System Development in Vietnam — 2026 Cost Model and 7 Questions

Before you start collecting competitive quotations for system development in Vietnam, there is one decision worth making first. The phrase covers two purchases with completely different characters. One is building the systems that keep a Vietnamese plant running day to day. The other is buying Vietnam software development capacity at a lower rate. Put both into a single estimate and the project usually breaks somewhere in the middle. This article lines up three delivery structures in a 50 person-month cost model, sorts through the regulatory changes that came into force in 2026, and sets out the questions to ask before you place an order.

System Development in Vietnam Covers Two Very Different Purchases

When a Japanese manufacturer asks us about development in Vietnam, the first thirty minutes always go to the same question. Is this a system that someone on the factory floor in Vietnam will touch every day? Or is it a system used in Japan or in a third country that you want Vietnamese engineers to build? Companies that can answer immediately and companies where the answer changes depending on who is in the room end up on very different paths.

Both are described with the same words, but as purchases they are separate things. What you are buying is different. The way they fail is different. The department that owns the outcome internally is different. And yet, by the time it reaches the approval memo, both have been compressed into a single line that reads something like “development of a system in Vietnam”. That single line is what later produces estimate inflation and schedule slippage.

Throughout this article we will call the first one the local operational system and the second one offshore development. Neither is superior to the other. They are simply bought in different ways, and the purchasing process has to reflect that.

Building a System to Run the Local Plant

The first type exists to run the Vietnamese plant itself. Production result collection, inventory control, shipping instructions, quality records, equipment monitoring. The users are local operators and staff, the screens are in Vietnamese, the printed forms follow Vietnamese practice, and the data sits under Vietnamese law.

The substance of this purchase is designing how work gets done. The software comes out the other end as a consequence. What it demands from the buyer is the ability to put shop-floor procedures into words, and the authority to decide how those procedures connect to the corporate standard. No matter how capable the development company is, only the people on the floor know how the work is actually performed today. Try to outsource that part and requirements definition spins without traction.

This type also never really finishes. People rotate out, products change, regulations move, and each time someone has to open the system again. You are buying on the assumption of five or ten years of use, which means the purchase includes who maintains it, whether there is a contact point you can talk to in your own working language, and whether anyone can physically reach the site. For the broader question of how to design a rollout across overseas plants, our article on overseas plant system rollout covers where to draw the line between the corporate standard and local optimisation.

Buying Development Capacity Offshore

The second type is a transaction for engineering resources. Work that belongs to a project in Japan, or to a system used across the whole group, is sent to a country with lower rates. What you receive is source code and a working application, and you operate it yourself afterwards.

The substance of this purchase is securing effort at a lower price. So the evaluation criteria are unit rate, productivity and quality, and the comparison set is not only other Vietnamese firms but also India, the Philippines, Indonesia and nearshore options inside Japan. The requirements come from the ordering side, and the more firmly the specification is fixed, the more efficiency you extract.

What deserves attention is that offshore development in Vietnam has stopped being a story about cheap labour. Vietnamese engineers have been moving from taking contracts for straightforward implementation towards work that reaches into design and technology selection. As we will see below, rates have risen accordingly. Choosing on price alone means misreading the change that is already under way.

Why One Estimate Covering Both Falls Apart

Why does combining the two into a single estimate break down? There are three reasons.

First, the requirements come from different places. Requirements for a local operational system come from the Vietnamese plant, while requirements for offshore development come from headquarters. Mix both into one project and the meeting that is supposed to freeze requirements ends up with the section manager from the Vietnamese factory, the corporate IT team and the business division all in the same room. Priorities never settle, and only the calendar moves.

Second, the acceptance criteria are different. A local operational system is accepted on whether the floor can use it. Offshore development is accepted on whether it matches the specification. The former contains a great deal that never made it into a written specification, and the latter deliberately does not deliver what was not written. Trying to judge both with the same acceptance test makes it hard to find a landing point that satisfies either side.

Third, maintenance has a different character. Maintaining a local operational system means keeping up with changes in how the local business operates. Maintaining offshore-built code means modifying source. Bundle both into a single annual maintenance contract and a request from the plant to add one more field to a screen goes into the same queue as code changes owned by headquarters.

So the decision to make before ordering is not which vendors to invite into the comparison. It is which of the two you are actually buying. If you need both, it is generally better to separate the estimates and the contracts.

Vietnam in 2026 — Unit Price Alone No Longer Decides

An approval memo written on the assumption that Vietnam is cheap can end up out of step with 2026 conditions. Rates have risen, and the kind of people you need has shifted as well. This section uses only published figures to set out what is actually happening.

As a premise, the rates discussed below are monthly rates charged when a Japanese company places work with a Vietnamese offshore development company. They are not Vietnamese domestic salary levels. We use the benchmark compiled by Offshore Kaihatsu.com for 2026, which cites the Offshore Development White Paper (2025 edition) and carries a last-updated date of 13 February 2026.

Monthly Rates Rose and the Gap With Japan Narrowed

According to that source, the monthly rate for a Vietnamese programmer is JPY 401,000, up 1.8 percent year on year. A senior engineer is JPY 500,000, up 3.5 percent. A bridge system engineer is JPY 590,000, unchanged from the previous year. A project manager is JPY 714,000, up 2.0 percent. Taken together, the level is described as roughly one third to one half of domestic Japanese rates.

What tends to be missed here is that the increases are not uniform across roles. Programmers and project managers rose gently, while senior engineers rose more sharply at 3.5 percent. Bridge system engineers were flat. One reasonable reading is that people who can carry design work are appreciating faster than raw implementation capacity.

And the range from one third to one half is a very wide band in practical terms. Against this article’s model premise of JPY 50,000,000 for 50 person-months from a domestic Japanese vendor, the difference between landing at one third and landing at one half is JPY 8,333,333. Comparing rate cards without understanding what determines where you land inside that band does not give you anything to decide on. The cost model in the next section is an attempt to explain that band in terms of where you place the delivery team.

The Minimum Wage Rose an Average of 7.2 Percent in January 2026

The floor under labour cost is moving too. According to the Japan Institute for Labour Policy and Training (JILPT), Vietnam’s National Wage Council decided on 11 July 2025 to raise regional minimum wages by an average of roughly 7.2 percent from January 2026. Region 1, which includes the urban areas of Hanoi and Ho Chi Minh City, is set at VND 5,310,000 per month. Minimum wages are defined across four regional categories, and this was the first increase in a year and a half.

On the legal side, CAST Vietnam notes that Decree 293/2025/ND-CP, which set out the increase, was promulgated on 10 November 2025 and took effect on 1 January 2026. It is already in force, so labour cost planning for the 2026 financial year should treat it as an existing condition rather than something ahead of you.

A minimum wage increase does not translate directly into higher rates for IT engineers, because engineer salaries sit far above the minimum. It does, however, feed into the calculation base for social insurance contributions, into indirect headcount costs, and into the cost of running an office. When an offshore development company’s cost base rises, that eventually reaches the rate card. Building a medium-term plan on the assumption that rate tables stay flat year after year is worth avoiding.

The Shortage Is Not Programmers but Upstream Roles and Bridge System Engineers

Vietnam has real depth in IT talent. TopDev’s Vietnam IT Market Reports put the number of people working in IT-related roles at around 560,000, with roughly 55,000 to 60,000 students entering IT-related faculties each year. On headcount alone, implementation capacity is not running dry.

The problem is distribution. Bridge system engineers who can take requirements in the client’s language, translate them into a specification that fits the client’s commercial practice, and then push that down to the Vietnamese team, together with upstream people who understand the business well enough to design it, are far scarcer by comparison. The fact that the bridge system engineer rate did not fall and instead stayed flat can be read as a reflection of that supply and demand. Some estimates suggest the shortage will continue, but the estimates vary widely enough that we will not put a number on it here.

The practical implication is clear. A project succeeds or fails not on how many programmers you can gather but on who carries the bridge and upstream roles and how much of their time you actually get. And those slots are less freely available than buyers tend to assume. When you read an estimate, checking how many person-months of bridge system engineer the structure contains, and what share of that is genuinely dedicated to your project, is worth far more than negotiating a few thousand yen off the rate.

A Cost Model for Vietnam Software Development at 50 Person-Months

From here we work with concrete numbers. Everything below is a model calculation. It is not an average of real projects and not a record of actual results. Change the assumptions and the results change. Read it as something to re-run against your own conditions.

The model case is a new build of an operational system covering production result collection and inventory control, for a Vietnamese plant with roughly 400 employees, at a development scale of 50 person-months. The rates are taken directly from the Offshore Kaihatsu.com benchmark cited above. The domestic Japanese vendor rate of JPY 1,000,000 per person-month is a model premise this article has set for comparison purposes, not a figure lifted from a source.

On that basis, we compare three patterns that differ only in where the team sits. The total volume is fixed at 50 person-months in every pattern, and pattern B reduces the Vietnamese side by exactly the eight person-months it places on the Japanese side. In other words, this is not a comparison where you buy quality by adding effort. It moves only one variable, which is where within the same total effort each role is placed.

Line up the differences and B exceeds A by JPY 4,544,000, C exceeds B by JPY 23,718,000, and C exceeds A by JPY 28,262,000. Judged only on those three lines, A looks overwhelmingly cheap and C looks indefensible. A great many approval memos stop right there.

What matters in practice comes after that. A’s average rate is JPY 21,738,000 divided by 50 person-months, which is JPY 434,760 per person-month. Now suppose that rework caused by requirement mismatches consumes 10.5 person-months, which is 21 percent of the scale. The additional cost is 10.5 person-months at JPY 434,760, or JPY 4,564,980. A’s effective total becomes JPY 26,302,980, which is JPY 20,980 above B’s JPY 26,282,000.

Put differently, the JPY 4,544,000 that B adds on top of A, which is the cost of the eight upstream person-months placed on the Japanese side, can be read as an insurance premium covering 10.5 person-months of rework. If rework exceeds roughly 21 percent of scale, B, with upstream work kept on the Japanese side, ends up cheaper. If it stays below that, A is cheaper. That level of roughly 21 percent is the break-even point for the structural decision.

It should be stated plainly that this comparison assumes zero rework on the B side. Keeping upstream work on the Japanese side does not eliminate rework. So the question worth asking is not which of A or B is cheaper. It is whether your own requirements definition is in a state that can hold rework below 21 percent of scale. If shop-floor procedures are undocumented, if the staff at the Vietnamese site cannot discuss requirements in a shared working language, and if nobody on the Japanese side can articulate the business process, then 21 percent is an optimistic number.

This way of thinking about a break-even point sits close to the practice of breaking cost down by phase. How to separate development cost into layers is covered in detail in our article on business system development cost and ERP integration, which is worth reading alongside this one if you want to sharpen how you read quotations.

The 2026 Regulatory Changes That Bite After Go-Live

Let us pause the cost discussion and turn to regulation. If you are building a system in Vietnam in 2026, there are changes worth confirming before you look at development cost at all. Every one of them shows up after go-live, as an operating cost.

All four items below are already in force as of the publication of this article. Read them not as things to prepare for, but as a checklist against which the systems you are already running should be inspected.

Interpretation of these laws and whether they apply to your own operation is an individual judgement. What follows is an outline based on published information only, and practical decisions should be confirmed with a local law firm or accounting firm.

The Personal Data Protection Law (91/2025/QH15) Took Effect in January 2026

Vietnam’s Personal Data Protection Law, Law No. 91/2025/QH15, was passed by the National Assembly on 26 June 2025 and took effect on 1 January 2026. It runs to five chapters and 39 articles. Its scope is broad. Any organisation handling the personal data of Vietnamese residents falls within it, regardless of where the processing takes place. If you process data on employees at your Vietnamese site using servers in Japan, you are in scope.

Three points matter in practice. The first is cross-border transfer, where a Cross-border Transfer Impact Assessment must be submitted within 60 days of the first transfer. The second is breach and loss notification, which must be made within 72 hours of detection. The third is sanctions, where violations relating to cross-border transfer carry penalties of up to 5 percent of the previous year’s annual revenue. A sanction based on revenue is an entirely different order of magnitude from the cost of modifying a system.

Factory systems are often assumed not to handle personal data, but in reality they do. Production result collection is tied to operator identifiers, attendance and access records carry names, and quality records retain the name of the inspector. If any of this is being pulled up into a business intelligence tool at the Japanese head office, that is a cross-border transfer. Taking an inventory of data items during the design phase costs far less than retrofitting an operational process to meet the 72-hour rule after the fact.

The AI Law and the Law on Digital Technology Industry

The other regulatory current concerns artificial intelligence. According to commentary from One Asia Lawyers, the Law on Digital Technology Industry was enacted on 14 June 2025 and took effect on 1 January 2026. It has six chapters and 51 articles, covers semiconductors, AI and digital assets, and is notable as the first Vietnamese law to regulate AI.

Vietnam Briefing further reports that the standalone AI Law took effect on 1 March 2026. It adopts a framework that sorts AI systems into four risk tiers, described as unacceptable, high, medium and low. Any AI system affecting Vietnamese users, the Vietnamese market or national interests brings foreign operators into scope as well, and providers based outside Vietnam are required to appoint a local legal representative.

If you plan to put AI into a factory system, this framework reaches into your requirements. Functions such as visual inspection judgement, demand forecasting and equipment anomaly detection carry different accountability obligations depending on which risk tier they land in. A configuration where a cloud AI service contracted by the Japanese head office is consumed at the Vietnamese site also raises the question of who counts as the provider. For the full picture of AI adoption in Vietnam, see our article on AI implementation in Vietnam, which is the better starting point when AI itself is the subject.

E-Invoicing Widened in Scope From June 2025

There is an accounting-side change to note as well. According to a JETRO business bulletin, Decree 70/2025/ND-CP was promulgated on 20 March 2025 and took effect on 1 June 2025. Under it, businesses with annual revenue of one billion dong or more are required to use electronic invoices issued from cash registers capable of transmitting data to the tax authorities.

For a manufacturing site, this becomes a question of whether the output of the systems involved in shipping and sales meets the tax authority’s requirements. You will need to establish whether invoices are issued directly from your core system, or routed through an accounting package or an external service, and whether that integration satisfies the current requirements.

For a new build, decide early where this integration lives. Absorb accounting requirements inside the production control system and you will be opening the production control system every time tax rules change. Separate them completely and you will be reconciling shipment records against billing by hand. Either way, whether the choice was made consciously at design time is what determines the cost of later modifications.

Provincial Mergers Changed Addresses and Administrative Codes

Of all the regulatory changes, the administrative reorganisation may be the one that hit systems most directly. Nikkei reported that from 1 July 2025, the previous 63 provinces and cities, comprising 57 provinces and six centrally governed cities, were reorganised into 34, comprising 28 provinces and six centrally governed cities. The district level, the second tier, was abolished, and the number of communes was sharply reduced.

On the system side, this touches address master data, administrative codes, customer and supplier masters, delivery destination masters, address formatting on printed forms, and any screen that uses an address as a search key. If you handle domestic logistics, it reaches the way delivery routes are segmented as well. Sometimes a bulk master update is enough. Sometimes the code scheme itself has to be reconsidered.

Whether procedural steps such as amending company registration are required is not something secondary sources can settle. How to align the registered address with the way addresses and administrative codes are held in the system is best confirmed with your own accounting firm or the tax authority. What we want to fix here is simply the fact that master data will need work.

Now let us add the cost of responding to these four changes to the earlier cost model. Annual maintenance is set at 15 percent of development cost. Regulatory response is identical across all patterns and consists of an initial JPY 5,300,000, made up of JPY 3,000,000 for personal data protection work, JPY 1,500,000 for e-invoicing work and JPY 800,000 to update address and administrative code master data, plus annual operating cost for personal data protection of JPY 600,000 over five years, which is JPY 3,000,000, for a combined total of JPY 8,300,000. These regulatory figures are also a model premise and are not taken from a source.

What stands out in this table is that the difference between A with 10.5 person-months of rework and B is only JPY 36,715. With no rework at all, A’s five-year total of JPY 46,341,500 sits JPY 7,952,000 below B’s JPY 54,293,500. But once rework reaches the break-even point, A becomes JPY 54,330,215 and now exceeds B by JPY 36,715. It is not the passage of five years that erases the gap. It is the precision of requirements definition that decides it. The deciding input is not the rate card but the maturity of your own requirements process.

The other thing the table shows is that the JPY 8,300,000 of regulatory response is identical in every pattern. Wherever you place the order, that portion does not disappear. Part of what you save through a cheaper development cost is handed straight back to compliance work. When you compare quotations, requiring every vendor to include this common cost in their proposal puts the comparison on level ground.

Where to Start With Factory Digitalisation in Vietnam

Enquiries about factory digitalisation in Vietnam usually arrive in the form of not knowing where to begin. A plant running on paper and spreadsheets gets a proposal for an integrated package, sees the price, and stops. The cycle repeats.

The ordering principle is simple. First make it possible to count what was produced. Then get physical balances to agree. Only then connect plan to actual. Concretely, start with production result collection, widen into inventory control, and finally extend into production planning and control. Do it in reverse and you end up with a mechanism for improving planning accuracy but no data to feed it.

There is another reason to put result collection first, which is that the return on investment is visible. Simply making daily output and defect counts visible by line and by item changes the conversation on the floor. Getting local staff comfortable with using a terminal at this stage makes the next step, inventory control, settle much faster. Skip this step and the gap between book inventory and physical counts never closes, leaving a residue of distrust in the system.

There are four main reasons a package built for Japan cannot be dropped in unchanged. The first is language. Screens and operating manuals have to work in Vietnamese. Interfaces available only in Japanese and English will not be used by operators on the floor. The second is printed forms. If the formats used for shipping, delivery and billing do not match local practice, someone will simply rebuild them in a spreadsheet.

The third is regulatory fit. As the previous section showed, personal data handling, e-invoicing, and address and administrative codes all carry Vietnam-specific requirements. These are typically absent from a package’s standard functionality and have to be built as add-ons. The fourth is connectivity and infrastructure. Depending on where the plant sits, the path to a head-office cloud can be thin and unstable. You then have to decide between a design that keeps result entry working offline and placing a server on site.

None of these four is an argument that packages are bad. They are the natural friction that appears when software refined around one country’s commercial practice is placed into another. So decide the line between what you use as standard and what you build locally at the very start of the evaluation. Since this line applies across overseas deployment in general, it is worth having a policy early if you plan to extend to several countries.

The entry point is the same when you are evaluating a production control system for a Vietnamese plant. Do not ask for process-level cost accounting or overall equipment effectiveness on day one. Build the state in which results are collected accurately. Deciding what you want to judge from the data after the data has started accumulating produces much smaller requirement gaps. The 21 percent rework break-even point from the cost model depends heavily on whether you can hold this sequence.

What Differs Between Hanoi, Ho Chi Minh City and Da Nang

System development in Hanoi, in Ho Chi Minh City and in Da Nang runs differently even within the same country. Few companies think about the city when they look for a vendor, but the relationship between where your site is and where the development company is affects how the project moves.

Hanoi in the north is the capital, and many firms there have built their track record on work for government bodies and large enterprises. The north has a dense cluster of industrial parks including Hai Phong and Bac Ninh, which gives Japanese manufacturers with sites in that area a practical advantage in being able to visit easily. Ho Chi Minh City in the south is Vietnam’s largest commercial centre, with more companies and greater talent mobility. It is easier to assemble a range of skills there, though we frequently hear that turnover is also higher. Da Nang in the centre does not match the other two in scale, but it is known as a city that has invested in developing IT talent, and development companies there often lead with relatively low living costs and higher retention.

These are tendencies rather than rules. Rather than deciding by city name, it is more reliable to check, company by company, how many hours it takes to reach them from your site, how much work they have done for Japanese companies, and how many people can hold a technical discussion in your working language. For a local operational system in particular, whether the development team is close enough to visit the plant has a direct effect on the quality of requirements definition.

Companies that already have a site in Thailand face one more consideration. Order system development across Southeast Asia country by country with no coordination and the number of maintenance contact points and contracts grows in step with the number of sites. On the other hand, expecting a Thai vendor to take on Vietnamese regulatory compliance is unrealistic. The practical answer is to split what is standardised at corporate level, meaning how masters are held, the code scheme, and how data is collected upward, from what is built locally in each country, meaning printed forms, regulatory compliance and interface language, and then consolidate only the design of the common layer with a single partner. Vendor selection on the Thai side is covered in our article on how to choose a system development company in Thailand.

Seven Questions to Ask Before You Request a Quote

Here are seven questions worth asking before you request quotations. In every case, how the answer is given reveals more than the content of the answer itself. A good answer comes back with the preconditions you did not think to ask about. A worrying answer simply agrees with the question and stops there.

Question 1. Is this project a local operational system or offshore development? This asks the vendor to make the distinction from the opening of this article for themselves. A good answer responds to your actual situation, along the lines of saying that your case is the former and that requirements definition should therefore start with interviews on the Vietnamese shop floor. A worrying answer is that they can handle either, which demonstrates breadth but may indicate they have not understood why the distinction matters.

Question 2. How many person-months of bridge system engineer are assigned, and how are they shared with other projects? As set out above, securing bridge capacity is the rate-limiting step on a project. A good answer gives specific person-months, the current allocation across projects, and the replacement plan if someone leaves. A worrying answer says only that a dedicated resource will be provided, without naming a headcount or a person.

Question 3. How is the cost of requirement mismatches allocated when they occur? As the cost model showed, the volume of rework determines which structure is correct. A good answer presents in advance the criteria for judging a specification change and the method for producing an additional quotation. A worrying answer defers everything to case-by-case discussion, which tends to produce friction in the back half of the project. This is a point worth writing into the contract.

Question 4. Does compliance with the Personal Data Protection Law fall inside your scope? This checks their position on a law that is already in force. A good answer separates what is guaranteed by the system from what is guaranteed by your own operating procedures, covering the cross-border transfer impact assessment and the 72-hour notification process. A worrying answer states that they are already compliant and stops, leaving it unclear what compliance consists of.

Question 5. In which master data do you manage post-reorganisation addresses and administrative codes? This measures the granularity of their implementation practice. A good answer can walk through how master updates were carried out on previous projects and what the impact scope would be at the next revision. A worrying answer is one where the conversation reveals that addresses are treated as a plain text field.

Question 6. What determines the pass criteria for acceptance? Acceptance criteria differ between a local operational system and offshore development. A good answer goes into the number of acceptance test scenarios, where the testing happens, and whether local operators take part. A worrying answer offers only that the system will behave as specified, which omits the question of whether the floor can actually use it.

Question 7. Who is maintaining this system five years from now? Finish by asking about the long term. A good answer covers the maintenance structure, the annual fee, coverage hours and the conditions for handing over source code. A worrying answer treats maintenance as an appendix to development and leaves both the amount and the structure until later. As the five-year total cost table showed, five years of maintenance reaches roughly three quarters of the development cost.

None of the seven takes long to ask. In fact, the split between the questions a vendor finds easy to answer and the ones they find hard tends to be very clear. Treat whatever they struggled with as the risk profile of that project.

What to Check on Every Quotation

After the questions comes the reading of the quotations themselves. Comparing only headline totals puts documents built on different conditions onto the same scale. Check that the following items are explicitly stated.

Items that do not appear in a quotation usually have not disappeared. They have simply been shifted onto the buyer. Acceptance testing and regulatory work in particular come back as your own internal effort if they are not written down. Before comparing amounts, start by levelling out what has and has not been documented.

Only after filling in this table should you compare amounts. If you took quotations from three companies, lay out where each of the three has left blanks. The position of the blanks says a great deal about where each company is strong and where it is not. A quotation with every field filled in is itself a piece of evidence about capability.

To sanity-check whether an amount is reasonable, it also helps to build a rough model yourself, in the manner of the calculation above. A sense of scale at 50 person-months, an average rate of JPY 434,760 per person-month, and a maintenance ratio of 15 percent of development cost. Those three reference points alone are enough to judge roughly where a quoted figure sits.

Frequently Asked Questions

These are the questions that come up most often in conversations about this topic. All of the answers are general. In an individual project the assumptions will differ.

What is the typical cost of system development in Vietnam?

On monthly rates, the levels published for 2026 are JPY 401,000 for a programmer, JPY 500,000 for a senior engineer, JPY 590,000 for a bridge system engineer and JPY 714,000 for a project manager. That places Vietnam at roughly one third to one half of domestic Japanese rates. Totals depend on scale and team structure, so for the 50 person-month model case this article showed a range of JPY 21,738,000 to JPY 26,282,000 with development placed in Vietnam, against JPY 50,000,000 for a single contract with a domestic Japanese vendor. These are model calculations, not averages of real projects.

Is Vietnam software development still cheap?

Compared with domestic Japanese rates, yes. But the gap has narrowed. Some roles, such as bridge system engineer, were flat year on year, while the roles that rose did so within a range of 1.8 percent to 3.5 percent, and the minimum wage went up by an average of roughly 7.2 percent from January 2026. On top of that, in a five-year total cost view, the JPY 8,300,000 of regulatory work applies wherever you place the order. It is safer to assume that a low development cost does not translate directly into a low total.

Can a Japanese package be installed as-is as a production control system in Vietnam?

Usually not without modification. There are four reasons, namely language, printed forms, regulatory fit and connectivity. Screens and operating manuals need to work in Vietnamese, shipping and billing formats need to match local practice, the requirements of the Personal Data Protection Law, e-invoicing and administrative codes need to be satisfied, and result entry needs to keep working when the plant’s connection is thin. Few packages meet all four with standard functionality. Drawing the line between what you use as standard and what you build locally, at the very beginning, is the realistic approach.

Should we order from Hanoi or from Ho Chi Minh City?

The city name alone does not settle it. If your plant is in the north, a company around Hanoi reduces the burden of site visits, and if it is in the south, Ho Chi Minh City is closer. The deciding factors are not the city itself but distance from your site, experience with Japanese manufacturing clients, how many people can hold the discussion in your working language, and what happens structurally if a key member leaves. It is more reliable to include Da Nang in the comparison and check those four points company by company.

Should we standardise systems across our Thai and Vietnamese sites?

Standardising everything and separating everything are both extremes. The realistic approach is to standardise the design of the common layer, meaning how masters are held, the code scheme and how data is collected upward, and build printed forms, regulatory compliance and interface language separately in each country. Consolidating only the design of the common layer with one partner makes it much easier to extend horizontally as sites are added. Since each country’s regulations change independently, trying to standardise compliance as well tends to break down.

Summary

The phrase system development in Vietnam covers two purchases with different characters. One is the local operational system that runs a Vietnamese plant, the other is offshore development bought as engineering capacity. Because the two differ in where requirements originate, in acceptance criteria and in the nature of maintenance, mixing them into one estimate causes the project to break partway through. The decision to make before ordering is not which vendors to compare but which of the two you are buying.

Looking at 2026 conditions on that basis, bridge system engineer rates were flat while the roles that rose went up by 1.8 percent to 3.5 percent year on year, and the gap with Japan narrowed. In the 50 person-month model, pattern A with all effort in Vietnam comes to JPY 21,738,000, pattern B with upstream work on the Japanese side comes to JPY 26,282,000, and pattern C with a single domestic Japanese contract comes to JPY 50,000,000. But at the point where rework reaches 10.5 person-months, equal to 21 percent of scale, A’s effective total of JPY 26,302,980 exceeds B’s JPY 26,282,000 by JPY 20,980, and even across a five-year total the difference is only JPY 36,715. The axis of the decision is not whether to choose on price, but how far your own requirements definition can hold rework down.

On the regulatory side, the Personal Data Protection Law (91/2025/QH15) and the Law on Digital Technology Industry have applied since 1 January 2026, the AI Law since 1 March 2026, and Decree 70/2025/ND-CP on e-invoicing since 1 June 2025. All are already in force. In addition, the provincial reorganisation of 1 July 2025 forced changes into address and administrative code master data. The cost of responding to these applies identically whichever pattern you order under. Decide a vendor on development cost alone and this portion ends up sitting outside the quotation entirely.

TOMAS TECH is a Bangkok-based integrator that has delivered production management and energy management systems to Japanese manufacturers in Thailand and across ASEAN. Whether you are still working out where to start at a Vietnamese site, or partway through comparing quotations from several companies, we are happy to talk it through in terms of what actually happens on the floor. Nothing needs to be firmed up before you get in touch, so please reach us through our contact page.

References

• Offshore Kaihatsu.com — Vietnam Offshore Development Monthly Unit Rate Benchmarks, 2026 Edition
• JILPT (Japan Institute for Labour Policy and Training) — Vietnam to Raise the Minimum Wage by an Average of Roughly 7.2 Percent in January 2026
• CAST Vietnam — Vietnam Minimum Wage 2026, Regional Amounts Under Decree 293/2025/ND-CP and Practical Responses
• Vietnam Law on Personal Data Protection, Law No. 91/2025/QH15 (full English translation)
• One Asia Lawyers — Vietnam, Introduction to Law on Digital Technology Industry 2025
• Vietnam Briefing — Vietnam AI Law, Regulatory Milestone and Business Implications
• JETRO Business Bulletin — E-Invoicing Mandate Extended From 1 June to Sole Proprietors With Annual Revenue of One Billion Dong or More (Vietnam)
• Nikkei — Vietnam Halves Its Provinces and Cities to 34 in a Sweeping Local Government Merger Aimed at Administrative Efficiency, but Confusion Persists on the Ground
• TopDev — Vietnam IT Market Reports