Bangladesh’s science and technology ministry used two events in Dhaka over the weekend to lay out what it wants from the country’s next phase of tech growth: dedicated artificial intelligence research hubs, a long-term push into chip design, and a business climate that makes private investors less nervous about putting money into AI.
Neither event produced a budget, a timeline or a named project. But together they give a clear picture of where officials and researchers think the gaps are, and the answer keeps coming back to three things: people, investment and electricity.
Investors first, then research
On Saturday, science and technology secretary Md Anwar Hossain spoke at a seminar on moving Bangladesh from digital infrastructure to an AI economy, organised by the American Chamber of Commerce in Bangladesh. His main point was that the government’s first job is to lower the risks private companies face and protect what they invest. He said policy interventions are needed to remove barriers that hold back business growth and innovation.
The ministry’s plans, as described at the seminar, include joint research between industry and universities, an incentive framework to reward that collaboration, and investment in the infrastructure needed for wider AI use. The panel, moderated by Oracle’s country head Rubaba Dowla, brought together Grameenphone chief executive Yasir Azman, Visa’s Bangladesh country manager Sabbir Ahmed, ShopUp co-founder Atau Rahim Chowdhury and Dhaka University computer science chair Professor Md Abdur Razzaq. Discussion centred on how AI could be used in financial services, telecoms and e-commerce, the sectors where Bangladesh already has scale.
A day later, Hossain was chief guest at a roundtable titled “Rooppur and the Fourth Industrial Revolution”, organised by StarTech Tribune, the science communication arm of the Bangladesh Council of Scientific and Industrial Research (BCSIR). There he urged researchers to stop working in silos and build domestic AI research hubs, pointing to uses in biotechnology, personalised medicine and cancer drug discovery. He said close to 100,000 science graduates finish university every year, and argued that curricula should be updated to steer more of them into nuclear science, advanced robotics and applied computing.
Purdue University professor Muhammad Mustafa Hussain called for long-term roadmaps for indigenous chip design and, eventually, local fabrication. Professor Risala Tasin Khan, director of Jahangirnagar University’s Institute of Information Technology, warned that training deep learning models needs an uninterrupted grid, since even brief micro-outages can corrupt costly datasets and damage hardware. A Rooppur Nuclear Power Plant official argued that data centres, robot manufacturing and large server farms need the kind of constant baseload supply that nuclear generation is designed to provide. BCSIR chair Dr Samina Ahmed said Rooppur “will serve as a generational catalyst for national industrial capability”.
What it means here
The message lines up with the government’s broader push, including its target of a million digital jobs and work on a national AI policy. The harder part is turning speeches into budgets. Chip fabrication plants cost many billions of dollars and take years to build, so design work, testing and packaging are the more realistic near-term entry points for Bangladesh, and they rely on the same engineering graduates the ministry wants to retrain.
The power argument is the most practical one. Anyone planning a data centre, a cloud service or a research lab in Bangladesh asks first about grid reliability, and recent disruptions have not helped that case. If Rooppur’s output does arrive steadily, it could matter more to AI ambitions than any new policy document. Until then, the investment-friendly climate officials describe will be judged on concrete steps: clear rules, predictable taxes and power that stays on.
