HANOI, 29 AUG 2026 — Vietnam's leadership has pressed Qualcomm and Samsung to deepen their investment in artificial intelligence, semiconductors and research. Qualcomm's chief executive said he wants Vietnam to become the company's third-largest AI research and development hub globally.
Officials noted that cumulative exports from Samsung's Vietnamese handset operations have reached US$500bn. That figure is being repeated without the word cumulative, and it matters.
Five hundred billion is a total, not a year
Vietnam's entire annual goods exports run to a few hundred billion dollars. A single company's phone factories cannot be exporting US$500bn a year from the country, and nobody has claimed they are.
The figure is the running total since Samsung began manufacturing there, reached at the end of June after roughly three decades. The number is remarkable, and it describes history rather than current output.
The distinction is not pedantry when the number is being used to argue about Vietnam's present position. Cumulative totals grow by construction and never fall, which is what makes them good material for an announcement. On the question of whether a relationship is expanding or contracting right now, they say nothing at all.
The ask has changed from assembly to design
What Vietnam requested at this meeting is different in kind from what it has received for twenty years.
Samsung's presence has been manufacturing: enormous facilities assembling handsets for export, employing very large numbers of people, and generating the export figure above. That is real economic development and it is the model Vietnam executed better than almost anyone.
Artificial intelligence research, semiconductor design and robotics are a different proposition, employing far fewer people at much higher value. This work depends on doctoral-level engineers rather than trained assembly workers, and it requires an intellectual property environment that a manufacturer's contract does not.
Moving from one to the other is the hardest transition in industrial development, and the countries that have managed it did so over decades with heavy investment in universities. What happened in Hanoi was a statement of intent from Vietnam's side, not evidence that the transition is under way.
Export value is not value captured
The export figure also overstates what Vietnam has gained for a structural reason rather than an arithmetical one.
When a phone is assembled in Vietnam from imported components and shipped abroad, its full retail-adjacent value counts as a Vietnamese export. What the country actually captures is the assembly margin: wages, local services, land, tax. The processors, the memory, the displays and the software were made and priced elsewhere, and most of the value went with them.
This is the well-documented shape of assembly-led development. Gross export figures are enormous while the domestic value added is a fraction of the total, which is why countries that succeed at assembly then spend decades trying to move upstream.
The US$500bn figure and the request for research investment belong to the same argument. One measures how much Vietnam has manufactured over thirty years. The other exists because so little of that value stayed in the country.
Third-largest by what measure
Qualcomm's ambition deserves the same reading as any superlative in an announcement.
Third-largest AI research hub is a ranking with no denominator. Engineer headcount, capital deployed, patents filed or projects owned end to end would each produce a different ordering, and a site can rank third on the first measure and be modest on the last.
It is also an aspiration voiced at a meeting with a head of state rather than a commitment with a figure and a date attached. Qualcomm already operates an AI research centre in Vietnam, so the direction is real. The magnitude is not yet a number anyone can check.
Vietnam legislated before it built
What makes Vietnam's position unusual is that the regulatory scaffolding arrived ahead of the industry it governs.
The country's AI law came into force in March, with grace periods running into 2027, making it one of the few jurisdictions anywhere with a comprehensive dedicated framework. It has also set out phased sovereign data centre capacity starting at 100MW.
Most countries build first and regulate afterwards, usually badly and late. Doing it in the other order is a coherent strategy for a state that wants investment on terms it has already defined, and it carries an obvious risk: rules written before an industry exists are written without the experience that would have made them practical.
Which of those it turns out to be will be settled by whether firms like Qualcomm expand under those rules or quietly route the work elsewhere.
The competition is regional and specific
Vietnam is not bidding against abstract alternatives. It is bidding against Malaysia, Singapore and increasingly Indonesia for the same engineering investment.
Each offers a different package. Singapore offers talent density and a settled legal system, at a cost that reflects both. Malaysia already runs semiconductor assembly and test at scale, on land and power Singapore does not have. What Vietnam brings is a manufacturing relationship with these same firms that is already twenty years deep, at a wage bill neither of the others can match.
The determining factor for research work, as opposed to manufacturing, tends to be the supply of senior engineers who can lead projects rather than execute them. That is the constraint every country in this group is trying to relieve, and it is the slowest one to fix, because it is produced by universities over a decade rather than by incentives over a budget cycle.
What it means from here
For businesses in the region the useful signal is that the competition for high-value technology work has moved from cost to capability, and Vietnam is arguing it has both.
If that argument succeeds, the practical consequence is regional salary pressure on exactly the roles that are already hardest to fill — machine learning engineers, chip designers, systems people. A Qualcomm research centre scaling in Ho Chi Minh City or Hanoi bids for the same people a Singapore or Kuala Lumpur employer wants, and it does so with a multinational's compensation band.
The figure to watch is not the cumulative export total or the ranking, but whether any of these companies announces a headcount or capital number for research in Vietnam within the next year. Ambitions get stated at meetings. Hiring requires a budget line, and no budget line has been announced.
