Indonesia is likely to remain predominantly a consumer and adopter of foreign-built artificial intelligence (AI) over the next several years, rather than a producer of frontier AI capabilities, according to BMI Country Risk and Industry Research said on last Friday.
The research firm said in a report that Indonesia’s ambition to become a “global AI player” remains aspirational, with the country still dependent on foreign hyperscalers, imported advanced chips and foundation models.
A shallow domestic startup pipeline, persistent digital-talent shortages and nascent sovereign AI initiatives are also limiting its ability to develop frontier capabilities.
AI adoption is expected to continue rising, supported by Indonesia’s young and digitally connected population, growing hyperscaler investment and government policy momentum.
However, BMI said the widening gap in AI diffusion between developed and emerging markets, uneven connectivity across the archipelago and shortages of skilled workers will constrain the pace and distribution of adoption.
Connectivity remains a structural challenge. Indonesia had about 212 million internet users, equivalent to roughly 66.5 percent of the population, as of late 2025.
However, its geography of more than 17,000 islands makes nationwide fiber and mobile coverage costly and difficult, leaving adoption concentrated in Java and major urban centers. Nationwide 5G coverage is not expected until 2029.
BMI also identified power availability as the key constraint on Indonesia’s AI and data center expansion.
Indonesia’s data centre market had 544MW of live capacity, 495MW under construction and a planned pipeline of 2.5GW as of the third quarter of 2026. Jakarta remains the dominant hub, with about 340MW of live capacity and a 1.7GW pipeline, while Batam has a pipeline exceeding 400MW.
However, coal accounts for about 60 percent-65 percent of Indonesia’s electricity generation, creating challenges for hyperscalers with net-zero commitments. Grid interconnection queues, competition for power and the lack of renewable transmission infrastructure could increasingly slow capacity deployment and deter environmentally conscious investors.
Foreign investment is expected to remain a major driver of Indonesia’s AI infrastructure. Microsoft has pledged $1.7 billion over four years for cloud and AI infrastructure, alongside training for 840,000 people. Nvidia and Indosat have committed $200 million to an AI center in Surakarta, while Tencent has pledged a further $500 million.
Domestic efforts are also emerging. Indonesia plans to establish a Sovereign AI Fund, expected to launch between 2027 and 2029 under a public-private financing model and managed primarily by state investment manager Danantara Indonesia.
Fiscal incentives include tax holidays of five to 20 years for pioneering industries, including AI and data centers, and a 300 percent super tax deduction for research and development.
BMI estimates Indonesia will require about US$3.2 billion by 2030 to meet national computing needs, along with another $968 million to train 400,000 AI professionals.
Meanwhile, cloud computing will remain the main delivery layer for AI, with spending forecast at $7 billion in 2026.
BMI expects growth to be supported by hyperscaler investments designed to meet data-localization requirements and provide low-latency connectivity.
The localization of government and financial data is also creating a structural source of demand for domestic data center capacity.
Oracle launched its Indonesia North Cloud Region in Batam in July 2025, while BMI expects Batam to benefit from demand spilling over from Singapore, where data center capacity remains constrained.
AI adoption is developing across several sectors, although maturity varies considerably. Banking, financial services and insurance (BFSI) is among the most advanced sectors, with institutions deploying AI for customer service, financial analysis and product development.
Healthcare is also showing early benefits, with a 2026 survey by Royal Philips finding that 88 percent of Indonesian clinicians reported improved workflow efficiency from AI, while 87 percent reported faster access to diagnostic results.
Other sectors remain at an earlier stage. AI applications in agriculture include precision farming, disease detection and resource optimization, but many reported results are based on vendor case studies and have yet to be independently verified or demonstrated at scale.
Logistics and infrastructure companies are using AI for route optimization and predictive maintenance, while oil and gas firms are exploring applications in upstream operations and enhanced oil recovery.
Indonesia’s domestic AI vendor landscape is concentrated mainly at the application layer, where local-language capabilities and knowledge of the domestic market provide an advantage.
Companies such as Nodeflux, Kata.ai, Prosa.ai and Verihubs are developing applications in areas including computer vision, natural-language processing and biometric verification.
At the underlying compute and cloud layers, however, the market remains dominated by foreign companies. NVIDIA, AMD and Intel supply advanced chips, while Microsoft, Google and Amazon Web Services provide much of the cloud infrastructure.
BMI said Indonesia’s sovereign AI efforts could gradually reduce dependence at the model and application layers.
Sahabat-AI, developed through collaboration between Indosat Ooredoo Hutchison, GoTo, NVIDIA and AI Singapore, is an open-source large language model trained on Indonesian data and dialects, including Bahasa Indonesia, Javanese, Sundanese, Balinese and Batak.
Despite these initiatives, Indonesia remains dependent on imported GPUs and foreign technology at the infrastructure level. Bahasa Indonesia itself is classified as a low-resource language in AI training data despite being spoken by around 270 million people, limiting the availability of digital text, labelled datasets and translation material for model development.
On regulation, Indonesia does not yet have dedicated, binding AI legislation and continues to rely primarily on the Information and Electronic Transactions Law and Personal Data Protection Law. The government is preparing regulations covering a 2026-2029 National AI Roadmap and AI ethics.
BMI said delays in formalizing the regulations represent a key governance risk. The draft framework is expected to adopt a risk-based approach, with greater oversight for applications involving public safety, personal data and fundamental rights.
The credibility and timing of the regulatory framework will be an important factor for future investment, particularly as AI is already being deployed in government programs before the new rules are formally in force.
Overall, BMI expects AI adoption in Indonesia to continue expanding over the next five years, supported by demographics, cloud investment and policy momentum. However, infrastructure, power, talent and connectivity constraints mean the country is likely to remain primarily an AI adoption market rather than a global producer of frontier AI capabilities in the near term.
Indonesia risks missing AI boom without shift into tech manufacturing – AMRO