The most expensive promise Prabowo Subianto made on his way to Indonesia’s presidency was lunch. His free-meals programme, budgeted at roughly $15bn to feed some 83 million children and pregnant women across an archipelago of thousands of islands, is the kind of undertaking that lives or dies on logistics.
Now Jakarta wants to point artificial intelligence at exactly that problem, embedding the technology into the meal scheme and a handful of other flagship programmes as it tries to make the machinery of the state work harder, says Reuters in an exclusive work.
The plan treats AI less as a moonshot than as plumbing. Indonesia’s national roadmap names a set of near-term, practical uses: tools to monitor the free nutritious-meal programme, models to forecast crop yields in support of the country’s food self-sufficiency drive, and systems to track financial reporting inside the Red-White cooperative initiative, another of Prabowo’s signature schemes.
The throughline is that these are not consumer products but back-office instruments aimed at waste, leakage, and the gap between what a programme promises and what actually reaches people.
That focus on a single, very large meal programme is telling. A scheme that size, run across remote districts, is precisely where money goes missing and food spoils before it arrives, and the government has already moved to refocus the rollout on more remote areas while trimming the pace of new kitchen construction.
An AI layer that can flag a kitchen falling behind, or a delivery that never lands, is a unglamorous use of the technology, but it is the kind that a finance ministry can actually measure.
The ambition sits inside a larger framework. Indonesia published a National AI Roadmap White Paper in 2025 and issued a presidential regulation directing the use of AI across public services, part of a push the government has tied to improving bureaucratic efficiency and easing a stubborn fiscal deficit.
The whole effort is yoked to Prabowo’s headline economic target of 8 per cent annual growth by the end of his term in 2029, and to the longer “Golden Indonesia 2045” vision of reaching high-income status by the country’s centenary.
There are obstacles the roadmap cannot wish away. Indonesia’s own AI regulation has slipped, with binding rules pushed into 2026 after the government missed an earlier target, which means the technology is being threaded into live programmes ahead of the legal framework meant to govern it.
That sequencing, deploying first and regulating after, is the opposite of the order most governments claim to prefer, and it raises obvious questions about data protection and accountability when the systems involved are touching welfare payments and food deliveries to tens of millions of people.
The country also has to build the underlying capacity, the data centres, the compute, the talent, in a region where AI infrastructure has become a contest in its own right, and where Indonesia is competing for the same chips, engineers, and cloud capacity as wealthier neighbours.
That race is one we have followed across Southeast Asia, from Thailand approving $29bn in data-centre projects to Kazakhstan’s $10bn deal with an Nvidia-backed builder, as governments try to avoid leaning entirely on a handful of US and Chinese clouds.
Indonesia’s twist is to lead with applications rather than infrastructure, putting AI to work inside programmes that already exist before the data centres behind them are fully built, an approach that risks running into the same sovereignty questions others are wrestling with.
What Jakarta is attempting is, in one sense, the least futuristic version of an AI strategy: not chatbots or frontier models but a state trying to use software to deliver lunch to 83 million people more reliably.
If it works, the proof will not be a demo. It will be a meal that arrives, in a village a long way from the capital, on a day it otherwise might not have.
Get the TNW newsletter #
Get the most important tech news in your inbox each week.