Three quotes for the same process, and the totals differ by more than a factor of two. This happens constantly to plants that have just started looking for a robot system integrator. The gap is not explained by how much discount each firm took on the robot arm. Part feeding, gripper, safety, connection to your production systems, and support after go-live — those are five places where a line gets drawn between what the supplier takes on and what stays with you. Move that line, and the total moves with it. This article sets out the five scope boundaries you should define for yourself *before* you compare prices, written for plants operating in Thailand.
A 2–3x Spread in Quotes Is Not a Price War
Evaluating industrial robots is no longer an unusual thing to be doing. According to the IFR’s World Robotics 2025 report, 542,000 industrial robots were newly installed worldwide in 2024, bringing the operational stock to 4,664,000 units. Asia accounted for 74% of new installations, with China at 295,000 units (54%) and Japan at 44,500. Demand has doubled in a decade, and you are stepping into that market now.
And yet most plants stall the moment they actually have quotes on the table. The reason is simple: the documents in front of them are not in a comparable form.
The Robot Itself Is Only 20–30% of the Total
When you think about the cost of a robot installation, the first number that comes to mind is the price of the arm. In real projects, the robot and its controller account for only 20–30% of the total. The rest is the base frame, safety equipment, the gripper, feed and transfer devices, the control panel, electrical installation, engineering (design, teaching, witnessed testing), and installation on site.
The table below breaks down one of the scenarios used later in this article — a two-robot cell awarded as a single package to a Japanese-affiliated integrator. The basis for these figures is set out in full in the estimate section further down.
| Line item | Amount (THB) | Share |
|---|---|---|
| Robots and controllers (2 units) | 1,800,000 | 19.9% |
| Gripper (EOAT) design and fabrication | 900,000 | 9.9% |
| Feeding, alignment and discharge equipment | 1,600,000 | 17.7% |
| Safety equipment and risk assessment | 750,000 | 8.3% |
| Control panel design/build and electrical work | 850,000 | 9.4% |
| Upstream integration (PLC, production management, data capture) | 700,000 | 7.7% |
| Engineering (design, teaching, witnessed testing, documentation) | 1,900,000 | 21.0% |
| Installation, commissioning and training | 550,000 | 6.1% |
| Total | 9,050,000 | 100% |
Note that engineering (21.0%) is a larger line than the robots themselves (19.9%). Robots are catalogue products; buy one from any integrator and the price difference is limited. What actually moves the total is how many person-months of design, fabrication and tuning are stacked around the robot — and that is determined by how much of the job the supplier agreed to own.
What Varies Is Scope, Not Price
When three quotes differ by a factor of two, what is usually happening is one or more of the following.
- Company A designed the part-feeding equipment. Company B priced on the assumption that parts arrive pre-aligned.
- Company A designed grippers covering all four part types. Company B designed for one representative part and wrote that the rest would be handled by changing fixtures — without saying who builds those fixtures.
- Company A included performing and documenting the risk assessment. Company B included the fencing only, leaving the assessment and the sign-off with you.
- Company A included feeding production data into your production management system. Company B stopped at a dry contact output for the completion signal.
- Company A assumed handover of program source and a full drawing set. Company B said nothing about handover at all.
None of them is lying. They simply drew the line in different places. And if you have not specified where that line goes, each firm will draw it wherever it is most comfortable and least risky for them. The result is three quotes that cannot be compared.
The “Cheap” Quote Is Not Cheap — It Is Narrow
The worst outcome here is picking the lowest quote and then discovering that everything outside the line has come back to you. You source the feeding equipment. You build the fixtures. You run the risk assessment. You tally output by hand. The robot is running, and the workload on the floor is roughly what it was before.
Japan’s own public guidance makes the same point: mid-sized and smaller manufacturers find it hard to carry a robot project on their own and generally need support from an outside body (Kanto Bureau of Economy, Trade and Industry, *Robot Implementation Policy Package*, May 2026 edition). The hard part is not the technology. It is scope definition.
So define the five boundaries yourself, first. Here they are.
| # | Boundary | What to decide | What happens if you leave it undefined |
|---|---|---|---|
| 1 | Infeed | Who builds the alignment, positioning and discharge of the workpiece | “It can’t pick the part” surfaces at commissioning, and blame gets passed around |
| 2 | Gripper (EOAT) | How many part types on how many grippers, and who pays for future additions | Every new part type triggers a redesign fee, and only the original vendor can quote |
| 3 | Safety | Who performs the risk assessment and who signs it | No document exists that you can produce in a customer audit or after an incident |
| 4 | Upstream integration | Where production data lands, and at what granularity | The robot runs but nobody can count output |
| 5 | After go-live | Source code, passwords, teaching rights, spares, response time | De facto lock-in — one firm controls both modifications and maintenance |
Boundary 1: Infeed — A Robot Can Only Pick What Has Been Placed
This is the first boundary and the one that causes the most friction.
What Robots Are Good At Is “Same Place, Same Orientation”
An industrial robot is fast and accurate when the workpiece is at a defined coordinate in a defined orientation. The corollary is that it is poor at workpieces whose position and orientation change every cycle. Bin-picked components. A pallet that has slumped. Cartons whose dimensions drift slightly from lot to lot. Who builds the mechanism that absorbs that variation is Boundary 1.
The classic misunderstanding on site runs like this. The buyer assumes: “We’ll stack it on the pallet exactly like we do today, and the robot picks from there.” The supplier has written in the quote: “Feed position and orientation shall be as specified in the specification document.” Neither party sees a contradiction in their own reading. Then, on commissioning day, someone discovers that below the fourth layer the cartons sink by 7 mm — and the gap in understanding finally becomes visible.
Four Things to Settle on the Infeed Side
- Positional tolerance — how many millimetres of variation in placement will be accepted, and who fixes it when that is exceeded
- Orientation variation — will unaligned orientations be tolerated, or will an alignment device present parts already oriented
- Part-to-part variation — carton dimensional tolerance, flash on moulded parts, colour variation between lots. Absorbed by vision, or absorbed mechanically
- The discharge side — where and how finished goods are stacked, what happens when the container is full, and who swaps it
If you decide to condition the infeed mechanically, what you are actually buying is bespoke machine design and fabrication: alignment conveyors, positioning stops, flipping units, infeed chutes. These are not catalogue items — they are designed around your specific workpiece, so design hours and fabrication cost become a substantial block within the quote. Handing Boundary 1 to the supplier means ordering that entire machine set.
The discharge side is the part most often forgotten. The “picking” end gets discussed enthusiastically; the “placing” end gets deferred. In palletizing applications the discharge side *is* the machine, so it is worth reading Palletizing Robot Costs and Selection alongside this article to get a feel for whether the money lands on the feed side or the discharge side.
The Weight of Choosing Vision
3D vision or bin-picking is often proposed as the way to absorb infeed variation. It works technically — but it moves Boundary 1 from “condition it mechanically” to “absorb it in software,” and both the cost structure and the risk structure change with it.
Choosing vision increases initial teaching effort and introduces sensitivity to lighting conditions, workpiece surface condition and background changes. It can work perfectly at handover and then, six months later, the material supplier changes and the surface gloss shifts, and recognition rates fall. Who pays for that re-tuning is something to settle at contract signature, not afterwards. It also connects directly to Boundary 5.
Conditioning mechanically raises initial equipment cost but reduces the number of moving variables. Neither answer is universally right. Choose based on which kind of variation your own organisation can manage over five years. Do not buy what you cannot manage — that is the practical rule.
Boundary 2: Gripper (EOAT) — Cost Scales With the Number of Part Types
The gripping device on the end of the robot is called EOAT (End of Arm Tooling). This is the second boundary.
Decide “How Many Part Types Per Robot” First
Gripper cost is driven almost entirely by the number of part types and the nature of what is being gripped. Handling five sizes of the same carton shape usually needs one vacuum gripper. Handling metal parts, plastic parts and soft bagged goods on the same robot means design and fabrication cost stacking up per gripper.
When comparing quotes, check three things:
- How many part types is the gripper designed for (representative part only, or all of them)?
- If gripper changeover occurs, is it automatic or manual, and how many seconds does it take?
- Who pays for the design change when a part type is added?
The third question matters most, and it is the one least likely to appear in the quote.
The Part-Type Addition That Always Arrives in Year Three
Factories are living things. Three years on, the mix has changed. Projects that never defined Boundary 2 end up here:
You want to run a new part. There are no gripper drawings. You ask the integrator who designed the gripper, and they tell you it needs a new design rather than a modification of the existing one. You try to get competing quotes — but you have neither the gripper drawings nor the robot teaching programs, so nobody else can price the work. You end up ordering from the original integrator with no negotiating position whatsoever.
That is what lock-in actually looks like. There is no clause in the contract that says “lock-in.” The simple fact that you do not hold the drawings and the source code makes the alternatives disappear.
How to Write Gripper Scope Into the RFP
The fix is straightforward — put the following into the RFP (request for proposal) stage:
- Attach a complete list of current part types (dimensions, weight, material, surface condition)
- State, as far as you can, what part types you expect to add within three years
- Make handover of gripper 3D drawings, bills of material and purchased-parts lists a condition
- Require unit rates for design changes on part-type additions (day rate plus estimated hours) to be quoted at contract time
The fourth is awkward to raise in negotiation, but any serious integrator can produce a rate. A firm that cannot, or that will only say “we’ll quote it when it happens,” may well be planning to make its margin at Boundary 2.
Boundary 3: Safety — Who Signs the Risk Assessment
The third boundary is safety. It is the heaviest issue both technically and legally, and it is the one most likely to be written vaguely in a quote.
ISO 10218 and ISO/TS 15066
Industrial robot safety is built around ISO 10218 (safety requirements for industrial robots and robot systems), with ISO/TS 15066 supplementing it for collaborative operation — where humans and robots share a workspace — by setting out how contact forces and pressures should be treated. The critical point is that these standards do not work on a “buy this model and you’re safe” basis.
Safety is assessed against the *robot system* — the robot plus peripheral equipment plus the workpiece plus the movement paths of the people around it — not against the robot alone. Which means the risk assessment is performed by whoever integrated the system, or by whoever uses it. It is not something the robot manufacturer can do on your behalf.
“Put Up a Fence” and “Perform a Risk Assessment” Are Different Line Items
This is where the boundary splits. Installing hardware — safety fencing, light curtains, area scanners, emergency stops, interlocks — and executing the process — hazard identification, risk estimation, decisions on risk reduction measures, documentation of residual risk — are two different pieces of work.
When a quote says only “safety measures, one lot,” it almost certainly covers the first and not the second. Check the following:
- Who performs the risk assessment, and what are that person’s qualifications and track record?
- Are the assessment outputs (hazard list, risk reduction records, residual risk register) included in the deliverables?
- What language are those documents in — Japanese, English or Thai — and which of those can your operators actually read?
- Who delivers operator safety training, in what language, and for how many hours?
- If the layout is changed after go-live, who performs the re-assessment?
For plants in Thailand there is an additional angle: customer audits. If you are a tier 1 or tier 2 automotive supplier, your customer’s process audit may ask to see the safety assessment records for the robot cell. At that point, “the fence is installed” is not an answer. You need a signed document.
Choosing a Cobot Moves Boundary 3
If you choose a collaborative robot (cobot) so that people can work alongside it without fencing, the position of Boundary 3 shifts substantially. The fencing cost disappears, but in its place come measurement and evaluation of contact forces and pressures, speed and force limit configuration, and the design of operator movement paths. It is closer to the truth to say that hardware cost has been replaced by process cost than to say safety measures are no longer required.
The cost structure and implementation sequence for cobots are covered in Collaborative Robot Implementation: Costs and Approach. It is worth reading at the point where you start making the Boundary 3 discussion concrete.
Boundary 4: Upstream Integration — Where Production Data Lands
The fourth boundary is the seam between the robot cell and your production management systems. It is the one most likely to be labelled “we’ll think about it later,” and thinking about it later is expensive.
The Robot Runs, But You Cannot Count Output
The classic failure: the robot cell runs exactly to specification. But you cannot pull from any system how many units were produced, how many times it stopped, or which part type ran between which hours. The floor is still filling in handwritten daily reports. Management asks why, after automating, the production report looks exactly as it did before.
The cause is an undefined Boundary 4. The integrator’s scope ended at “output a completion signal on a dry contact.” Who receives that signal and where it gets recorded was nobody’s job.
Decide Granularity and Destination
What to settle first about upstream integration is not features — it is granularity.
| Level | What you get | What it requires |
|---|---|---|
| Level 1 | Completed-unit count only | Contact output plus a counter, or increment in an existing PLC |
| Level 2 | Output by part type and by time slot | Part-change signal exchange, clock synchronisation |
| Level 3 | Downtime broken out by stoppage cause | Alarm code definitions with external output, shared code table |
| Level 4 | Two-way integration with production management / MES | Receiving production orders, returning results, consistent ID numbering rules |
Levels 1 and 2 can usually be handled entirely inside the existing PLC and control panel, so the incremental cost is limited. From Level 3 upward, the alarm code definitions have to be agreed between the robot manufacturer, the integrator and you, which generates design hours. Level 4 also means changes on the production management system side.
The problem is that adding Level 3 or above after the fact means modifying the control panel and rewiring. Something that would have cost a few thousand THB — running one spare signal line at installation — becomes a job in the hundreds of thousands of THB two years later once panel modification, downtime scheduling and re-validation are included. Ordering Control Panel Design and Manufacturing in Thailand sets out what to specify at order time regarding what goes inside the panel.
Do Not Turn “Not Now” Into “Not Ever”
In practice, plenty of projects do not need Level 4 built in from day one, and budgets are finite. The right response is not to lower the level but to leave room at the outset to expand upward.
Concretely: spare I/O points, space inside the control panel, a defined alarm code scheme, a reserved communication port. At the initial stage all of these cost very little. One line in the RFP — “assume future expansion to Level 3; reserve 20% spare I/O” — changes what the installation work costs two years from now.
Boundary 5: After Go-Live — Where Japanese-Owned Plants in Thailand Have the Most Regret
The fifth boundary is the one this article most wants to emphasise. This is where Japanese-owned companies operating in Thailand most often end up regretting the decisions they made — and it is also the least likely to appear in the quote.
What Gets Handed Over, and What Does Not
Once commissioning is finished and acceptance is signed, what do you actually hold? A great many projects never define this in advance.
| Item | With handover | What happens without handover |
|---|---|---|
| Robot teaching programs | Received in editable form | Every motion change needs a vendor visit |
| PLC program | Source with comments | Maintenance staff cannot read the logic; modification impossible |
| Controller password | Received with administrator rights | No configuration change is possible at all |
| Electrical drawings and panel wiring diagrams | CAD data plus PDF | No other firm can quote a modification |
| Gripper and fixture 3D drawings / BOM | CAD data plus purchased-parts list | You cannot source consumable parts yourself |
| Operating manuals and maintenance procedures | In a language the floor can read | Knowledge stays with individuals and is lost at handover |
“Our policy does not allow us to release source code” is a real answer you will sometimes get. As a business decision it is not necessarily dishonest. The problem is signing the contract without knowing that condition and finding out three years later. If you are going to contract on a no-handover basis, get the unit rate for modification work and the response times written into the contract in exchange. Framed as an equivalent trade, the later negotiation still works.
Who Holds Teaching Authority
What bites harder in day-to-day operation is authority to change the teaching. Minor adjustments — the feed position has drifted a few millimetres, a vacuum pad was replaced and the height changed — come up monthly on a running cell.
Whether your own maintenance staff can touch those adjustments completely changes your line downtime. If they cannot, you call the integrator every time. Near Bangkok they may come same-day or next-day. In Rayong, Chonburi, Ayutthaya — or Khon Kaen or Lamphun — travel time simply becomes downtime.
Write into the RFP that the scope includes granting your maintenance staff the authority to change teaching, plus the training to do it. And specify the language of that training. Thai maintenance technicians trained from a Japanese-language manual will not retain it. You need Thai-language procedures and hands-on training in Thai on the actual machine.
Local Spares and Response Time
The other item is spare parts. Three things to check:
- Which parts are held locally — of vacuum pads, cables, sensors and servo motors, which ones actually fail?
- Procurement lead time — with no local stock, how many days from Japan, China or Europe? Does that figure include customs clearance?
- How response time is defined — is it a contractual SLA in hours, or a best-efforts target?
This is the point at which the choice between a Japanese-affiliated integrator and a local Thai integrator turns into concrete numbers. A local integrator inside the same industrial estate gets there fast. A Japanese-affiliated integrator brings Japanese-language communication and home-office backup. Which matters more depends on your location and your maintenance organisation, so decide on the actual road distance from your plant to each firm’s base, not on generalities.
Criteria Specific to Selecting a Robot Integrator in Thailand
The five boundaries above apply equally in Japan, Thailand or Vietnam. Four factors are specific to Thailand.
There Simply Are Not Enough SI Engineers
Thailand’s system integration sector is growing, but on the talent side it is still early. The Thai Automation and Robotics Association (TARA) has more than 120 automation SI member companies, and the government has a plan to grow the pool of specialist SI personnel from roughly 200 today to 1,400. The association has certified 280 automation system analysts (aSA).
Look at the order of magnitude. Roughly 200 specialist SI engineers is plainly small relative to the scale of Thai manufacturing. What it means in practice is that good engineers are being fought over, and when projects overlap, schedules and support slip.
The picture in Japan is similar. The Japanese robot system integrator association has around 220 member firms, and 98% of them cite engineer shortages as a problem. Going with a Japanese-affiliated integrator does not mean people are plentiful.
There are two practical implications. First, plan your order timing to avoid peak periods. Second, make the deliverables package thicker so that you can do more yourself. That takes us back to Boundaries 2 and 5. In a market short of people, reducing vendor dependency is the single biggest risk control available to you.
BOI Incentives — But Check the Scope and Conditions
Thailand’s Board of Investment (BOI) offers incentives for automation investment. Notification No. 4/2569, published 31 March 2026, sets out — for automotive manufacturing (activity category 3.6) and HEV/PHEV manufacturing (3.8) — a three-year corporate income tax exemption for automation and robotics investment, capped at 50% of the investment amount, plus import duty exemption on machinery. Furthermore, if 30% or more of the value of the machinery installed is linked to Thailand’s domestic automation machinery industry, the cap rises to 100% of the investment amount. The minimum investment is 1,000,000 THB (excluding land and working capital), and the application deadline is the end of 2027.
Three cautions. First, this measure targets automotive activity categories — it does not apply automatically across all industries. Check which activity category your business falls under before anything else. Second, because there is a 30% local-content requirement, which integrator you use and whose machinery they build with feeds straight into the incentive. Importing a complete turnkey line from overseas, versus having a local integrator fabricate it, can produce different incentive amounts. Third, other general productivity-improvement measures exist (machinery replacement on existing BOI projects, for example), but conditions vary case by case — always confirm your own situation directly with the BOI or a specialist.
Either way, confirm your BOI eligibility before you request quotes, not after. The incentive requirements affect where you draw the boundaries — specifically, how much gets built locally.
How to Look at Payback Under a 400 THB Minimum Wage
Thailand’s minimum wage was revised to 400 THB per day across all of Bangkok on 1 July 2025, and has been held at that level for 2026. Wage growth at Japanese-owned companies has been reported at 3.8% in 2023, 4.58% in 2024, and a projected 4.64% for 2025.
How should you read that? Labour costs remain considerably lower than in Japan. Which means a payback calculation whose only numerator is labour cost reduction is harder to make work in Thailand than in Japan. The lower the annual cost per operator, the smaller the saving.
So automation in Thailand needs a numerator beyond headcount reduction. Specifically:
- Reduced quality variation (defect rate, customer complaints, sorting labour)
- Reduced staffing risk (recruitment difficulty, turnover, peak-season temporary labour)
- The possibility of 24-hour operation (operating hours you cannot reach by adding people)
- Removing people from hazardous work (injury risk, insurance premiums, audit exposure)
- Standing in customer audits (in some sectors, automation level is a condition of award)
Most of these are hard to convert into money. Which is exactly why producing a payback period using only the convertible items yields a number worse than reality. Conversely, be wary of proposals that monetise unsupported benefits to shorten the payback figure. The estimate below takes the same position.
Put Thai-Language Teaching and Maintenance Training in the Specification
Finally, an unglamorous point that pays. The language of the robot operator panel, the language of alarm messages, the language of maintenance procedures, and the language of training. If these are not in the specification, you will be delivered Japanese or English.
The people who touch the robot daily in a Thai plant are, in most cases, Thai operators and maintenance technicians. If an alarm appears in a language they cannot read, the only available response is “call the Japanese manager.” On a night shift, that time is added directly to downtime.
In Thai plants, a large share of on-the-job training runs through Japanese or English materials, and the language barrier becomes a direct learning burden — something we see regularly ourselves. A robot installation sharpens that problem. Put a Thai-language alarm code table, Thai-language daily inspection procedures and Thai-language teaching-change procedures on the deliverables list. Spending a few tens of thousands of THB there pays for itself in avoided downtime.
Our broader thinking on automation investment in Thailand is set out in How to Approach Factory Automation in Thailand.
Our Own Estimate: How Five-Year TCO Changes Across Three Contracting Models
The section below shows how the placement of the five boundaries turns into money. Everything here is TOMAS TECH’s own estimate, not actual results from any named company. The figures are standard levels we have seen across multiple projects, offered as a frame you can rebuild with your own numbers.
Assumptions
| Item | Value used | Note |
|---|---|---|
| Location | Industrial estate near Bangkok | Assembly and packing process at a Japanese-owned manufacturer |
| Configuration | Two 6-axis robots | One for part picking, one for boxing |
| Operation | Two shifts, 16 hours/day, 250 days/year | 4,000 hours per year |
| Part types | Four current types | Two additional types assumed in year three |
| Operator labour cost | 216,000 THB per person per year | Based on 400 THB/day including overtime, statutory benefits and overhead |
| In-house engineer rate | 120,000 THB per person-month | Production engineering / maintenance. Assumed internal transfer rate including overhead |
| Opportunity cost of line stoppage | 3,500 THB per hour | Assumed gross margin contribution per hour for the line concerned |
| Evaluation period | 5 years | Initial cost plus five years of running cost |
All figures are in THB. Because exchange rates move, avoid comparing in yen — decide in THB.
Three Scenarios
- Scenario A: single package to a Japanese-affiliated integrator — one firm takes all five boundaries. Source code and drawings handed over. Negotiation in Japanese.
- Scenario B: single package to a local Thai integrator — lowest initial price. Upstream integration stops at contact output. No agreement on source code handover.
- Scenario C: split award — robots, grippers and safety to a local integrator; upstream integration in-house or to a separate vendor. Specification definition and management done by you.
Initial Cost Comparison
| Line item (THB) | A Japanese-affiliated, single package | B Local Thai, single package | C Split award |
|---|---|---|---|
| Robots and controllers (2 units) | 1,800,000 | 1,700,000 | 1,700,000 |
| Gripper (EOAT) design and fabrication | 900,000 | 600,000 | 750,000 |
| Feeding, alignment and discharge equipment | 1,600,000 | 1,100,000 | 1,300,000 |
| Safety equipment and risk assessment | 750,000 | 500,000 | 650,000 |
| Control panel design/build and electrical work | 850,000 | 600,000 | 700,000 |
| Upstream integration | 700,000 | 150,000 | 550,000 |
| Engineering (design, teaching, witnessed testing, documentation) | 1,900,000 | 900,000 | 1,000,000 |
| Buyer-side management effort | — | — | 480,000 |
| Installation, commissioning and training | 550,000 | 350,000 | 400,000 |
| Initial total | 9,050,000 | 5,900,000 | 7,530,000 |
Looking only at this point in time, B is dramatically cheaper. The gap between A and B is 3,150,000 THB, roughly 35%. B gets chosen in a great many projects because this is the only table anyone looks at.
The 480,000 THB of buyer-side management effort in Scenario C assumes half of one in-house engineer for eight months — four person-months at 120,000 THB each. In a split award you carry the boundary coordination yourself, so this line always exists. A split-award estimate that puts it at zero does not match reality.
Note that buyer-side effort is not zero in A and B either (specification review, witnessed testing, acceptance). But the gap against C lies in where coordination accountability sits, and to keep the comparison simple we have not entered this line for A and B. Read the A and B figures as slightly conservative — that is, slightly unfavourable to C.
Five-Year Running Costs
| Line item (THB, 5-year cumulative) | A Japanese-affiliated, single package | B Local Thai, single package | C Split award |
|---|---|---|---|
| Scheduled maintenance and consumables | 900,000 | 750,000 | 800,000 |
| Spare parts inventory | 250,000 | 200,000 | 220,000 |
| Year-three part-type addition (2 types), modification cost | 450,000 | 1,450,000 | 600,000 |
| Retrofitted upstream integration work | — | 900,000 | — |
| Opportunity cost of downtime | 210,000 | 1,050,000 | 420,000 |
| Training and handover | 150,000 | 250,000 | 180,000 |
| Buyer-side ongoing management effort | — | — | 400,000 |
| Running total (5 years) | 1,960,000 | 4,600,000 | 2,620,000 |
Scenario C’s 400,000 THB of ongoing management effort assumes roughly 3.3 person-months over five years (at 120,000 THB per person-month) spent maintaining the specification, coordinating between vendors, and obtaining competing quotes for modifications. A split award needs your hands not only at the start but throughout. Choose C without seeing this, and management collapses the moment the responsible person transfers.
The Five-Year TCO Result
| Category | A Japanese-affiliated, single package | B Local Thai, single package | C Split award |
|---|---|---|---|
| Initial cost | 9,050,000 | 5,900,000 | 7,530,000 |
| 5-year running cost | 1,960,000 | 4,600,000 | 2,620,000 |
| 5-year TCO | 11,010,000 | 10,500,000 | 10,150,000 |
| Gap vs A at initial cost | — | −3,150,000 | −1,520,000 |
| Gap vs A at 5 years | — | −510,000 | −860,000 |
The 3,150,000 THB gap between A and B at the outset narrows to 510,000 THB after five years. About 84% of the initial price difference disappears over five years.
How to Read This Table
Three line items inflate Scenario B.
First, the 1,450,000 THB modification cost for the year-three part-type addition. Because the gripper drawings and teaching programs are not in your hands, you cannot obtain competing quotes and end up ordering at the original vendor’s price. A and C stay within 450,000–600,000 THB because holding the drawings and the teaching rights lets competition work. Boundaries 2 and 5 are what is operating here.
Second, the 900,000 THB of retrofitted upstream integration work. Because the initial scope stopped at contact output, the moment you discover in year two that you need results by part type and by stoppage cause, you are into control panel modification and rewiring. Reserving spare I/O at installation would have made this line far smaller. That is Boundary 4.
Third, the 1,050,000 THB opportunity cost of downtime. Modelled as 60 hours of downtime per year × 5 years × 3,500 THB/hour. It assumes you cannot change teaching yourself and must call the vendor each time. A is modelled at 12 hours per year, C at 24. That is Boundary 5.
In other words, what drives up B’s five-year TCO is not workmanship — it is signing a contract without defining Boundaries 2, 4 and 5. Award to the same local integrator with those boundaries defined and the numbers can land close to C. Read A/B/C as differences in where the boundaries were drawn, not as a ranking of companies.
What This Estimate Deliberately Leaves Out
We do not give a payback period. There is no defensible basis for the numerator. Assume the removal of three operators and you get 648,000 THB per year — but in practice, people remain for monitoring, changeover and exception handling, so how many positions actually disappear depends on the process. On top of that, as noted above, Thailand puts more weight on the benefits that are hard to convert: quality, staffing risk, operating hours. When the denominator (TCO) can be set but the numerator cannot, the honest thing is not to publish a payback figure.
For the same reason, we have not costed workplace injuries or customer audit findings — there is no basis for assigning a probability. When you see a proposal with a crisp payback period, always ask where its numerator came from.
The RFP Question List — Mapped to the Five Boundaries
Here is all of the above in usable form. Send these questions, in identical wording, to every firm you are asking to quote, and the quotes become comparable. RIPS (Robot System Integration Process Standard, proposed in 2017 by Japan’s Ministry of Economy, Trade and Industry together with the Japan Robot Association) was created as a process framework for preventing misalignment between buyer and integrator. It works best when you use it not as “a standard the integrator must follow” but as “a skeleton the buyer uses to write scope.”
| Boundary | Question to put in the RFP | How to read a vague answer |
|---|---|---|
| 1 Infeed | What is the tolerance on feed position and orientation, in mm and degrees? Whose scope covers exceeding it? | If it says “pre-aligned assumed,” alignment equipment is yours to source |
| 1 Infeed | Whose design scope covers the discharge destination, full-container handling and swap frequency? | No mention of the discharge side means it is effectively undefined |
| 2 Gripper | How many part types are covered? How many gripper variants will be built? | “Designed for representative part” means the rest are out of scope |
| 2 Gripper | Are gripper 3D drawings, BOM and purchased-parts lists included in deliverables? | If not, modifications go to the original vendor by default |
| 2 Gripper | Can you quote a day rate and estimated hours for design changes on part-type additions? | If not, your year-three costs are unknowable |
| 3 Safety | Who performs the risk assessment, and what are their qualifications and record? | “Safety measures, one lot” usually means hardware installation only |
| 3 Safety | Are the hazard list, risk reduction records and residual risk register delivered? In what language? | Without documents you cannot answer a customer audit |
| 3 Safety | Safety training for operators and maintenance staff — what language, how many hours? | With no language specified, you get Japanese or English |
| 4 Upstream | Which of Levels 1–4 is the production data granularity? (Attach the table from this article) | “Completion signal output” is Level 1 |
| 4 Upstream | Who defines the alarm code scheme, and can it be output externally? | Vague ownership means stoppage-cause analysis is impossible later |
| 4 Upstream | What percentage of spare I/O and panel space will be reserved? | If none, future expansion means modifying the panel |
| 5 After go-live | Will robot teaching programs, PLC source and electrical drawings be handed over? | If not, get modification unit rates written into the contract |
| 5 After go-live | Will the controller administrator password be handed over? | If not, every configuration change needs a vendor visit |
| 5 After go-live | Does scope include granting our maintenance staff teaching-change authority, plus training? In what language? | If not, even minor adjustments extend downtime |
| 5 After go-live | Local spare parts held, procurement lead times, and the definition of response time | If it is a “best-efforts target,” it is not an SLA |
Put these fifteen questions out and the responses separate clearly. Firms that answer immediately, firms that come back within a few days, and firms that come back asking what the question means. None of those is inherently bad — but any item that never gets answered should be treated as not included in that firm’s quote.
If assembling this question list on your own is difficult, having someone outside help structure the concept phase is a legitimate option. That approach is covered in Making Use of Factory Automation Consulting.
Sequencing for Smaller Plants
Finally, a note on how plants without a dedicated production engineering department should handle these five boundaries.
For the First Robot, Pick the Process With the Fewest Boundaries
The classic first-project failure is going straight at the most labour-intensive process. Labour-intensive processes are usually complex, which means Boundary 1 (infeed variation) and Boundary 2 (number of part types) are at their most demanding.
For the first robot, pick the process with the fewest boundaries: few part types, stable workpiece geometry, feeding and discharge already in place, and simple upstream requirements. Palletizing, boxing and simple machine tending often qualify. The measured benefit may be modest, but what you keep is the experience of having operated a robot — and whether you have that experience completely changes the precision of your boundary definitions on robot number two.
Build One Person In-House Who Can Touch Teaching
As covered in Boundary 5, whether you can change teaching yourself drives your running cost. On the first installation, name one maintenance technician and have them attached to the integrator’s commissioning work from the very beginning. Not sent on a course — present during the work. Three weeks on site during commissioning sticks better than any training curriculum.
Ideally, that person should be Thai. Concentrate the knowledge in a Japanese expatriate manager and it leaves with them at the end of their posting. Build it up in a Thai maintenance technician who can pass it on to a successor in Thai.
Three Quotes — But the Same RFP
Three firms is the practical upper limit for competitive quoting. Beyond that, the effort of comparison collapses. What matters is not the number of firms but giving all three the same RFP. Send the fifteen questions above in identical wording and you can tell whether a price difference is a scope difference or a rate difference.
And do not mechanically pick the lowest bid. As the five-year TCO estimate shows, most of the initial price gap disappears within five years. Choose on where the boundaries sit and how thick the handover package is — not on the initial price.
Frequently Asked Questions
What does a robot system integrator actually do?
A robot system integrator is not a company that sells robots. It designs, builds, installs and commissions the complete production system that includes the robot. In practice that means process analysis, robot model selection, gripper design and fabrication, design and fabrication of feed and discharge equipment, safety measures, control panel design and build, PLC and robot programming, installation, commissioning and training. Robot manufacturers make arms and controllers; embedding one into a factory process is the integrator’s job. (In the Japanese market these firms are commonly called “SIer,” short for system integrator — you will see the term in Japanese-language material and from Japanese-affiliated suppliers.) The five boundaries in this article are the lines that divide how much of that embedding work the integrator takes on.
What does a robot system integrator cost?
There is no single market rate, because complexity varies enormously by process. Structurally, though, the split is consistent: the robot and controller are 20–30% of the total, with the remainder in grippers, feed and discharge equipment, safety, control panel, electrical work, engineering and installation. In this article’s estimate (near Bangkok, two robots, assembly/boxing process) initial cost ranged from 5,900,000 to 9,050,000 THB depending on the contracting model. These are assumed values in our own estimate, and they will move with your part-type count, operating conditions and existing equipment. Rather than asking for a market rate, insist on a quote broken out by line item. A quote that says only “robot system, one lot” cannot be compared with anything.
What is the difference between an integrator, a trading company and a robot manufacturer?
They play different roles. The robot manufacturer builds the arm, the controller and genuine peripherals. Trading companies and distributors handle sales and delivery of those, and on some projects will introduce an integrator. The robot system integrator designs and builds a system around your process and hands it over in working condition. The largest cost movement sits in the integrator’s territory, and all five boundaries in this article are lines drawn between you and the integrator. Because one company sometimes plays several of these roles, judge by “who carries which line item in the quote” rather than by the label on the door.
Can a smaller manufacturer implement robots?
Yes — but the choice of process and the design of what stays inside your organisation decide the outcome. Japan’s own public guidance treats support from an outside body as a prerequisite for mid-sized and smaller manufacturers taking on robot projects (Kanto Bureau of Economy, Trade and Industry, Robot Implementation Policy Package). The practical sequence is: start the first robot on a low-boundary process (few part types, stable feeding and discharge), and attach one maintenance technician to commissioning so that you end up with someone who can touch the teaching. From the second robot onward, the accuracy of your boundary definitions depends on whether you have that experience. Going after the highest-value process can wait until the experience has accumulated.
What should we watch for when selecting a robot integrator in Thailand?
Four things. First, response time — judge by the actual road distance from your plant to each firm’s base. At an upcountry site, travel time is loss. Second, language — require alarm code tables, inspection procedures and teaching-change procedures in Thai, and training delivered in Thai. Third, alignment with BOI incentives — Notification No. 4/2569, published March 2026, sets out corporate tax exemption for automation investment in automotive and HEV/PHEV manufacturing, with the exemption cap raised where 30% or more of machinery value is linked to Thailand’s domestic automation machinery industry. Because where the machinery is built affects the incentive, confirm your eligibility before requesting quotes. Fourth, the talent squeeze — TARA data puts Thailand’s specialist SI workforce at roughly 200 today, with a government plan to grow it to 1,400. Assume schedules and support will slip when projects overlap, and thicken your handover package so you can do more in-house.
How should we evaluate return on a robot investment?
In Thailand, a numerator consisting only of labour cost reduction rarely works. With the minimum wage revised to 400 THB per day across Bangkok in July 2025 and held there for 2026, the saving per person is not as large as in Japan. The numerator needs to include reduced quality variation, reduced staffing risk, the possibility of 24-hour operation, removal of people from hazardous work, and standing in customer audits. But those are hard to convert into money. Which is why the realistic approach is to set the denominator (five-year TCO) precisely, and present the numerator as two separate lists — what can be converted and what cannot. Crushing it into a single payback figure always builds a lie into one side or the other.
What should we prepare before requesting competitive quotes?
The single most important thing is giving all three firms the same RFP. The RFP should describe the current state of the target process (part list, dimensions, weight, material, daily output, operating hours, existing equipment) and include the fifteen questions from this article. On top of that, specify the format of the quote: require a line-item breakdown across robots, grippers, feed and discharge equipment, safety, control panel, electrical work, engineering and installation, plus a deliverables list stating whether drawings, source code and procedures are included and in what language. Without a specified format, each firm writes at whatever granularity suits it and comparison becomes impossible. Building a level playing field beats adding more bidders.
Is it acceptable to demand handover of program source code?
Yes, it is. It is also true that some firms will decline, and that is a legitimate business decision on their part. What matters is getting that condition stated explicitly before contract. If you are going to contract on a no-handover basis, get the day rate for modification work and the response time (SLA) written into the contract instead, so the terms are fixed in numbers. The worst case is a contract that says nothing about handover and a first “we can’t release that” three years later. Treat gripper 3D drawings and BOMs the same way.
Does a collaborative robot remove the need for safety measures?
No. The hardware cost of fencing and light curtains falls, but process work rises in its place: measuring and evaluating contact forces and pressures, configuring speed and force limits, and designing operator movement paths. A risk assessment under ISO 10218 is still required for collaborative operation, and ISO/TS 15066 supplements it with the treatment of contact forces and pressures. Safety cost does not disappear; it moves from hardware to process. Which makes deciding, at Boundary 3, who performs the evaluation and who signs it *more* important with a cobot, not less.
Summary
Robot integrator quotes vary by a factor of two or three not because of differences in discount on the arm, but because of where each firm drew the line across five boundaries.
- Boundary 1, infeed — who absorbs workpiece position, orientation and part-to-part variation. Do not forget the discharge side
- Boundary 2, gripper — how many part types on how many grippers, and who pays for the year-three addition. Are the drawings handed over
- Boundary 3, safety — who performs the risk assessment, who signs it, and in what language the documents survive
- Boundary 4, upstream integration — at which of Levels 1–4 production data is captured, and whether spare I/O was reserved
- Boundary 5, after go-live — source code, passwords, teaching rights, spares, response time. Lock-in is decided here
In our own estimate, the 3,150,000 THB initial-cost gap between A and B narrowed to 510,000 THB after five years. About 84% of the initial price difference is consumed by part-type addition modifications, retrofitted upstream integration work, and the opportunity cost of downtime. Choosing on initial price means not seeing that structure.
The Thailand-specific factors are a specialist SI workforce of roughly 200 people, BOI automation incentives aimed at automotive activity categories with a 30% local-content condition, the difficulty of building a payback case under a 400 THB minimum wage, and the questions of local maintenance capability and language. Each of them reaches the total through where you draw the boundaries.
The first thing to write is not a request for quotation. It is your own answers to the five boundaries. Once you can write those, three quotes finally stand on the same ground.
If you are still at the stage of working out where the boundaries should sit in your own process, or what belongs in the RFP, that is a perfectly good place to start — no commitment required. TOMAS TECH is based in Bangkok and builds FA systems, control panels and production management systems for manufacturers operating in Thailand, and we are happy to talk from the concept stage onward. If you would like to compare notes before deciding whether to proceed at all, get in touch through our contact page.
References
- IFR World Robotics 2025: Global robot demand in factories doubles over 10 years
- Thailand unveils new incentives for automotive and HEV/PHEV manufacturing (Notification No. 4/2569, Tilleke & Gibbins)
- Thailand BOI: Automation incentives
- Thai Automation and Robotics Association (TARA)
- The state of Thailand’s system integrator industry (M Report)
- Japan FA and Robot System Integrator Association
- Survey on engineer shortages among robot system integrators (Newswitch)
- Explainer on RIPS, the Robot System Integration Process Standard (KOTO ONLINE)
- Kanto Bureau of Economy, Trade and Industry, Robot Implementation Policy Package (May 2026 edition)
- Thailand minimum wage and wage trends (Kuno CPA Thailand blog)
- Thailand BOI manufacturing investment trends (Pertama Partners)
