Why ‘cheaper’ no longer wins development deals for Vietnam

I’M GOING to say something that will probably annoy some people: Vietnam’s “we’re cheaper” software pitch is running out of road.

I started thinking seriously about this a few months ago, while I was listening to a company pitch for a software development contract. The firm opened with the hourly rate before telling me about the team, its engineering capability, or what it could actually deliver.

As someone who hss spent 18 years running an outsourcing business (my company, Saigon Technology, now has more than 350 engineers), I have sat through a lot of these pitches. But this one stuck with me.

The rate was roughly US$20 to US$25 an hour for a junior developer. That would have sounded perfectly reasonable 10 years ago, and maybe it still is to some today.

In an era when AI can spit out code practically for free, however, it is not.

Vietnam did not suddenly become expensive, and its engineers did not become less capable out of the blue. It is just that the thing we spent the last decade selling – cheap engineering labour – is becoming less differentiated, and AI is accelerating that change.

The old math is breaking down

Software outsourcing has traditionally been a fairly simple calculation.

If a developer in the US costs US$100 an hour and a developer in Vietnam costs US$25, the client saves 75 per cent. Multiply that across a team and a year, and the business case is obvious.

Vietnam has built a very successful industry around this model. Now the math is being rewritten.

SEE ALSO

Vietnam pitches its real economy, not a Singapore replica, in two-city IFC push

Can Vietnam’s small-business backbone survive a compliance squeeze – all at once?

Vietnam workers keenest on AI in South-east Asia; Singapore employees among most sceptical: survey

A Microsoft Research field experiment involving 4,867 developers across Microsoft, Accenture, and a Fortune 100 company found that developers using an AI coding assistant complete 26.08 per cent more tasks on average.

This is not happening at the edge of the industry. Stack Overflow’s 2025 Developer Survey found that 84 per cent of developers use or plan to use AI tools, while 51 per cent of professional developers turn to them daily.

In light of this, let us reimagine that same pitch meeting. A company says, “We have five developers. They’re US$25 an hour.”

The buyer’s next question should be, “What can those five engineers deliver now that they have AI?”

That now becomes a very different conversation. Because AI is changing how much engineering capacity a team can produce, the hourly rate becomes a much less useful proxy for value.

Being cheap isn’t the point

I do not think the answer to this shift is pretending Vietnam is no longer a low-cost engineering market. Depending on seniority, specialisation, and delivery model, the country remains broadly competitive with other major offshore markets.

In my experience, typical mid-level development rates in Vietnam can run about US$25 to US$45 an hour, compared with roughly US$20 to US$40 in India, US$25 to US$40 in the Philippines, and higher ranges in markets like Eastern Europe.

There have always been cheaper markets than Vietnam, so developers here have never competed solely on price. But it is becoming clear that being cheaper is no longer a competitive advantage.

I have had versions of this conversation with several founders in the country. The reaction is usually the same pushback, and I understand why.

If you have spent 10 years building a company around software outsourcing, trained hundreds of developers, conducted sales around offshore delivery, and won clients because your team is cheaper than their local alternative, the foundation of your business is changing.

What AI can’t do

One founder pushed back at me directly, stressing that clients still care about price.

“If we’re not cheaper, why would they choose Vietnam?” he asked.

It is a fair question, and I do not think clients are disregarding price considerations. But if price is your value proposition, then AI will weaken your position year after year.

This is where I think the industry needs to make an important distinction.

AI excels at implementation by generating boilerplate, writing tests, explaining code, refactoring functions, creating API endpoints, producing documentation, and investigating errors.

But while these are all important, none of them answer the harder questions, such as:

• Should we build this feature at all?
• Does this architecture make sense for the business?
• How does this new service interact with a 10-year-old system nobody fully understands?
• Is the generated code secure?
• Will it scale?
• What happens when the AI-generated solution is almost right but fails in production?

Stack Overflow’s 2025 survey also captures this tension: 46 per cent of developers said they distrust the accuracy of AI output, while 66 per cent shared that their biggest frustration is AI-generated answers that are “almost right, but not quite”.

“Almost right” is where engineering begins. While AI can initiate implementation, it cannot be accountable for the decision. That responsibility still belongs to human engineers, and this is where Vietnam’s next opportunity lies.

I saw this play out on one of our own projects, where an AI tool generated part of an API’s data-handling logic. The first output looked fine – it passed functional tests and handled the normal transaction flow.

It was only during the senior engineer review and edge-case testing that we found the problem: Under a specific scenario, two near-simultaneous requests could cause the same record to be processed twice.

Fixing it took six to eight engineering hours, which involved reproducing the issue, correcting the code, adding tests for the edge case, and running regression tests.

The AI was right enough to pass the obvious checks but wrong enough to cause a serious failure if no one had looked closer.

The next advantage

Vietnamese firms have spent years competing on the number of developers and the rate they could offer. But the next generation of successful engineering companies will compete differently, with engineers who understand architecture, systems, security, business requirements, and trade-offs.

For example, a Singapore fintech client recently approached my company with a loan comparison platform already mapped out. However, the requirements had been AI-generated and lacked a clear roadmap or a sense of which technologies actually fit the goals.

Before writing a line of code, we spent two weeks restructuring the requirements into a validated MVP scope, mapped against a proper work breakdown structure.

Only then did a nine-person team – a project manager, a business analyst, six developers, a designer, and a quality control engineer – start building. They delivered the full platform in three months.

At Saigon Technology, AI-generated output moves through AI-assisted implementation, automated testing, senior engineer review, and quality assurance before release.

A recent project involved eight backend APIs, five front end components, and several third-party integrations – work that would typically take about 10 engineering days. We completed it in just around six without skipping any of those review steps.

That is what I mean by engineering leverage.

AI can generate a set of requirements in minutes, but it cannot tell you whether they are buildable, secure, or the right MVP for a three-month timeline. That judgment is what the engineers on the project are actually paid for.

The goal is not having fewer engineers because AI has replaced them. Instead, it is having better engineers who can accomplish more because AI multiplies their capacity.

That is very different from just saying, “We have 500 developers at a lower rate”. TECH IN ASIA

Thanh Pham is the co-founder and CEO of software development firm Saigon Technology.