Bangladesh’s semiconductor success will depend on what it owns

Illustration: TBS

In an upper-floor office in Dhaka, a young engineer is at her desk at half past ten, long after the sun has gone down. The regression suite has been running since the afternoon; one block keeps failing timing, but she will, as usual, settle this before the morning call with the client. The part she is signing off will ship in tens of millions of units, and her name will appear nowhere on it, nor will her country's.

There is nothing wrong with that. It is how almost every engineer in this industry begins, and she is paid better than most of her graduating class. 

The question worth asking, days after the country adopted its first national semiconductor strategy, is what she will own in 15 years. That question sits beneath the debate currently flooding timelines and opinion pages: whether any of this is real.

The next logical frontier

When the National Semiconductor Symposium and BEAR Summit 2026 concluded late last month at the Novo Theatre in Dhaka, it left behind a strong sense of momentum. Its eight strategic frameworks, covering talent development, ecosystem building, global partnerships and diaspora engagement, set a rigorous foundation. 

The Prime Minister framed semiconductors as Bangladesh's next sovereign growth engine after ready-made garments. This alignment, reinforced by keynotes from academic pioneers and the Bangladesh Semiconductor Industry Association, marks a profound shift from the fragmented discussions of two years ago. The organisers deserve credit for institutionalising this ambition.

The framework rightly prioritises capability, partnerships, and talent pipelines. Now that these pillars are in place, the next test is who owns the intellectual property (IP) this ecosystem creates. Building talent first was necessary, but as teams mature, strategy must evolve towards retaining domestic IP assets. 

Without policy incentives for local ownership, global value chains will steer Bangladesh towards a service-heavy model. The aim should be to move local pioneers from back-office execution to real equity in the global market.

Two roads to the same place

The semiconductor landscape offers two entry points with different trade-offs. The first road is talent engineering: supplying design services to global corporations. India has taken this route for three decades, employing 20% of the global chip-design workforce, yet Invest India notes that very little of the IP generated is domestically owned. A Marvell India executive described the needed shift as moving from services to product ownership: designing one's own chips rather than supplying engineers so somebody else can design theirs. That shift must be engineered.

The key decision is how to design the promised cash incentive on design-service export earnings. As framed, it rewards billable volume - the more hours a firm sells, the more it receives, subsidising the service model India is now trying to escape.

The second road targets foreign direct investment (FDI) in advanced packaging and testing. Vietnam pursued it aggressively, attracting 170 high-tech FDI projects worth $11.6 billion, led by Amkor's Bac Ninh facility. Yet most revenue goes to foreign firms, while Vietnam has only 15,000 specialists against a target of 50,000 by 2030.

For Bangladesh, mimicking this FDI model unmodified is unrealistic. We cannot yet attract capital on that scale, meaning we risk accepting dependency without gaining the leverage that makes it viable. Costa Rica shows the fragility of this high-tech tenancy. When Intel arrived in 1997, it eventually drove 80% of tech exports. With the industry fully dependent on foreign presence rather than local ownership, Costa Rica was powerless when assembly shifted to Asia in 2014, wiping out 1,500 jobs overnight.

The terms of engagement

Both roads should move together, but on clearer terms: design services remain Bangladesh's only semiconductor foreign-exchange source, while foreign investment in packaging and testing would represent real progress. 

The goal is not to reject either model, but to recognise their shared limit: neither automatically creates domestic equity. That is why the partnership terms must be shaped now, while Bangladesh's footprint is still small enough for policy shifts to be economically and politically manageable.

Neither hype nor destiny

The louder debate has spilled beyond the press. On social media, and more carefully in print, some argue that Bangladesh's semiconductor ambition is manufactured excitement. Sceptics note that the foundations — talent, wage gaps and domestic markets, shared by a dozen countries — have not helped build their industries. That history matters: the transistor's co-inventor watched his own California firm fail, while a radio-repair shop in Tokyo was licensed to use the same technology to build Sony. Bangladesh's IT push, despite the hi-tech parks and expanded computer-science departments, never scaled into a 'second garments' industry after India absorbed the wage advantage it was counting on.

But that record does not make the dream hollow. It shows that precursors are necessary but insufficient: execution, not potential, is the issue. The hundredfold export target for 2030 is certainly very steep, and admitting it is not defeatism; it is the condition for spending the next five years on capability rather than announcements. Hype and fatalism make the same mistake: both skip the unglamorous middle. 

The sober view is that a real but narrow opening exists, and walking through it depends on decisions no summit will celebrate. Turning the task force into an audited national mission with a published scorecard would turn ambition into accountability; without it, ambition becomes theatre.

The arithmetic of ambition

Malaysia offers a useful model because it addressed this problem with instruments, not intentions. Its industrial master plans linked fiscal incentives to technology transfer, requiring multinationals to work with local suppliers to qualify for long-term tax benefits. The government and industry also founded the Penang Skills Development Centre instead of leaving training to universities alone. 50 years later, the machinery continues: in early 2026, the state allocated land worth about RM40 million for a shared-equipment campus pairing 14 multinationals with 18 local firms.

This is what supplier development looks like as a disciplined programme rather than a hope, and it shows the scale required. One Malaysian state committed about $9 million in land to one campus. By contrast, Bangladesh's Tk500 crore startup fund is the total pool, while individual grants are capped at Tk5 crore, about $400,000 short of a single mature-node tape-out once mask costs are counted. 

This is arithmetic, not criticism: limited capital should back two or three high-potential efforts rather than be scattered across a crowded room.

Beyond the billable hours

The key decision is how to design the promised cash incentive on design-service export earnings. As framed, it rewards billable volume — the more hours a firm sells, the more it receives, subsidising the service model India is now trying to escape.

Part of that incentive should instead reward ownership: IP registered and held by a Bangladeshi entity, or royalty and licensing income rather than service billings. The paperwork is no harder, but the signal changes: what the state pays for is what it gets.

Alongside that is something the state cannot legislate but the industry association can standardise. Design contracts often assign everything to the client, but reusable assets, verification environments, methodologies and non-client-specific building blocks need not be included. Keeping them costs nothing and is how service firms can build proprietary assets. Local firms negotiate poorly because they do so in isolation; a shared template and legal capacity would change that fairly quickly.

Terms for tomorrow

The national conversation keeps conflating two objectives. Bangladesh now earns roughly $8 million a year from design services, supported by around 1,200 engineers. That is a real starting point, but it is not the same as competing in a global semiconductor market projected to be worth about $1.51 trillion in 2026, driven largely by memory and AI compute — areas Bangladesh may never touch.

The near-term opportunity is embedded systems and electronics design: firmware, control software, board-level engineering and integration. This is high-margin work and a training ground for future chip designers. Chip-design revenue can grow separately as export services mature into licensable IP. These are two tracks on different timelines, and pretending otherwise serves neither.

Within five years, the engineer working on that upper floor will become very good at her job. Whether she is still doing someone else's work at 40 depends on decisions made in the next 18 months, in contract clauses and incentive schedules. Bangladesh has the strategy; now it needs the less glamorous work of ensuring that, if this industry succeeds, part of it belongs to those who built it.

Dr Sabbir Ahmad, a tech executive with global experience in digital connectivity, sustainable infrastructure and energy, now shaping Bangladesh's semiconductor landscape as the CEO of Silicon Array Ltd. Email: sabbir@ieee.org