Is there room for anyone other than the US and China in the global AI race? Vikram Sinha, CEO of Indonesia’s second-largest mobile operator Indosat Old Hutchison (IOH), believes there should be.
“What works out in the United States or China may not work in Indonesia,” he said. luck In early April, he pointed out the differences in culture and language in this country. That leaves room for companies like Indosat. “We are in pole position in terms of how we can deliver connectivity and computing, or intelligence, to millions of people around the world in a sovereign way,” he continued.
Sovereign AI has become a buzzword for almost every government concerned about leaving the AI field solely to US and Chinese research institutes such as OpenAI, DeepSeek, and Moonshot AI.
Sinha is betting that the next stage of AI, running models closer to the end user and responding to local problems in local languages, will belong to telcos like Indosat in the so-called Global South. Indosat’s CEO, who came to Indonesia after working in India, Seychelles and Myanmar, is keen to drive that development through Sahabat AI, the country’s platform for startups backed by Indonesia’s large-scale language models, which he claims can avoid the blind spots of models trained in the US or China.
Still, even Sinha doubted whether “sovereignty” could be turned into a business. “If you ask my team if they can make a business case for Sahabbat, they don’t know how to do it,” he admitted.
From India to Indonesia via Yangon
Born in Jamshedpur in eastern India, Sinha entered the telecom industry in 2005 when he joined Bharti Airtel. Seven years later, the company sent him, then just 37, to lead operations in Seychelles, a small island nation of just 120,000 people off the east coast of Africa. He then moved to another island nation, heading up Oredoo’s operations in the Maldives, and then to Myanmar, just as the Southeast Asian nation was in the midst of an (ultimately short-lived) democratization and opening-up.
What Sinha remembered from his time in Myanmar was the average age of the team. They were all 27-year-olds, in his words, “young players.” Still, he found his time in the country a rewarding experience. “When I was in Myanmar, people warned me about the capability gap,” he said. “But if you invest in bringing out the best in people, you will find a lot of talent.”
In 2021, Ooredoo appointed Sinha to lead the newly formed IOH, which was formed through the merger of Indosat and Hutchison 3 Indonesia, owned by Hong Kong-based conglomerate CK Hutchison. Ooredoo and CK Hutchison together own 65.6% of IOH’s shares. The Indonesian government holds a 9.6% stake, a holdover from Indosat’s earlier state-owned enterprise.
Most mergers end in disappointment, with consulting firm McKinsey estimating that up to 70% of such mergers fail to deliver on their promises. (Sinha puts this figure even higher, claiming that 95% of telecom mergers fail). However, Indosat is an exception, as the company continues to grow its revenue, profits, and user base after the deal.
“The most important guiding principle we wanted to follow was to look at the merger from a maximization perspective rather than an optimization perspective. How do we make one plus one equal 11?” Sinha explains. “When investors and analysts look at mergers, they only talk about synergies, but employees and customers don’t care about that. They care about growth and experience.”
outperform a declining market
Indosat reported that its revenue in 2025 will increase by 1.1% year-on-year to 56.5 trillion Indonesian rupiah ($3.3 billion) and profit will rise 12.2% to 5.5 trillion rupiah ($320 million). But these numbers mask a tough year. Sinha points out that the company’s performance was weak in the first half, but things only improved in the second half.
This strong performance continued into the first quarter of 2026, with sales increasing 12.1% year-on-year. (Indosat announced its first quarter results on April 29th. luckConversation with Sinha). Indosat also achieved average revenue per user (ARPU) of Rp 45,000 ($2.59), the highest since the merger.
During an earnings call for analysts, Sinha highlighted Indosat’s new partnership with Google to bring the US tech company’s Gemini AI product to users. “We believe there is even greater opportunity for upside in ARPU,” Sinha told analysts.
Still, Indosat’s stock price is down 9% for the year. That’s still better than the broader market, which has been depressed for months over fears of a downgrade to “frontier market” status. (The Jakarta Composite Index has fallen 17% since the beginning of the year)
Indonesia’s technology sector has been in a state of prolonged turmoil. Investors were once enthused by the country’s potential to serve hundreds of millions of young, upwardly mobile and digitally savvy Indonesians. After that, that optimism disappeared. “The problem was that there were a lot of unicorn startups,” Sinha said. “They were following the wrong metrics. They were all on the ratings game.”
“That mindset has to change. We need to build our business around a more sustainable model with something more pragmatic,” he added.
Building an AI stack
Indosat taps into every layer of Jensen Huang’s “AI layer cake,” the framework Nvidia’s CEO uses to describe the layers of AI infrastructure, from energy and chips to infrastructure, models, and ultimately applications. The company collaborates with Nvidia to offer GPU-as-a-service, providing on-demand processing power to businesses in Indonesia. Indosat’s AI factory is centered around a cluster of Nvidia’s H100 processors and is already attracting customers across banking and mining.
“Inference needs to happen close to the edge,” Sinha said, referring to deploying AI models closer to end users rather than in centralized data centers. “Telecoms like us can bring intelligence to the edge with low latency and develop applications for a country that are made in that country, rather than just buying them from China or the United States.”
Sinha argued that countries like Indonesia have structural advantages that Western countries do not have. “A country like Indonesia has electricity, land and water. Indonesia currently has nearly 800 megawatts of approved electricity,” he said. “America has no power.”
Still, he acknowledges that raw infrastructure alone is not enough. “Without human capital, you can never be sovereign,” Sinha said. “Sovereignty is not just about investments and money.”
your friendly AI
Sahabat AI is at the heart of Indosat’s AI strategy. This open source, large-scale language model was developed in collaboration with Indonesian ride-hailing and tech giant GoTo and is built around Indonesian languages such as Indonesian and Batak. (“Sahabat” means “best friend” in Bahasa.)
For locally built models, this is easier, at least in principle. “The LLM is not neutral, and if it’s not written in your language, it’s going to have biases, cultural nuances, etc.,” says Sinha. “Every country will focus on protecting data and cultural sovereignty.”
Several other countries are trying to build their own local models. South Korea’s Naver has developed a Korean model, and AI Singapore’s SEA-LION initiative has built a family of open-source models for 11 Southeast Asian languages, including Indonesian, based on models from Meta, Google, and Alibaba.
There are also practical reasons that go beyond principled reasons. “The ability of governments, banks, and other regulated entities to use AI depends on accuracy,” explains Pak-Sun Ting, co-founder of Hong Kong-based startup Votee AI, which has developed LLM that works in the Chinese dialect of Cantonese. “If there is no precision and people are speaking a language that the model cannot understand, there is no use case.”
However, it is difficult to build models in languages for which there is not a lot of documentation. “Low resources, by definition, means there is not enough data to build large language models with natural language processing,” Ting says. This designation has nothing to do with the number of speakers (Indonesian is spoken by around 300 million people), but rather with the amount of digitized text (which is small for most languages).
For Mr. Sinha, Sahabhat looks more like a public service than a business, at least initially. He called it a “platform for innovation and cooperation” and said it contributed to strengthening new AI startups in Indonesia. “We haven’t promoted this in a way that increases daily or monthly users,” he said.
“We are very confident that a business case will emerge. But certainly at this early stage there will be questions,” he acknowledged. “You have to give it your all and believe in it.”
In Fortune’s twice-monthly Asia Agenda column, we speak to Asia’s top business leaders about how they’re building for the future and the lessons they’ve learned from leading companies in one of the world’s fastest-growing and most dynamic regions. See our full profile here.
