackwell-powered AI data center in Johor cost RM10 billion and sits on 1,640 acres. YTL also built ILMU, Malaysia's first sovereign large language model. The government created the conditions, through the New Industrial Master Plan 2030 and ministerial presence at deal signings, and then let industry move. Two hundred local startups are already plugged into YTL's APIs building their own applications. That's a productive arrangement, but it works when you have a private actor willing and able to take on the infrastructure burden at that scale.

The third model is the one South Korea has bet on: picking national champions and making them compete. The Ministry of Science and ICT put $381 million behind five AI consortia and structured it as a tournament. By 2027, only two survive. SK Telecom, the telco in the group, is arguably the most interesting case study. It has reorganized its entire company structure around AI, with seven divisions, a 519-billion-parameter foundation model called A.X K1, and an "AI Infrastructure Superhighway" plan involving gigawatt-scale data centers and GPU-as-a-service. It won a government GPU-leasing contract this year. It leads Korea's K-AI Alliance with 30 AI startups. It's not a telco that dabbles in AI. It's an AI company built on top of telco infrastructure, and it's explicitly framed as national infrastructure, not just a commercial play.

What connects these three models isn't the mechanism but the underlying conviction that AI capability is too important to be incidental. Whether it's Singapore's government driving from the front, Malaysia's productive handshake between state and industry, or South Korea's competitive forcing function, each country has made an active choice about who carries the responsibility.

That's the question Indonesia is answering, and it isn't answering it alone. With the likes of Telkomsel, Gojek and others, Indosat Ooredoo Hutchison, with over 100 million subscribers, Nvidia cloud partnerships, the Sahabat-AI open-source LLM built for Bahasa Indonesia, and a national AI Center of Excellence launched with Nvidia and Cisco, is building toward the same position SK Telecom occupies in Korea. The infrastructure is real. The ambition, per CEO Vikram Sinha, is to be an "AI nation shaper," not just a connectivity provider, part of its three pillar strategy — AI-native telco, sovereign AI TechCo, and nation shaper, each one a layer further from the connectivity business Indosat started as.  

Indonesia is a country of 280 million people spread across 17,000 islands. Whether its AI future is driven by the state, by industry, or by some productive combination of both, it needs someone willing to carry the weight. Indosat has decided that someone is them.