Factory Automation in Vietnam 2026: Payback Comes from Output, Not Headcount

Search for “Vietnam factory automation” and you will find articles that place a low robot density next to low labour costs. But the reason these projects stall in the investment committee of a Japanese-affiliated plant is not technical. It is that the denominator, the capital investment, is priced in dollars and comes to the same amount as in Thailand, while the numerator, the labour cost you can remove, is only 67.5% of Thailand’s. This article sets out a model assembly plant near Hanoi and compares two ways of using the same equipment. To say it up front, what pays back is not labour saving.

The numbers to write down first: the investment is identical, and Vietnam’s total labour cost is 67.5% of Thailand’s

When you discuss the payback of automation, the first number to write down is not the number of robots or the automation ratio. It is how the currencies of the numerator and the denominator move.

Equipment is the same machinery wherever you buy it in the world. Robot arms, servos and vision sensors are, as a rule, priced in dollars. The labour cost they displace, on the other hand, is set in local currency. What breaks when you carry a Thai design into Vietnam comes down to this single point: only the denominator is fixed in dollars, and only the numerator shrinks in local currency.

Let us fix the model used in this article first. Everything below is a model value used in this article, not a measured figure from any specific company and not a market rate. When you apply it to your own site, replace the wages, operating hours and electricity tariff with your own.

Shared assumptions (all of them model values used in this article)

Check: 7,000,000 x 0.215 = 1,505,000 VND / 7,000,000 + 1,505,000 = 8,505,000 VND / x 12 = 102,060,000 VND

Check: 15,000 x 0.05 = 750 THB (equal to the 750 THB cap), giving 15,750 THB / x 12 = 189,000 THB / x 800 = 151,200,000 VND

Divide one by the other and you get the ratio this whole article turns on.

Check: 102,060,000 / 151,200,000 = 0.675, so total operator labour cost in Vietnam is 67.5% of the Thai figure

In other words, when you replace one person with automation, the amount freed up in Vietnam is only two thirds of what it is in Thailand. The investment decision starts with a numerator that has already shrunk to 67.5%. That is the structure running through this entire article.

How the minimum wage of 5,310,000 VND relates to the model’s base pay of 7,000,000 VND

A word of clarification before we go further. The statutory minimum wage for Region I in Vietnam is 5,310,000 VND/month (25,500 VND per hour) from 1 January 2026. The base pay and allowances of 7,000,000 VND/month assumed in this model are higher than that. This is not a contradiction.

The minimum wage is a floor, not the amount at which people can actually be hired in an industrial park. If a Japanese-affiliated assembly plant near Hanoi wants to keep two shifts of operators staffed on a continuing basis, the going rate once allowances (perfect attendance, meals, commuting, shift premium) are stacked on top of base pay is normally above the minimum wage. This article adopts 7,000,000 VND/month as an assumed value close to that going rate, then adds the 21.5% of employer social insurance contributions to reach a total labour cost of 8,505,000 VND/month.

So when you read the calculations in this article, please keep these two apart.

• 5,310,000 VND/month: the statutory floor. An official figure set by decree. Covered in the section on wage increases below.
• 7,000,000 VND/month: the pay level assumed by this article’s model. An assumed value, not a measured one.

If you redo the calculation for your own site, replace the 7,000,000 VND with the amount you actually pay. The numerator is very largely determined by that single figure.

Equipment (identical in both countries; the investment is assumed to be the same because it is dollar-denominated)

Check: 6,312,000,000 / 26,300 = USD 240,000 / 6,312,000,000 / 800 = 7,890,000 THB

Check: 6,312,000,000 x 0.035 = 220,920,000 VND

Assuming the same investment of 6,312,000,000 VND in both countries is not a crude simplification. Robot arms, controllers, vision systems and safety equipment are all imported, and the engineering cost of installation and commissioning also moves at close to international prices. Local sourcing makes a difference on some peripheral items such as frames and wiring, but not enough to move the price of the cell as a whole. That is the basis for the assumption.

Running costs (per year)

Check (Vietnam): 60,720,000 + 220,920,000 = 281,640,000 VND

Check (Thailand): 26,400 x 4.2 = 110,880 THB / x 800 = 88,704,000 VND / 88,704,000 + 220,920,000 = 309,624,000 VND

Look at running costs alone and Vietnam comes to 281,640,000 VND a year against Thailand’s 309,624,000 VND, so Vietnam is cheaper. That is because the electricity tariff assumed in this article is lower. Taken in isolation, this makes Vietnam look like the better case. The conclusion, however, runs the other way. The gap of 27,984,000 VND in running costs is far smaller than the gap in the numerator, the labour cost. The next section puts numbers on it.

The same equipment, a payback 1.81 times longer: 19.09 years in Vietnam against 10.56 years in Thailand for the labour-saving design

Take a labour-saving design that worked in Thailand and carry it into Vietnam unchanged. Same equipment, same headcount reduction of 6 people (3 people per shift on two shifts).

Check (Vietnam): 612,360,000 – 281,640,000 = 330,720,000 VND / 6,312,000,000 / 330,720,000 = 19.09 years

Check (Thailand): 907,200,000 – 309,624,000 = 597,576,000 VND / 6,312,000,000 / 597,576,000 = 10.56 years

Check: 19.09 / 10.56 = 1.81 times

The project is rejected by arithmetic, not by technology

Thailand’s 10.56 years is by no means short either, but if you take the useful life of the equipment and the life of the process to be more than ten years, it is within the range you can put through an internal approval. Vietnam’s 19.09 years, by contrast, will not pass any manufacturer’s investment criteria. Most Japanese-affiliated companies look at five to eight years for capital equipment, ten at the outside. The 19.09-year figure presupposes that this process will still be making the same product in the same quantity nineteen years from now. No one can guarantee that.

What matters is that this 19.09 years is not a technical failure. Robot performance, cycle time and the accuracy of visual inspection are exactly the same as in Thailand. Only the numerator is different. When a Vietnamese project is stopped at the investment committee, the site tends to conclude that “Vietnam is not ready for automation,” but in reality the rejection comes from what was placed in the numerator of the division.

Look one level deeper into the breakdown. As shown above, running costs are lower in Vietnam, 281,640,000 VND against Thailand’s 309,624,000 VND. And payback still takes 1.81 times as long. In other words, effort spent lowering the denominator (electricity, maintenance costs, squeezing the equipment price) will not close this gap. Closing it would require cutting the investment of 6,312,000,000 VND by close to half, and that does not happen for a cell with the same functionality.

The thinking behind the Thai labour-saving case that this comparison is built on, how the costs are stacked up and how the headcount reduction is set, is laid out in labour saving and automation: case examples and cost-effectiveness. A design that passes in Thailand does not pass in Vietnam. The quickest way to see where that difference comes from is to put the two side by side.

“Waiting until wages rise” is not an answer

The next move that usually comes up here is to wait until labour costs in Vietnam catch up. As an investment decision, though, this misses. As covered in the section on wage increases below, the minimum wage was raised by an average of 7.2% in January 2026, but if you wait for wage growth to pull 19.09 years down into the five-year range, that process runs on manual labour, un-automated, for the several years you spend waiting. And while you wait, equipment prices move as well, in dollars.

There is a third move that neither cuts the denominator nor waits for wages. Swap the numerator.

Vietnam’s robot density of 31 units per 10,000 employees is not a sign of lagging technology

Let us check a figure that gets quoted often. Vietnam’s industrial robot density is estimated at 31 units per 10,000 employees, up 28% year on year (Ken Research).

For comparison, World Robotics 2025 from the IFR (International Federation of Robotics) gives a global average of 132 units, Asia at 131, Western Europe at 267 and North America at 204. Vietnam’s 31 units sits at roughly a quarter of the global average.

Note: this article does not give a robot density figure for Thailand, because the research behind it did not confirm a reliable primary source. Putting an unverified number into a comparison table would undermine the credibility of every other number in the article.

Attributing the low density to a technology gap leads to the wrong move

The usual reading of a figure of 31 units is that Vietnam is behind on automation and therefore has room to grow. The growth part is correct: the market is projected to expand from USD 112.0 million in 2025 to USD 231.7 million in 2031, a CAGR of 12.88% over 2026 to 2031 (Ken Research). Foreign direct investment into manufacturing is put at USD 25.58 billion in 2024, with manufacturing accounting for 24.1% of value added. Plants are being built.

But reading it as “low because it is behind” leads to the wrong move. As shown above, even with the same equipment, the same process and the same level of technology, payback comes to 19.09 years as long as labour saving sits in the numerator. The reason the equipment does not go in is that the approval does not go through. It is not a shortage of engineers, or the availability of robots, or resistance on the floor. The figure of 19.09 years is simply screened out automatically by the investment criteria.

Take that view and the density figure reads differently. Vietnam’s 31 units is not “a number that will rise once the technology catches up,” but “a number that is hard to raise as long as a labour-saving ROI test cannot pass at local wage levels.” Put the other way round: companies that swap the numerator of that test can install equipment regardless of where the national average sits.

Large manufacturing investments are in fact moving without waiting for the average to rise. Foxconn has been accelerating its investment in Vietnam, going as far as moves into humanoid robot manufacturing. It is hard to see investment on that scale as explicable by the arithmetic of reduced labour cost alone, and the natural reading is that production capacity and quality sit in the numerator (the substance of such investment decisions is not published, so this remains this article’s interpretation).

Do not build the case for overseas plant automation on averages

When you are weighing automation at an overseas plant, other companies’ adoption rates and average robot density are not grounds for a decision. An average is the result of “what has passed investment criteria in that country,” and has nothing to do with whether your own process pays back.

There are only three things to look at: the investment (the denominator), the annual benefit that equipment generates (the numerator), and how many years that benefit lasts. From here on, we rebuild the contents of the numerator.

Swapping the numerator: automation in a Vietnamese plant reaches 7.19 years with a volume-growth design

Exactly the same equipment and exactly the same investment of 6,312,000,000 VND as the labour-saving case, with only the use of it changed. Instead of using it to remove people, it is used to raise capacity and raise quality.

An important premise: Design 1 (labour saving) and Design 2 (volume growth) are two separate ways of using the same equipment. The two sets of benefits must never be added together. Read the tables below as two branches taken by one and the same investment of 6,312,000,000 VND.

Check: 3,600,000 x 0.024 = 86,400 units / 3,600,000 x 0.011 = 39,600 units / 86,400 – 39,600 = 46,800 units / 46,800 x 3,800 = 177,840,000 VND

Check: 777,600,000 + 177,840,000 = 955,440,000 / + 204,120,000 = 1,159,560,000 VND

Check: 1,159,560,000 – 281,640,000 = 877,920,000 VND / 6,312,000,000 / 877,920,000 = 7.19 years

Check: 19.09 / 7.19 = 2.66 times faster

The +18% capacity gain and the 2.4% to 1.1% defect rate are assumptions

To be explicit: the capacity gain of +18%, the defect rate moving from 2.4% to 1.1%, the unit gross margin of 1,200 VND/unit and the loss per defective unit of 3,800 VND are all model values used in this article. They are neither sourced measurements nor industry averages. Please replace them with your own figures. Unit gross margin in particular varies by an order of magnitude between products, so putting a real number there alone can change the conclusion.

One note on how the +18% was arrived at. When you automate a boxing and palletising process, capacity rises not because the robot is faster than a person, but because the variation in cycle time disappears. Human work speeds up and slows down around shift changes, around breaks and as fatigue accumulates. That variation creates waiting time downstream. Most of the benefit of a robot cell comes not from the fastest value but from lifting the slowest value. So when applying +18% to your own site, the logical first step is to measure how much variation the current process actually has.

Preventing double counting (two points)

Let us close off the two places in this calculation that are easiest to get wrong.

One: do not deduct the material and processing costs of the additional 648,000 units a second time.

The unit gross margin of 1,200 VND/unit is defined as selling price minus variable cost. Material costs and processing variable costs have already been deducted at the point that 1,200 VND is derived. Subtracting material costs again from the additional gross margin of 777,600,000 VND means deducting the same cost twice.

Two: the volume-growth design removes only 2 people. Do not carry over the 6 from the labour-saving design.

The volume-growth design raises capacity by 18% while keeping headcount broadly unchanged. That is why the additional gross margin materialises. Bring the 6-person reduction of the labour-saving design into it and you create a world where “6 people are removed while production rises 18%,” which contradicts the premises of the model. Labour cost reduction in the volume-growth design is 204,120,000 VND (2 people). It is not 612,360,000 VND (6 people).

Commit the second of these in particular and the net annual benefit swells to 777,600,000 + 177,840,000 + 612,360,000 – 281,640,000 = 1,286,160,000 VND, and the payback looks like 6,312,000,000 / 1,286,160,000 = 4.91 years. That benefit does not exist.

Check: 777,600,000 + 177,840,000 = 955,440,000 / + 612,360,000 = 1,567,800,000 / – 281,640,000 = 1,286,160,000 / 6,312,000,000 / 1,286,160,000 = 4.91

The breakdown of the benefit changes the priorities

Here is the annual benefit of 1,159,560,000 VND, ranked by amount, largest first.

Check: 777,600,000 + 177,840,000 + 204,120,000 = 1,159,560,000 VND / – 281,640,000 = 877,920,000 VND

About two thirds of the benefit is additional gross margin. That is the strength of this design and, at the same time, its greatest weakness. The next section looks at the weakness.

The volume-growth design only stands on the assumption that the output sells: 62.92 years without demand

The 7.19 years above stands on the assumption that all 648,000 additional units are sold. If they are not, the additional gross margin of 777,600,000 VND goes to zero in its entirety. What you produce becomes inventory, that is, working capital rather than margin.

Here are the numbers for that case, without hiding them.

Check: 177,840,000 + 204,120,000 = 381,960,000 VND / 381,960,000 – 281,640,000 = 100,320,000 VND / 6,312,000,000 / 100,320,000 = 62.92 years

62.92 years means it does not pay back. However generously you estimate the life of the equipment, you will not get there. The difference between 7.19 years and 62.92 years is not a difference in machine performance. It is whether the orders exist.

Three things to confirm before choosing the volume-growth design

The volume-growth design therefore only holds together as a package with demand verification. What should accompany the approval request is not the robot specification but these three points.

1. Whether real enquiries exist for the additional 648,000 units. State whether they are backed by customer forecasts, informal orders or the existing order backlog. “Sales will make it happen” is not a premise.
2. How many years that demand lasts. A design that pays back in 7.19 years presupposes that the demand continues for more than seven years. A one-off spot increase does not pay back.
3. How far it sinks if the output does not sell. In this model, 62.92 years. Whether you can write that number into the approval document is the fork in the road on whether you can choose the volume-growth design at all.

We cannot write that “the volume-growth design will always pay back.” What can be written is only the conditional conclusion that the volume-growth design pays back in 7.19 years if demand follows, and takes 62.92 years if it does not.

A middle course when demand cannot be read

It is more usual for demand not to be fully readable. Here is one practical way of handling that case: draw the minimum payback line using defect reduction and labour cost reduction only, and treat the additional gross margin as upside.

In this model, the demand-independent benefits are defect reduction of 177,840,000 VND plus labour cost reduction of 204,120,000 VND, totalling 381,960,000 VND, and after deducting running costs of 281,640,000 VND, 100,320,000 VND remains. That is a state in which the running costs at least cover themselves. A design that turns negative here, that is, one that cannot even pay its running costs without demand, is a step riskier even as a volume-growth case. Simply setting the minimum bar for the investment decision at “net annual benefit positive even with zero demand” changes the screening of projects considerably.

The real payback killer is downtime: with overseas plant automation, the difference in maintenance capability alone adds 2.61 years

After demand, the next thing that breaks payback is unplanned downtime. And this one is something almost nobody puts a figure on at the approval stage.

From this model’s production capacity, here is the value of one hour of downtime.

• Hourly output = 3,600,000 / 4,800 = 750 units/h
• Lost gross margin per hour of downtime = 750 x 1,200 = 900,000 VND/h

Note: the annual downtime figures of 120 hours and 380 hours are assumptions in this article. They are not measured values.

A rule within the model (to prevent double counting)

This is easy to get wrong, so let us state the rule of the calculation explicitly. The annual benefit of 1,159,560,000 VND above is the figure after Case 1 (120 hours of downtime per year) has been factored in. In other words, the 108,000,000 VND lost across those 120 hours is treated as already reflected in the benefit.

Case 2 therefore only means deducting a further 234,000,000 VND from the 7.19-year side. Do not deduct the whole 342,000,000 VND. That would deduct the 120 hours twice.

Check: 877,920,000 – 234,000,000 = 643,920,000 VND / 6,312,000,000 / 643,920,000 = 9.80 years

Note: the chart does not include a bar for 62.92 years (the case where the volume-growth premise breaks down), because the scale gap would flatten the other three. The figures are as given above.

“Same machine, different capability” alone is worth 2.61 years

Between Case 1 and Case 2, not a single machine differs. Only these three things differ.

• Whether key spares are held locally (servos, reducers, hands, vision system parts)
• Whether first-line maintenance staff are present on site
• Whether recovery decisions can be made locally (rather than waiting for an engineer to arrive from outside the country)

The difference on those three points becomes 260 hours of downtime a year, 234,000,000 VND in money, and +2.61 years in payback. Taking that same 2.61 years out through a price reduction on the equipment would require cutting the investment by 6,312,000,000 – 643,920,000 x 7.19 = roughly 1,682,000,000 VND (about 26.7% of the investment), which is not realistic. Through the design of the maintenance capability, however, it can be taken out.

In practice, moreover, these 2.61 years often arise as the result of “cutting the maintenance contract and the spares to buy the machine cheaply.” Cutting the denominator and breaking the numerator is the most frequent failure mode.

Downtime you have not measured cannot be used as a negotiating position

One more point. Whether it is 120 hours or 380 hours a year cannot be discussed unless it is measured. At many sites, unplanned downtime ends up as a note in the remarks column of the daily report saying “minor stoppages occurred.” If a 15-minute stoppage happens three times a day, that is 225 hours over 300 days. Converted at this model’s rate of 900,000 VND/h, that alone amounts to a loss equivalent to 202,500,000 VND.

How to capture stoppages, and at what granularity to record them so that they become a negotiating position on maintenance capability, is covered in equipment monitoring at plants in Vietnam. The discussion of payback becomes realistic from the day you start measuring downtime.

Wage increases in Vietnam and automation: Decree 293 and the risk of calculating on a static hourly rate

The numerator set out above is calculated on today’s wages. Let us now add the time axis.

Vietnam’s statutory minimum wage was raised by an average of 7.2% from 1 January 2026 under Decree 293/2025/ND-CP. By region, the figures are as follows.

This is the first January effective date in six years, since 2020. Because the revision now falls at the start of the year, the relationship with how the annual budget is put together needs to be checked.

Thailand’s minimum wage, meanwhile, moved to 400 baht per day across all industries in Bangkok from 1 July 2025 (up from 372 baht before the revision), with a national range of 337 to 400 baht, and it has remained unchanged into 2026.

The 67.5% is today’s ratio, not a fixed value

This is the key point of this section. The ratio of 102,060,000 / 151,200,000 = 0.675 derived earlier is the value on the assumption that wages in both countries are frozen at a point in time. If the Vietnamese side rises every year while the Thai side stays flat, that ratio narrows.

The labour-saving payback of 19.09 years therefore needs a footnote as well. That 19.09 years is the simple payback on the assumption that wages stay at 7,000,000 VND/month indefinitely. In reality the numerator grows year by year, so the actual payback moves in the direction of being shorter than 19.09 years.

That is not, however, grounds for writing “so labour saving is fine after all.” There are two reasons.

1. You cannot fix the rate and the number of years of increase as a settled value in an investment committee. A payback figure that builds in future wage growth can be made as short as you like depending on the growth rate assumed. Even if it passes the approval, accountability becomes vague if the assumption turns out to be wrong.
2. Wages rise for the case where you do not automate as well. Wage growth is not a tailwind only for the automated side; it simultaneously pushes up the cost of running that process manually. Both sides of the comparison move, so it is not fair to move only one side and claim that payback has improved.

This article presents 19.09 years as a static value in order not to build in future wage growth to suit the argument. Put the other way round: if a calculation at your own site shortens the payback by assuming “wages rise by a certain percentage every year,” be sure to check what the figure becomes once that growth rate is removed. Changing a single assumption about the growth rate easily moves the payback by several years.

The right way to use wage growth as a reason for automation

“We are automating because labour costs in Vietnam are rising” is, in itself, a sound problem statement. When translating it into an investment decision, though, put it in this form.

• Produce the payback on today’s static wages (in this model, 19.09 years for labour saving and 7.19 years for volume growth)
• State as a direction, not a number, which design becomes more favourable as wages rise
• Do not build a specific growth rate into the approval figures, or, if you do, put it in a separate table as a sensitivity analysis

Keep that order and the wage discussion can be used without distorting the investment decision.

Electricity assumptions: the range of industrial tariffs, and treating power as a precondition

This model assumes industrial electricity in Vietnam at 2,300 VND/kWh. Let us check the basis for that assumption and its range.

In Vietnam, the average retail electricity price was raised by 4.8% to 2,204.07 dong per kWh (excluding value added tax) from 10 May 2025. Tariffs applied to the industrial sector fall in a range of 1,146 to 3,640 VND/kWh, varying by voltage class and time of day. Combined with the October 2024 revision, that is a rise of 9.8%, and as the fourth revision since 2023, an increase of 18.2% against the start of that year.

The 2,300 VND/kWh used in this article is an assumed value placed within that range of 1,146 to 3,640 VND/kWh. The tariff actually applied varies with contracted voltage, time of day and location, so please substitute the effective rate on your own invoice.

Doubling the electricity tariff does not change the conclusion

The point to hold on to here is how much influence electricity has on payback. Of this model’s running costs of 281,640,000 VND/year, electricity is 60,720,000 VND and maintenance and spares are 220,920,000 VND. Maintenance is larger than electricity.

Movements in the electricity tariff are therefore not a primary driver of payback. Set against the downtime difference of 234,000,000 VND/year described above, the place of electricity at 60,720,000 VND/year becomes clear. The money is larger in the capability to reduce downtime than in the effort spent negotiating the electricity tariff. That is the order of priority this model points to.

Even so, write electricity into the contract as a precondition

Separately from the size of the amounts, electricity has another significance: when it stops, production goes to zero.

An automation cell is more sensitive to power quality than a manual process. A voltage dip that puts the robot into an alarm stop and costs 30 minutes for homing and resetting does not appear on the electricity invoice. What it appears in is the downtime discussed above. At this model’s rate, one hour of downtime costs 900,000 VND.

What needs checking about electricity at the design stage is therefore not the tariff but these three points.

The third item, measurement, is needed so that calculations like the ones in this article can be checked against reality later. This model assumes 5.5 kW x 4,800 hours = 26,400 kWh/year, but whether that was right can never be known unless it is measured at cell level. For how to view equipment status and power consumption on the same time axis, see equipment monitoring at plants in Vietnam, cited above. Downtime and power are two sides of the same measurement.

Lowering expectations on incentives: who Decree 182/2024 is for, and what import duty exemption actually affects

When the payback stretches out, the next hope to appear is that an incentive scheme might close the gap. Lowering expectations here in advance makes the later disappointment smaller. What follows is general information as of August 2026.

Decree 182/2024/ND-CP (Investment Support Fund)

Decree 182/2024/ND-CP, effective 31 December 2024, sets out provisions on the Investment Support Fund. The outline of the scheme is as follows.

• It is aimed mainly at companies subject to the global minimum tax (consolidated revenue of EUR 750 million or more)
• The company must operate in a high-tech field designated by the government, hold an IRC (Investment Registration Certificate) and conduct actual business
• Support covers up to 50% of initial investment costs, plus subsidies for annual operating expenses

Read those conditions and many mid-sized and smaller Japanese-affiliated manufacturing sites fall outside the main target. The threshold of EUR 750 million in consolidated revenue is tied to the global minimum tax framework, and the scheme is designed with large multinationals in mind. You cannot take the general statement that “Vietnam has generous support schemes” and apply it directly to the payback of a single cell.

Import duty exemption (Law 107/2016/QH13 Article 16 and Decree 134/2016/ND-CP Article 14)

What is relevant to a wider set of companies is exemption from import duty. Under Article 16 of Law 107/2016/QH13 on import and export duties and Article 14 of Decree 134/2016/ND-CP, there are provisions exempting import duty on imported equipment forming fixed assets, as well as on raw materials, supplies and components for production.

The body of an automation cell may in many cases qualify. Care is needed, though, about how this feeds into payback.

The investment of 6,312,000,000 VND in this article is set as the amount that finally leaves the company, including the effect of any such exemption. You therefore cannot shorten the 19.09 years or the 7.19 years by adding the effect of import duty exemption afterwards. That would count the same effect twice.

Where exemption works is strictly at the stage of fixing the denominator. When you take quotations, check

• how items split between those eligible for exemption and those not
• what procedures the exemption application requires and how long it takes
• how much the investment increases if the exemption is not granted

and place the amount fixed as a result of that in the denominator. That is the correct order.

⚠️ The above is general information as of August 2026. Please be sure to confirm applicability with the investment registration authority and your tax advisor. This article is not tax or legal advice.

Do not build a payback out of incentives

In summary, a realistic expectation of the incentive schemes looks like this.

Incentives are not a tool for pushing through a project that does not stand up. What turns 19.09 years into 7.19 years is not a scheme, it is the design of the numerator.

Five translations to apply when bringing a Thai design to automation in a Vietnamese plant

Let us turn all of this into a practical procedure. Here are five things not to carry across unchanged when rolling out to Vietnam an automation design that succeeded at a Thai site.

Translation 1: swap the numerator

The single most important point. With the same equipment, the labour-saving design gives 19.09 years and the volume-growth design 7.19 years, a gap of 2.66 times. Before selecting equipment, check whether this process has gross margin that can be increased and defects that can be removed. If it has neither, deciding to take that process out of scope for this round of automation is a legitimate outcome. This is not a conclusion that you should not automate in Vietnam. It is the conditional conclusion that putting labour saving in the numerator will not pass.

Translation 2: do not be greedy about the headcount reduction

The warning above about not mixing the 6 people of the labour-saving design with the 2 of the volume-growth design is a point about the arithmetic and, at the same time, a point about operations. A plan to remove 6 people while raising output by 18% does not hold together as a staffing plan in the first place. In the volume-growth design, moving the freed-up people to the surrounding processes for the extra volume and to the next line is part of the design. Put it through approval not as a plan to cut people but as a plan to make more with the same number of people.

Translation 3: decide the maintenance capability at the same time as the equipment

The +2.61 years described above does not appear in an equipment specification. At the quotation comparison stage, make sure the following are lined up side by side.

• Whether first-line response staff are present on site (own employees, or a local system integrator stationed there)
• The scope of key spares held locally, and the cost of holding them
• The target time to recovery (within how many hours the line is back up)
• The lead time if an engineer has to be called in from outside the country

Comparing only the machine price and placing the order with those four items blank is what ends up most expensive.

Translation 4: back the demand with numbers

The volume-growth design only stands on demand. If the output does not sell, it is 62.92 years. Set out in the approval where the additional 648,000 units are going.

Translation 5: do not get the ordering wrong

Finally, sequence. What you often see in practice is the order “decide the robot model, then calculate the payback.” Do it that way and, when the resulting figure fails to meet the criteria, you start wanting to add benefits after the fact. That is where double counting happens.

The correct order is (1) select the process, (2) measure the numerator (the gross margin you can add and the defects you can remove), (3) decide the ceiling on the investment required, and (4) select equipment within that range. How to sequence a capital investment plan, where to take quotations and what to decide first is set out in how to build a capital investment plan. Take quotations before measuring the numerator and the price will drag the decision along with it.

The payback figures at a glance (this article’s conclusion table)

All five rows use the same equipment and the same investment of 6,312,000,000 VND. Only the numerator differs.

FAQ: common questions about automating a plant in Vietnam

Does factory automation in Vietnam pay back?

It depends on what you put in the numerator. In this article’s model, putting labour saving in the numerator gives a net annual saving of 330,720,000 VND and a payback of 19.09 years (the same equipment in Thailand gives 597,576,000 VND and 10.56 years, a gap of 1.81 times). Putting added volume and quality in the numerator gives a net annual benefit of 877,920,000 VND and 7.19 years. The reason it fails to pay back is not “because it is Vietnam.” It is that the denominator is the same dollar-denominated amount (6,312,000,000 VND = USD 240,000) while only the numerator, labour cost, comes to 67.5% of the Thai level. Note that the volume-growth design depends on demand: if the output does not sell, it falls to a net annual benefit of 100,320,000 VND and a payback of 62.92 years. Whether you can put both ends of that range into the approval document is the deciding factor.

Can AGVs be used in Vietnam?

They can, but the investment case for an AGV has the same structure as this article. What an AGV displaces is mainly the hours of people engaged in transport, so putting labour saving straight into the numerator shrinks the numerator in Vietnam and stretches the payback. If you want it to pay back, the practical approach is to measure the benefit an AGV creates as the reduction in line stoppages caused by waiting for transport. At this model’s rate, one hour of downtime causes a lost gross margin of 900,000 VND. Measuring “how many hours of stoppage are avoided” rather than “how many people are removed” changes the evaluation of an AGV. Note also that floor flatness along the travel route, aisle widths and mixed operation with people depend heavily on the local building conditions, so measuring the actual layout comes before selecting a model.

Vietnam or Thailand: which should be automated first?

If you compare them on the same labour-saving design, the numbers say Thailand first. In this model, Thailand is 10.56 years against Vietnam’s 19.09 years, a gap of 1.81 times. That is because the numerator, total labour cost, is 102,060,000 VND/year in Vietnam against 151,200,000 VND/year in Thailand (Vietnam at 67.5% of Thailand). But that is only the ordering when labour saving sits in the numerator. If the Vietnamese site has room to increase volume and reduce defects and the demand is verified, the volume-growth design at 7.19 years is faster than Thailand’s labour-saving design at 10.56 years. Decide by the order of processes, not the order of countries.

What incentive schemes are available for automation in Vietnam?

As general information as of August 2026, Decree 182/2024/ND-CP (Investment Support Fund, effective 31 December 2024) is aimed mainly at companies subject to the global minimum tax (consolidated revenue of EUR 750 million or more) that operate in government-designated high-tech fields, hold an IRC and conduct actual business, and provides for support of up to 50% of initial investment costs plus subsidies for annual operating expenses. Many mid-sized sites fall outside the main target. More broadly relevant is import duty exemption: under Article 16 of Law 107/2016/QH13 on import and export duties and Article 14 of Decree 134/2016/ND-CP, imported equipment forming fixed assets and raw materials and components for production may be covered. Note, however, that exemption is something to build in at the stage of fixing the investment (the denominator), and adding it on top of a payback already fixed is double counting. Please be sure to confirm applicability with the investment registration authority and your tax advisor.

Who should you talk to about automating an overseas plant?

Before choosing who to talk to, having three numbers in hand makes the conversation faster. (1) the annual output and unit gross margin of the target process (3,600,000 units and 1,200 VND/unit in this model), (2) the current defect rate and the loss per defective unit (2.4% and 3,800 VND in this model), and (3) annual unplanned downtime (120 hours in Case 1 and 380 hours in Case 2 in this model). With those three, you can have a discussion about the numerator with any system integrator. Without them, the conversation can only be about models and prices. When selecting who to work with, make sure to include whether they can hold first-line maintenance capability and a spares stock locally as a condition. In this model, the presence or absence of that capability alone accounts for 234,000,000 VND a year and +2.61 years of payback (7.19 years to 9.80 years).

Is labour saving meaningless in Vietnam?

We cannot write that it is meaningless. What can be written is that putting labour saving alone in the numerator makes it hard to clear investment criteria. The labour-saving design in this model gives a net annual saving of 330,720,000 VND and a payback of 19.09 years, which will not pass the investment criteria of most Japanese-affiliated companies. There are, however, situations where labour saving is effective. First, processes where hiring is difficult (heavy items, high temperatures, dust and so on): if people cannot be recruited at all, the size of the saving is not the deciding factor. Second, processes with high turnover where training costs accumulate. Third, even within the volume-growth design, the labour cost reduction of 204,120,000 VND (2 people) is included in the benefit. The point is not to discard labour saving, but not to make labour saving the numerator on its own.

Summary

• The denominator is the same dollar-denominated amount; only the numerator is 67.5%. The investment in this model is 6,312,000,000 VND (= USD 240,000 = 7,890,000 THB) and identical in both countries, while total operator labour cost is 102,060,000 VND/year in Vietnam against 151,200,000 VND/year in Thailand, a ratio of 0.675.
• Same equipment, same reduction of 6 people, and the payback is still 1.81 times longer. The labour-saving design gives Vietnam a net annual saving of 330,720,000 VND and 19.09 years, against Thailand at 597,576,000 VND and 10.56 years.
• A robot density of 31 units per 10,000 employees (up 28% year on year) is not a technology gap. The distance from the global average of 132 units and Asia’s 131 should be read as the result of an investment test with labour saving in the numerator failing to pass.
• Swap the numerator to gross margin and it becomes 7.19 years. The volume-growth design takes an annual benefit of 1,159,560,000 VND (additional gross margin 777,600,000 plus defect reduction 177,840,000 plus labour cost reduction 204,120,000), deducts running costs of 281,640,000 VND for a net annual benefit of 877,920,000 VND, and pays back 2.66 times faster than the labour-saving design.
• But the volume-growth design stands only on the assumption that the output sells. Without demand, the additional gross margin of 777,600,000 VND goes to zero, leaving a net annual benefit of 100,320,000 VND and a payback of 62.92 years, that is, no real payback.
• The real payback killer is downtime. The gap between local maintenance with spares in stock (120 hours a year) and importing spares as needed with engineers flown in (380 hours a year) is 234,000,000 VND a year. Payback deteriorates from 7.19 years to 9.80 years, +2.61 years.
• Do not build a payback out of incentives. The main target of Decree 182/2024 is high-tech companies with consolidated revenue of EUR 750 million or more, and import duty exemption is something to use in fixing the denominator, not to add on top of a payback already fixed. Please confirm applicability with the investment registration authority and your tax advisor.

A note on double counting. Design 1 (labour saving) and Design 2 (volume growth) in this article are two separate ways of using the same equipment costing 6,312,000,000 VND, and their benefits must not be added together. Three further points. (1) The material and processing variable costs of the additional 648,000 units are already included in the definition of the unit gross margin of 1,200 VND/unit (selling price minus variable cost). Do not deduct them again. (2) The volume-growth design removes 2 people (204,120,000 VND). Do not carry over the 6 people of the labour-saving design (612,360,000 VND). (3) The annual benefit of 1,159,560,000 VND is the figure after Case 1 (120 hours of downtime a year) has been factored in. Case 2 only deducts a further 234,000,000 VND from it; deducting the whole 342,000,000 VND would deduct the 120 hours twice.

Note that of the figures in this article, only those on minimum wages, electricity tariffs, robot density, market size and the regulatory schemes are based on externally published information. The investment, output volume, unit gross margin, defect rate, downtime, electricity tariff, exchange rates and the 18% capacity gain are all model values used in this article, neither measured results nor market rates.

Starting from a conversation before you rebuild the payback case

Swap in your own output volume, unit gross margin, defect rate and downtime, and the ranking of the five payback figures listed here can change. Unit gross margin in particular, and whether demand is verified, move the conclusion considerably. TOMAS TECH works on factory automation, robot deployment and equipment monitoring for Japanese-affiliated manufacturers in Thailand and Vietnam, and we also take enquiries at the stage of “we have not chosen a model or a budget yet and want to work out whether the numerator even stands up in our process.” If you would like to go through how to set the gross margin and the downtime for a target process before taking quotations, please get in touch through our contact page.

References

• 最低賃金は2026年1月に平均7.2％引き上げの正式決定 (official decision to raise Vietnam’s minimum wage by an average of 7.2% in January 2026, JETRO Business Brief, November 2025)
• 2026年1月に最低賃金を平均約7.2％引き上げ (minimum wage up by an average of about 7.2% in January 2026, JILPT, the Japan Institute for Labour Policy and Training)
• 電気料金を4.8％引き上げ、7カ月ぶりの改定 (Vietnam raises electricity prices by 4.8%, the first revision in seven months, JETRO Business Brief, May 2025)
• バンコクの最低賃金、日額400バーツに引き上げ (Bangkok’s minimum wage raised to 400 baht per day, JETRO Business Brief, July 2025)
• Smart factories or bust: decision time in 2026 (Vietnam Investment Review)
• Vietnam Industrial Robotics Market 2025-2031 (Ken Research)
• Robot Density Surges in Europe, Asia and the Americas (International Federation of Robotics, World Robotics 2025)
• 政令182/2024/ND-CP：ベトナム投資支援基金に関する新たな規定 (new provisions on Vietnam’s Investment Support Fund under Decree 182/2024/ND-CP, Manabox Vietnam)
• 外資に関する奨励 (incentives for foreign investment, JETRO country information for Vietnam)
• Foxconn accelerates investment in Vietnam (The Investor / VAFIE)