AGV Price and Total Cost — AMR Break-Even Is Set by Layout Changes

“How much does one AGV cost?” We field that question from Japanese-owned plants in Thailand almost every month. Yet searching for AGV prices only turns up the going rate for the vehicle itself, which tells you nothing about how many times larger your own quotation will be. The reason is straightforward. A quotation for an automated guided vehicle puts a price on the route, not on the unit.

There is a second argument that surfaces in every comparison. Which is cheaper, a guided AGV or an autonomous AMR? The usual explanation says that AMRs pull ahead once you add more units, but when we line up the total cost of real projects, that is not what happens. What decides whether the totals cross over is not the number of vehicles but how many times you change the layout.

Rather than listing vehicle prices, this article takes you all the way to a purchasing decision using a single division problem — the gap in upfront cost divided by the cost gap per layout change. From there we set out what changes when you buy in Thailand, and what the buyer needs to settle before requesting competing quotations.

Why “how much is one AGV” has no answer

The vehicle price is public. The total still is not.

Start with the published figures. Pulling together the Japanese market data, an AGV vehicle is generally quoted at JPY 2 million to 5 million per unit, a forklift type (AGF) at around JPY 15 million, and an autonomous AMR at JPY 1.5 million to 5 million.

None of those numbers is wrong. The problem is that comparing sheet after sheet of them never brings you closer to a total. OptiMax’s 2026 breakdown of logistics robot deployment costs contains one decisive sentence. It is not unusual for the cost of integrating with an ageing core system to exceed the price of the vehicles themselves.

In other words, on a project that installs five vehicles at JPY 2 million each, it is perfectly normal for that JPY 10 million of hardware to come to less than half the total. A price table tells you the cost of the single most legible line item. It does not tell you which line items actually move the money in a quotation.

The six line items that actually move

When we line up competing quotations for automated guided vehicles in Thailand, the six items below are where the numbers swing wildly from vendor to vendor. Put the other way round, quotations that do not share the same assumptions on these six cannot be compared at all.

The column to watch is the one on the right. Of the six items, only the vehicle scales directly with unit count, with charging equipment and commissioning contributing in part. The remaining three are set independently of unit count by route length, by the number of systems you connect, and by the number of zones. “How much is one AGV” has no answer because most of the quotation is driven by variables other than unit count.

What makes the fourth item awkward for the buyer is that nobody inside the company can estimate host system integration. How old are the existing WMS and MES, and which interfaces can be opened up? The figure that would cover that investigation is carried through the proposal stage marked “to be quoted separately”, and later reappears at a level comparable to the vehicle price. For the wider picture of AGV deployment, see our basic guide to AGV and AMR deployment.

Guidance method decides the structure of the price

What gets locked in before you choose a vehicle is the guidance method. This is where the structure of the price is settled, not its amount. What is paid once at the start, and what is paid again every time you touch the layout? That split feeds directly into the break-even discussion later.

Magnetic tape guidance

Magnetic tape is laid on the floor and the vehicle follows it. The on-board configuration is simple, so the vehicle is cheap, and the method itself is mature. The trade-off is that the route is physically fixed to the floor. Move equipment, reroute an aisle, reorganise a process, and each time you have to peel the tape up, lay it again, and re-verify travel and stop positions.

Tape wear is the other factor that bites on the shop floor. In aisles crossed by forklifts, or stretches where pallets are dragged, the tape degrades and read failures start causing stoppages. This is not an accident but routine consumption, and it lands on the maintenance workload.

Laser guidance

Reflectors are mounted on walls and columns, and the vehicle’s laser scanner determines its own position from them. Nothing is laid on the floor, so the method is resistant to surface wear, and because the route is held in software it is more tolerant of change than magnetic tape.

The premise, however, is that the reflectors stay visible. Stacked inventory hiding a reflector, or a layout change that alters sightlines column by column, means repositioning and re-surveying the reflectors. Understand it as a method with no floor work but with environmental preparation still required.

SLAM (AMR)

The vehicle scans the surrounding geometry to build a map and drives autonomously by matching against it. Because nothing is installed on the floor, it is in principle the lightest method to follow a layout change — you rebuild the map and you are done.

In exchange, the vehicle is more expensive because sensors and computing sit on board. And “no guide path” does not mean “no preparation”. Wide featureless spaces, temporary staging areas whose appearance changes constantly, mirrored surfaces and transparent partitions all reduce map stability. Map creation and on-site tuning at deployment always occur, both as cost and as workload.

When the money lands, across the three methods

The third row of that table, “on layout change”, is the only cost that recurs. The rest is essentially one-off. A cheap vehicle and the amount you pay at every change enter into a trade-off the moment you pick a method. Route design itself is covered in detail in how to approach AGV layout design.

The belief that AMRs fall under a separate standard is wrong

Let us kill off a common misconception at this point. ISO 3691-4:2023, the safety standard for driverless industrial trucks, lists guided AGVs and autonomous AMRs within the same standard. The claim that “AMRs are a new category so a different standard applies” is not correct.

More importantly, what the standard looks at is the system, not the vehicle in isolation. Annex A of the same standard covers preparation of the operating area, which in practice means that the condition of the floor and the securing of aisles fall within the buyer’s scope of responsibility.

That boundary feeds straight into price. Whether conditions such as “unevenness in the floor is to be corrected by your company” and “this aisle width is to be secured” are written into the assumptions of the quotation determines how large the extra work billed later turns out to be. The cheaper a quotation looks, the more likely those assumptions are missing.

The crossover comes from change frequency, not unit count

Here is the heart of it. Where do the totals for a guided AGV and an autonomous AMR swap places?

Model assumptions (these are assumed figures, not an actual quotation)

Assume a Japanese-owned plant in Thailand installing a five-vehicle transport system. The figures below are assumed values placed here for explanation. They are neither a quotation for a specific project nor a market rate. Real figures vary widely with layout, existing systems and operating conditions. What to look at here is not the amounts themselves but how they line up.

Note that charging equipment itself is needed with either method. The second row of this table bundles the AGV-side guide path installation together with the floor and electrical work it entails, and assumes that no equivalent work arises on the AMR side. In a real quotation the cost of charging stations and spare batteries lands regardless of method, so check that alongside the running costs discussed later.

The upfront cost is 1,100,000 baht higher for the AMR. Up to this point the picture matches the quotations most plants actually receive. And most projects conclude right here that the AGV is cheaper.

Only three items create the gap

Break down where that 1,100,000 baht comes from.

• Vehicle difference — the AMR is 1,500,000 baht more expensive
• Guide path installation and charging equipment work — 700,000 baht lands on the AGV side only
• Map creation and on-site tuning — 300,000 baht lands on the AMR side only

1,500,000 − 700,000 + 300,000 = 1,100,000 baht, which matches the table total.

Now note the 1,000,000 baht for host system integration. It lands equally on both methods, so it has no effect whatsoever on the gap. There is a reason it is still in the table. This is exactly where the disappointment of “we chose the cheaper hardware and the total barely moved” comes from.

Seeing a 1,500,000 baht difference in the vehicles feels like “the AMR costs 1.5 times as much”, but in total terms it is 5,800,000 against 4,700,000 — a gap of just over 20 percent in ratio. Because costs that land regardless of method push the total up, the impact of the method choice on the total is smaller than the vehicle price suggests. The same structure applies to automated warehouses, as covered in the price structure of automated warehouses.

There is only one division to do

Next, set out what happens when the layout is changed once. These are assumed figures as well.

The gap per change is 200,000 − 40,000 = 160,000 baht. At every change, the AGV side alone pays that much extra.

All that is left is to divide.

1,100,000 ÷ 160,000 = 6.875, so roughly 6.9 changes

At around 6.9 layout changes, the cumulative totals for the AGV and the AMR swap places. Applied to the pace at which a plant actually changes, that comes out as follows.

The numerator of the division is the gap in upfront cost, and the denominator is the cost gap per change. Do not put a total into the denominator. Get that wrong and the order of magnitude changes.

Adding units does not change the direction of the crossover

This is the part we most want to convey in this article.

You will see it explained that “AMRs become advantageous as the number of units grows”, but that is not what happens in this model. Add units and the vehicle gap — the amount by which the AMR is more expensive — widens in proportion as well. That is, the numerator, the upfront gap itself, grows. The denominator, meanwhile, is the difference in layout change cost, which is set by tape laying distance and the scope of re-verification, in other words by the scale of the layout, and so does not grow at the same multiple as the unit count.

The result is that adding units does not shrink the number of changes needed to cross over. If anything it pushes the crossover further away. At the very least, it does not run in the direction of “more units make the AMR advantageous”.

What the unit count determines is the size of the upfront gap, meaning the distance to the crossover, not its direction. The direction is set by one thing only, which is how many times you touch the layout.

That difference changes where the evaluation starts. Beginning from “how many units do we need” produces nothing you can use to judge the method. Begin from “over the next three to five years, how many times do we plan to change our product mix and layout?” Once that is settled, the division is a single line.

And the party who knows the answer to that question best is the buyer, not the vendor. The timing of the next model changeover, plans for expansion, whether there is any intention to move to mixed-model production. Request competing quotations without presenting that, and every vendor will hand you the cheapest configuration built on an unstated assumption of no change.

Three running costs that are not in the quotation

Once you have the upfront cost and the change cost pinned down, what remains is the money that leaves every year during operation. This is rarely written up in one place in a proposal, and is the classic case of surfacing after the internal approval has gone through. Check the following three.

1. Batteries and charging (utilisation is set by charging, not by unit count)

The “number of units required” is worked back from the transport capacity of a single vehicle. Yet that calculation sometimes leaves out the time spent standing still on charge.

For an AGV with a lithium-ion battery, the rule of thumb is three to six hours of continuous operation and around 1.5 hours for a full charge. On a site running around the clock, those conditions mean four battery swaps a day. The swap itself may be quick, but who performs it, how many spare batteries you hold, and how many charging stations you place and where, are all questions of money and floor space.

The choice of battery type matters too. Lithium-ion offers a cycle life of over 2,000 cycles, said to be three to five times that of lead-acid. The upfront cost is higher, but because the replacement interval lengthens, it is an item where the totals can reverse on equipment with a long service life.

The important point is that spare batteries and charging stations are costs that sit outside the unit count. They appear neither in the unit calculation nor in the vehicle quotation, yet they bear directly on utilisation. Always check whether charging wait time is built into the calculation behind “five units will be enough”.

2. Maintenance (5 to 10 percent of the vehicle price per year)

The rule of thumb for maintenance is 5 to 10 percent of the vehicle price per year. For a JPY 3 million vehicle that works out at JPY 150,000 to 300,000 a year. On a five-vehicle configuration, five times that leaves the business every year.

What to check in a maintenance contract is the content rather than the amount. Frequency of scheduled inspections, whether consumables such as drive components and batteries are included, response time on breakdown, availability of a replacement machine. On projects in Thailand, one more question is added — where does the engineer travel from? A contract that assumes engineers are called in from Japan effectively loads travel time and accommodation onto the maintenance cost.

3. Software updates and rebuilding the host connection

The third item, easily overlooked, is on the software side. It is not only vehicle firmware. The interface connecting the fleet management system to the host system also becomes subject to updates.

The thing to know here is VDA 5050. This is a specification defining the communication interface between the fleet management system and AGVs or AMRs, drawn up by VDA and VDMA under the supervision of the IFL at KIT (Karlsruhe Institute of Technology). VDMA and VDA formally adopted version 3.0.0 on 17 February 2026, and VDA announced its publication on 20 April 2026, introducing a zone concept for handling robots with a higher degree of autonomy, the addition of CRITICAL and URGENT error levels, and a mechanism for route sharing.

Why does this belong in an article about price? Because the more the interface is standardised, the greater the chance that you can replace a vehicle with one from a different manufacturer. And being able to replace it is precisely what gives you negotiating power on price at the next expansion or renewal.

Conversely, choose a configuration where fleet management and vehicles are tightly coupled over a proprietary protocol, and every expansion has to be bought from the same manufacturer. The configuration with the cheapest opening quotation becomes the most expensive purchase five years on. That pattern is not rare.

The countermeasure is simple. Require the purchase specification to state the supported interface specification and its version. A proposal that says only “VDA 5050 compliant” tells you nothing about which version and how much of it is implemented. Have them write the version number and, if there are unsupported functions, a list of those as well.

The option of not buying (leasing and RaaS)

Where the upfront cost will not clear internal approval, leasing and service-based models (RaaS) are options. In Japan, the AGV Handbook lists monthly rates for a five-year lease on the ZMP CarriRo series.

Laid out like that, it looks as if “JPY 80,000 a month buys you automated transport”. But anyone who has read the first half of this article will already see the catch. What is included in that monthly rate is the vehicle. Host system integration and floor work are charged on top.

In terms of the assumed model above, the only row the monthly rate replaces is “vehicles”. Construction, map creation and host system integration do not disappear because you leased. Make your comparison on the premise that the monthly rate alone does not put the system into operation.

That said, there are genuinely situations where leasing and RaaS fit. A start-up phase where transport patterns have not settled, a fixed-term response to a production increase, a stage where you want to validate the method itself. While change frequency is unpredictable, choosing not to carry the equipment as an asset is a rational judgement.

Conversely, for a plant with low change frequency that knows it will run the same routes for five or ten years, purchasing settles at a lower total. The deciding axis here is the same one — change frequency.

What changes when you buy in Thailand

So far the discussion has been about method and cost structure. If your plant is in Thailand, some of the underlying premises differ from Japan.

Labour savings alone will not pay it back

Thailand’s minimum wage was revised on 1 July 2025, bringing Bangkok to 400 baht per day, with a national range of 337 to 400 baht per day. And that level has been held unchanged for 2026 as well. (Note that these are daily figures. Misreading them as hourly rates throws a payback calculation out by an order of magnitude.)

At that level, a payback calculation grounded solely on cutting a few transport operators simply does not stand up. It is a conclusion we write repeatedly — automation in Thailand has to be justified by effects other than the substitution of labour cost.

For transport, that means not “fewer people” but “fewer trips” and “less stagnation between processes”. Once transport timing is stable, work in progress ahead of each process falls and the interruptions caused by emergency deliveries disappear. The effect shows up in lead time and work-in-progress inventory rather than in headcount. The relationship between labour cost and automation is set out in detail in how to face rising labour costs in Thailand. For measuring the effect of transport, see improving in-plant logistics.

Understand BOI incentives together with their conditions

Under the Thailand Board of Investment (BOI) measure for Smart and Sustainable Industry, the baseline for automation and robot deployment is a three-year corporate income tax exemption capped at 50 percent of the investment. Further, where at least 30 percent of the value of the upgraded machinery is sourced from the Thai automation industry, the cap is raised to 100 percent of the investment. An exemption from import duty on machinery is also provided.

Always take this together with its conditions. Write an internal approval on the strength of hearing only “BOI gives you a 100 percent exemption” and the premise collapses later when the condition is not met. It does not become 100 percent unconditionally. The existence of a local sourcing ratio requirement means that where you buy what, at the supplier selection stage, affects the size of the incentive. It is a point to settle before requesting quotations.

Note that eligibility and procedure differ from project to project. For an actual application, work from BOI primary sources and confirmation from a specialist.

Local maintenance capability and spare parts stock

Once you are operating in Thailand, what ends up mattering is who turns up when it stops. There are three things to check.

First, are there engineers inside Thailand? If the assumption is calling them in from Japan, recovery time becomes unpredictable once visas and travel arrangements are factored in. Second, are spare parts stocked inside Thailand? If wear and failure-prone parts such as drive wheels, batteries and sensors have to wait on an import, you are down for days at a time. Third, are engineer accommodation costs during commissioning included in the quotation? This tends to be handled as “actual expenses, billed separately”, and contracts get signed with the amount unknown.

How to evaluate this organisational side is also covered in how to choose a robot system integrator.

Six items the buyer settles before requesting quotations

Finally, the practical part. Request competing quotations having settled nothing, and each vendor writes its quotation on its own assumptions. Quotations built on different assumptions cannot be compared no matter how you line up the amounts. Settle the following six on the buyer’s side, then put the same conditions to every vendor.

1. What is transported and in what unit — what, at what weight, in what form (pallet / cart / container). Until the load handling method is settled, the vehicle type cannot be.
2. Planned frequency of layout change — the centre of this article. How many times over the next three to five years do you expect to touch the routes? Without presenting this, vendors will quote the cheapest configuration assuming no change.
3. Operating hours and charging approach — how many hours will it run? Will charging waits be tolerated, or will you operate with battery swaps? This is what changes the number of units required.
4. Scope of host system connection and interface specification — which systems, connected how far? If VDA 5050, require the supported version to be stated.
5. Division of safety responsibility — as Annex A of ISO 3691-4:2023 indicates, preparation of floors and aisles is within the buyer’s scope. Draw the line between what you provide yourself and what you ask the vendor for.
6. Maintenance and spare parts capability — will engineers and stock inside Thailand be part of the requirements? If so, write it into the request for quotation.

A single sheet setting out these six items is enough to put every vendor’s quotation on the same footing. And the gaps the vendors return can be dropped straight into the division in this article.

Frequently asked questions

How much does an AGV cost?

The published market figures put the vehicle itself at JPY 2 million to 5 million per unit, a forklift type (AGF) at around JPY 15 million, and an AMR at JPY 1.5 million to 5 million. That is only part of the total, however. Guide path and floor work, charging equipment and spare batteries, host system integration, safety measures, and commissioning and training are added to it. Host system integration in particular can easily exceed the vehicle price when the counterpart is an ageing core system. Ask not “how much per unit” but “how much for this route and this connection scope”.

Which is cheaper, an AGV or an AMR?

The general structure is that the guided AGV is cheaper upfront and the AMR is cheaper each time you pay for a layout change. Which is cheaper therefore depends on how many times you change the layout. In the assumed model in this article, dividing the 1,100,000 baht upfront gap by the 160,000 baht cost gap per change gives a crossover in total cost at roughly 6.9 changes. For a plant that changes twice a year that is about 3.4 years, and about 6.9 years at once a year. With no change at all over five years there is no crossover. Note again that these are assumed figures for explanation, not an actual quotation.

How should I compare automated guided vehicle manufacturers?

Compare on the following three points rather than on vehicle specifications. First, can they put in writing what happens and at what cost on a layout change? Second, can they state the interface specification and version for the host system (for VDA 5050, how much of v3.0.0 is implemented)? Third, do they have maintenance capability and spare parts stock inside Thailand? Vehicle speed and payload do not differ much between vendors, but these three separate them clearly. And as a precondition for any comparison, the buyer has to issue the same specification sheet to every vendor.

What is included in AGV running costs?

Three things, broadly. Batteries and charging (three to six hours of continuous operation on lithium-ion, 1.5 hours for a full charge, four swaps a day on round-the-clock operation, and the cost of spare batteries and charging stations), maintenance (5 to 10 percent of the vehicle price per year, so JPY 150,000 to 300,000 a year for a JPY 3 million vehicle), and software updates and rebuilding the host connection (modification work to keep up with updates to the interface specification). The third of these often has no figure attached at proposal time, so confirm how it is handled before contracting.

What is different about deploying an AGV in Thailand compared with Japan?

Three things. First, with a minimum wage of 337 to 400 baht per day (400 baht in Bangkok, held unchanged for 2026), a payback calculation grounded solely on labour savings is hard to sustain, so the effect has to be taken through fewer transport trips and less stagnation between processes. Second, the BOI Smart and Sustainable Industry measure offers a three-year corporate income tax exemption capped at 50 percent of the investment, with the cap rising to 100 percent where at least 30 percent of the value of the upgraded machinery is sourced from the Thai automation industry. Third, local maintenance capability and spare parts stock determine recovery time.

Summary

The reason AGV prices have no single answer is that a quotation puts a price on the route rather than on the unit. The vehicle is only part of the total, and host system integration can even exceed the vehicle price.

The totals for a guided AGV and an autonomous AMR do not cross over when you add units. They cross over when the number of layout changes reaches the upfront gap divided by the cost gap per change. On the assumed figures in this article that is roughly 6.9 changes. A plant that changes twice a year crosses over in about 3.4 years, and a plant that makes no changes over five years never does. Unit count only alters the size of the upfront gap. It does not alter the direction of the crossover.

So the first thing to settle is not the number of units but “how many times will we change the layout over the next three to five years?” Settle that, then produce a single condition sheet covering what is transported, operating hours, connection scope, division of safety responsibility and maintenance capability, and request competing quotations on identical conditions. Only once you have done that do the vendors’ figures become comparable.

If you are deploying in Thailand, put two premises at the front of the internal approval — that labour savings alone will not pay it back, and that the BOI incentive comes with conditions.

TOMAS TECH implements production management systems and factory automation and robotics for Japanese-owned manufacturers in Thailand. We are happy to talk at the stage of “we have not decided on a method or a number of units yet” or “we want to know what working back from change frequency would look like”. If you bring your existing layout drawings and the current state of transport, we can start by working through the differences in cost structure between the methods together. You can reach us here.

References

• Market rates for AGV deployment cost (OptiMax, 2026 edition)
• Automated guided vehicle (AGV) handbook, prices by manufacturer
• Differences between AGV and AMR and how to choose (Moriya Sangyo)
• Differences between AGV and AMR (Okamura)
• How to choose an AGV battery (Nabtesco)
• Charging challenges for AGVs running around the clock (Musashi Energy Solutions)
• Bangkok minimum wage raised to 400 baht per day (JETRO, July 2025)
• Held unchanged for 2026 at 337 to 400 baht (Bangkok Shuho)
• BOI Measure for Industrial Upgrades towards Smart and Sustainable Industry
• VDA 5050 v3.0 published (VDA, 20 April 2026)
• VDA 5050 3.0.0 Release (Idealworks, March 2026)