What Is Factory Automation — The Full Picture of FA and How to Roll It Out 2026

If you have just started looking into the question “what does factory automation actually cover”, here is the conclusion first. FA is not about buying equipment. It is about handing the tasks and judgements people have carried out over to machines, one at a time, in a deliberate order. And almost every site that gets real results starts not with full automation, but with a small start on a single process. This article walks through the building blocks of FA, why it has become necessary in Thailand and Southeast Asia, and how to phase in automation step by step.

What Factory Automation Really Is — A Question of Sequence, Not Equipment

Factory Automation, or FA, translates literally as automating the factory. But if that is the only definition your team carries into the discussion, the conversation will almost certainly slide into “which robot should we buy” — a model-selection exercise. When you are actually designing FA in practice, a far more useful framing is this.

FA is the work of deciding in what order to hand five things people have been doing — moving, measuring, transporting, judging and recording — over to machines.

This definition is practical because it shifts the axis of the investment decision from equipment performance to sequence. With the same budget, the wrong sequence produces no benefit. Automate the moving (machining and assembly) without handing over the measuring (inspection and gauging), and you get defects mass-produced at high speed. Automate machining while leaving the transporting (inter-process movement) with people, and work in progress piles up in front of the machine so utilisation never rises. The success or failure of FA is decided far more by the design of this sequence than by the merits of any individual machine.

How FA Is Defined in Manufacturing — Think in Three Layers

When people on the shop floor talk about FA, three layers with quite different characteristics are usually mixed together. Discussing them as one blurred whole is why conversations fail to connect.

• Machine layer (moving and transporting) — robots, dedicated machines, conveyors, AGVs and AMRs, the parts that physically move the object
• Control layer (measuring and judging) — PLCs, sensors, vision inspection systems and safety devices, the parts that detect state and decide whether an action may proceed
• Information layer (recording and connecting) — HMIs, SCADA and production management systems, the parts that record results and pass them to people and to upper-level systems

When someone says “let’s get a quotation for automation,” what is normally in scope is the machine layer and the control layer. Yet most of the projects that end up judged as “we automated and nothing improved” are missing the information layer. The equipment runs, but how many units were made, why it stopped, and which process is the bottleneck are never captured as numbers. As a result, there is nothing to base the next investment decision on. TOMAS TECH handles both factory IT and FA precisely because we have seen so many automation projects with that third layer missing.

A Short History of FA — From Lights-Out to Sharing the Work with People

FA is not a new term. Its starting point was the 1970s, when PLCs (programmable logic controllers) spread and control that had been wired as relay circuits could be rewritten as programs instead. Through the 1980s and 1990s, the mainstream was automating dedicated lines built on the assumption of mass production. The goal of FA in that era was unambiguous — a lights-out factory.

From the 2000s, however, high-mix low-volume production became the norm and the premise behind dedicated lines fell apart. Equipment that needs a major changeover every time the product variant changes never reaches the utilisation required to pay back. That is where the thinking about automation shifted. Instead of handing every process to machines, you keep with people what people are good at (judgement, setup, handling exceptions) and hand over only what machines are good at (repetition, precision, heavy loads). Designing that division of labour became the central theme of FA.

Collaborative robots that can sit next to a person without safety fencing became practical in the 2010s, and autonomous AMRs and cloud-connected SCADA spread in the 2020s, making this “shared” form of automation more realistic still. A site starting its review today should take its reference not from lights-out line success stories of the 1990s, but from this division-of-labour design philosophy.

The Building Blocks of FA — Seven Parts That Only Become Automation Together

There are seven main elements that make up FA — controllers (PLC/PAC), sensors, industrial and collaborative robots, AGVs and AMRs, HMIs (touch panels), SCADA, and the communication network. Below we cover the role of each, limited to the points you need to settle during your review. The details of each individual technology are left to their own dedicated articles.

Controllers (PLC) — The Core of FA That Sets the Order

A PLC is an industrial computer that receives signals from sensors and, following a program written in advance, decides which actuator to move and when. It is the central nervous system of FA.

There are three points to settle at the review stage. First, whether to standardise on the PLC brand already used in your existing equipment. Mixing Japanese and non-Japanese brands means protocol conversion, which raises both cost and the difficulty of maintenance. Second, who owns the program source code. If the machine builder keeps it as a black box, every later modification becomes a purchase order. Third, whether anyone locally can touch the program. At sites in Thailand, this third point is usually the most binding real-world constraint.

We explain the basic structure of PLCs and how to think about ladder programming in detail in What Is a PLC — The Basics of Sequence Control and How to Choose One, so start there if you want to understand control from the inside out.

Sensors — Automation Quality Is Decided by What You Can Measure

Sensors are the components that convert physical quantities such as position, presence, temperature, pressure, dimension and colour into electrical signals. They are an unglamorous element, but the quality of your automation is decided here. A person can notice an abnormality by sight and touch, but a machine cannot notice what it is not measuring.

A common trap in automation reviews is leaving out of the requirements the checks that operators perform unconsciously in the current process. The orientation of the workpiece, the presence of burrs, reading the stamp applied in the previous process. When people are “sort of looking at” these things, they never make it into the specification, the process gets automated without them, and the gap surfaces after go-live as defects escaping to the customer. When you choose a target process, observe what the operator is actually looking at — not only what the work instruction says — before you translate it into requirements.

Industrial and Collaborative Robots — Replace the Person, or Stand Beside Them

An industrial robot performs fast, high-precision repetitive motion inside a fenced-off area. Payload and reach determine the model, and typical applications are welding, painting, palletising and machine tending. For processes with stable volume that repeat the same motion for long stretches, it is still the most cost-effective choice available. We have summarised how the types are classified and how to select between them in Types of Industrial Robots and How to Select One — What to Decide Before You Buy.

A collaborative robot, by contrast, is designed to meet safety standards such that, when the conditions are satisfied, it can be installed next to a person without safety fencing. It moves more slowly than an industrial robot, but it takes up a small footprint and adapts easily when the layout changes. At high-mix low-volume sites, or at sites that want to try a single unit first, this is often the entry point. For a picture of the processes it actually suits, see Collaborative Robot Use Cases — Which Process to Start With.

Which one to choose comes down to how stable your volume is and how often your layout changes. At a site where product changeovers happen several times a month, installing an industrial robot that depends on dedicated fixtures means the changeover effort eats up the benefit of automating in the first place.

AGVs and AMRs — Automating the Transporting Between Processes

An AGV (automated guided vehicle) is a transport vehicle that follows a fixed route along magnetic tape or a guide wire on the floor. An AMR (autonomous mobile robot) estimates its own position using onboard sensors and decides its own route, avoiding obstacles as it goes.

Automating transport tends to be pushed back behind automating machining, but it is in fact an area where the benefit is easy to see. Moving material between processes adds no value, and it is also where the largest share of manual labour goes. We have set out the prerequisites and the common pitfalls in What Is an AGV — Automated Guided Vehicle Basics and What to Decide Beforehand. For how they are actually used and in which processes, AGV Implementation Case Studies — Ways of Using Them That Delivered Results is a useful reference.

One caution — when you review transport automation, opening with the vehicle itself is not a good move. In-plant logistics improves through three layers: reducing storage points, shortening the distance travelled, and automating the act of transporting. Introducing vehicles belongs to the last of those layers. We write about this sequence in detail in In-Plant Logistics Improvement Happens in Three Layers, so if transport is where you feel the pain, read that first.

HMI (Touch Panel) — Where It Is Decided Whether the Floor Can Use It

The HMI is the operating screen through which operators check equipment status, set conditions and respond to abnormalities. Because it has no direct bearing on machine performance it tends to be treated lightly, yet it is one of the elements with the biggest impact on utilisation.

The reason is simple — automated equipment always stops. How much real running time you get depends on whether, when it stops, the operator on the floor can read the cause from the screen and recover on their own. A design that displays only an error code and sends someone hunting for the manual, versus a design that shows graphically which sensor responded and how, differ by an order of magnitude in time to recovery.

At sites in Thailand and Southeast Asia, a language problem is layered on top of this. Go live with Japanese-only screens and local staff cannot operate the machine, so the line stops every time an expatriate has to be called. We have summarised the practical essentials of screen design in Touch Panel Screen Design — Building an HMI That Keeps the Line Running.

SCADA and Upper-Level Systems — Making Automation Visible

SCADA collects operating data from multiple machines and PLCs and monitors, records and visualises it. At the stage where you have automated a single machine it looks unnecessary, but once the count grows to three or five machines, human observation can no longer keep track of where the bottleneck is.

The important thing here is not to file SCADA away as “something to add later”. Trying to extract operating data after the equipment is installed means modifying the existing PLC program and pulling out signals, which generates extra cost. Include at least the list of signals to be output in the specification at the time of your first machine, and later expansion becomes dramatically easier. Bringing this decision forward is exactly the kind of question that sits on the boundary between FA and factory IT.

Communication Network — Equipment That Does Not Connect Is Not Automated

Industrial Ethernet and fieldbuses are the foundation of communication that links machine to machine and machine to upper-level system. It is an element that rarely surfaces in discussion, but at sites with equipment from multiple vendors it is the single biggest source of unplanned cost.

During your review, we recommend compiling a list of the communication standards your existing equipment supports. With that list in hand when you select new equipment, you can choose models that avoid adding converters.

Organising the seven elements by role and by the order in which to consider them gives the following.

What deserves your attention in this table is that the high-priority elements are not necessarily the expensive ones. Sensors and the HMI account for a small share of the total quotation, yet they decide whether things work after go-live. Conversely the robot itself, the big-ticket item, is a relatively low-risk area for model selection as long as you do not pick the wrong process.

Why FA Is Needed in Thailand and Southeast Asia Right Now

Many sites started moving because head office in Japan said “look into automation”. But the reasons FA is needed in Thailand and Southeast Asia are somewhat different from the drivers inside Japan. Let us check them against the numbers.

More Than Half Are Already Moving — 27.9% Underway, 27.5% Planned

According to the FY2023 Survey on Business Conditions of Japanese Companies Overseas, published by JETRO on 16 May 2024, among Japanese companies operating in Thailand, 27.9% were already working on automation and 27.5% answered that they planned to do so. Together that is 55.4% — in other words, more than half are either executing or planning automation.

This number needs to be read carefully. There is no need to panic that “more than half are doing it, so we are behind”. What matters more is that the underway and planned shares sit at almost the same level. That tells you two things: a site beginning its review now is by no means a latecomer, and because so many sites are running the same review at the same time, the resources of machine builders and system integrators are easily stretched thin. You need to allow generous lead time from quotation to installation.

40.4% Face Labour Shortages, 72.8% Cite Rising Labour Costs as a Risk

In the same survey, 40.4% of Japanese companies in Thailand answered that they face a shortage of personnel, and 72.8% named rising labour costs as a management risk.

The particular difficulty of operating in Thailand lies in these two numbers being high at the same time. If you simply could not hire, one available move would be to raise wages and become more competitive as an employer. If labour costs were simply rising, one available move would be to raise productivity and absorb them. But in an environment where both are happening at once, you raise wages and still cannot fill the roles, while your current headcount cannot support an increase in output — a deadlock. FA is needed because there are not many other ways to break it.

Wage levels are being pushed up by policy as well. According to JETRO Business Briefs, a minimum wage of 400 baht per day has applied to all industries within Bangkok since 1 July 2025. The minimum wage level itself directly affects only a limited number of processes, but because the whole wage table gets pushed upward, the indirect effect reaches the entire plant.

Digital Adoption at 52.1% — Where That Puts You

There is one more figure worth holding on to. According to the FY2025 Survey on Business Conditions of Japanese Companies Overseas (Asia and Oceania edition), published by JETRO on 26 November 2025, 52.1% of Japanese companies in ASEAN were using digital technology. The same survey shows that while Australia, South Korea and India all exceed 60%, ASEAN sits below that level.

In the context of FA, what this figure signals is the weakness of the information layer described earlier. Even where equipment automation has progressed, many sites have not reached the point where operating data is recorded and used. Flip that around, and a site reviewing FA today can start one step ahead of its peers simply by designing the machine layer and the information layer as a set.

Investment Incentives as a Tailwind

Thailand has an investment incentive scheme run by the BOI (Thailand Board of Investment), under which investment in automation and robotic systems can qualify for benefits such as corporate income tax exemption. Exemption rates and eligibility conditions vary with the content of the investment and with local sourcing, so we recommend checking the conditions early in your review.

We have organised the specific exemption rates, the practicalities of applying, and how to look at costs in Factory Automation in Thailand — Implementation Sequence, Costs and BOI Incentives, so turn to that when you reach the stage of building an investment plan for a Thai site. Whether you build your plan on the assumption of incentives, or make it stand up without them, changes which processes you should target.

What Is Happening in Neighbouring Countries

Similar pressures are reported across other ASEAN countries as well. In Vietnam, average manufacturing wages are said to be trending upward, and the market for industrial robots is reported to be expanding. These come mainly from secondary sources and conditions vary widely by site, so they should not be treated as settled fact. Still, at the very least the shared direction is clear enough — building an equipment plan on the assumption that a production model premised on cheap labour will continue for a long time is a risky bet.

Demand for industrial robots is itself concentrated in Asia. World Robotics 2025, published by the International Federation of Robotics (IFR) on 25 September 2025, reported annual worldwide installations of industrial robots at 542,000 units in 2024, with Asia accounting for 74% of new installations. In the sense that both the supply chain for equipment and parts and the pool of engineering resources are deep within Asia, the conditions for pursuing FA in this region are coming into place.

How to Roll Out Automation — Build Factory FA From a Small Start

This is the heart of the matter. The way to pursue FA is built in five steps, from selecting the target process through to horizontal rollout.

Why You Should Not Jump Straight to Full Automation

First, three reasons to avoid automating an entire line in one go.

• Scaling up before the requirements are settled makes rework costs grow in proportion to that scale. A specification gap found on one machine costs one machine’s worth of modification; found after ten machines have been ordered, it costs ten.
• If the unit of measurement is too large, you cannot tell what actually worked. Change a whole line at once and, whether productivity rises or falls, you cannot pin down the cause. You end up with no basis for the next investment decision.
• The shop floor cannot climb the learning curve fast enough. Automated equipment stops most often right after installation. When several machines stop at once there are not enough hands to recover them, and the floor ends up deciding to revert to manual work. Once you revert, restarting becomes markedly harder.

A small start is not about making the investment small for its own sake. It is a way of working that finds mistakes in the requirements early and cheaply, and buys the floor time to get used to the equipment.

Step 1 — Analyse the Current State and Define the Target Process and Objective

The first thing to do is not to compare machine models but to understand where you stand. Concretely, map out the work time, headcount, defect rate and overtime hours of each process, plus the answer to “why is a person doing this task?”

At this point we strongly recommend narrowing to a single objective. Headcount reduction, quality stability, capacity for higher output, or eliminating hazardous work. When there are several objectives, the equipment specification bloats and the result is that none of them is achieved properly. Once the objective is set, the target process narrows down naturally.

The criteria for choosing the target process are as follows.

• A process whose work is repetitive, with little variation in conditions
• A process that absorbs a lot of labour, so the saving can be shown in numbers
• A process where the incoming condition from the previous step is stable
• A process positioned such that a stoppage does not halt downstream steps for long
• A process whose key people on the floor are cooperative

That last item cannot be quantified, but in practice it is the most important. If the first automated machine is rejected by the floor, FA at that site stalls for years.

Step 2 — Estimate the Return on Investment

Next, estimate how much benefit automating the target process will generate. What matters here is not to overestimate the benefit and not to underestimate the cost.

On the benefit side, the only things you may legitimately count are those where headcount allocation actually changes or overtime hours actually fall. Qualitative benefits such as “the work becomes easier” or “quality becomes stable” are weak as a basis for an investment decision, so convert them as far as possible into numbers such as defect rate or rework hours.

On the cost side, load in the following items from the start, not just the price of the equipment itself.

• Fabrication of fixtures and grippers
• Signal interfacing with, and modification of, existing equipment
• Electrical, air and installation work
• Production losses during the commissioning period
• Spare parts stock and annual maintenance fees
• Man-hours for floor training and preparing manuals

When you get to the stage of collecting quotations from several suppliers, unless you align whether these items are included in each quotation, you cannot compare the figures side by side. We set out how to align the assumptions concretely in Comparing Automation Quotations — The Assumptions That Put Them on the Same Footing.

Step 3 — Run a Small Pilot on One Line or One Process

Once the estimate looks promising, narrow the scope to one process, or one line, and install a pilot. The purpose here is not to raise productivity but to surface the errors in your requirements.

Here is what to verify during the pilot.

• Whether the cycle time you assumed is actually achieved
• How long changeover takes when you switch product variants
• How much variation from the previous process it can absorb
• The frequency of abnormal stoppages, and the share the floor can recover unaided
• Whether local staff can operate it and handle basic maintenance

Of these, the one that betrays you most often in practice is the third. Time and again we see cases where everything ran fine in a lab environment or during the builder’s witness test, but variation in the real incoming material causes frequent stoppages. Run your pilot with real production material and at the rhythm of real production, without exception.

There is also the option of not building fully automatic equipment straight away, and starting instead with a semi-automatic configuration where a person loads and the machine processes. It keeps the investment down while letting you verify what the machine side can really do, which makes it a rational approach for a first unit. For details, see What Is a Semi-Automatic Machine — When Not Going Fully Automatic Is the Right Call.

Step 4 — Verify in Live Operation and Run PDCA

Fold the pilot equipment into live production, run it for a set period (at least around three months), and reconcile your original estimate against actual results.

What you must record at this stage is a breakdown of stoppages by reason. Stoppages caused by equipment failure, by changeover, by waiting for material and by quality checks each call for entirely different countermeasures. Without records by reason, all you are left with is the impression that “it did not deliver as much as we hoped”, and you cannot make the next decision. The SCADA and operation monitoring mechanisms described earlier start paying off at this stage.

And if actuals fall short of the estimate, identify the cause before moving on. Deciding at this point that “adding more units should surely produce the benefit” only lines up more machines carrying the same problem.

Step 5 — After Confirming the Benefit, Roll Out Horizontally to Adjacent Processes

Once the pilot has demonstrated benefit and the floor is used to running it, expand to adjacent processes. As a rule, the order of horizontal rollout follows the process flow, spreading upstream and downstream. Jump to a distant process and the know-how gained in the pilot (how fixtures are approached, the maintenance setup, how training is run) cannot be reused.

What starts to pay off only at the rollout stage is the standard you created on the first machine. PLC program structure, HMI screen layout, the error code scheme, the maintenance procedure. Get these in order on machine one and the commissioning period for the second and subsequent machines shrinks substantially. Build the first machine ad hoc, on the other hand, and you end up with as many different operating methods as you have machines.

Organising the five steps by rough duration and deliverable gives the following.

These durations are indicative only and will shift with the difficulty of the target process and the engineering resources available locally. Between Step 2 and Step 3 in particular sits the process of selecting and placing an order with a machine builder, so in busy periods it takes longer than assumed. We have summarised what kind of builder to choose, and how, in How to Choose an Automation Equipment Builder — Look at Their Setup Rather Than Their Track Record.

How to Think About Payback — A View Particular to Phased Automation

Payback on automation tends to be discussed as a simple calculation — divide the capital cost by the labour cost saved to get the number of years. That calculation is useful as a starting point, but when you proceed in phases, both the numerator and the denominator need correcting.

For the denominator (the investment), the first rule is, as described above, not to leave costs beyond the equipment price unaccounted for. On top of that, if horizontal rollout is the premise, separate out the portion of the first machine that is the cost of creating the standard. The screen design and maintenance procedures created on machine one can be reused from machine two onward, so evaluating payback on the first machine alone produces an unduly harsh result.

For the numerator (the benefit), do not look only at labour cost savings. Benefits that are straightforward to count in practice include the following.

• Redeployment of direct headcount, or reduction in overtime hours
• Reduction in defect and rework man-hours
• Reduction in quality assurance man-hours through automated inspection records
• Increased capacity through unattended running at night and at weekends
• Lower insurance premiums and downtime losses from reduced occupational accident risk

Of these, the one most often overlooked at Thai sites is the fourth. Approach the review purely as reducing daytime headcount and the benefit looks small, but if the same equipment lets you extend running hours into the night, the numerator changes dramatically. Note, however, that night running requires a response structure for abnormalities and a quality assurance design as a set, so verify what the equipment can really do unattended during the pilot before you write it into the plan.

The target payback period depends on your site’s investment policy, but setting the target too tight has a side effect — only the processes where benefit comes easily get considered, and automation ends up as a one-off. Agree internally up front on a two-stage view, where the first machine is an investment in verifying requirements and payback comes from the second machine onward, and your budget is far less likely to be blocked at the rollout stage.

Common Stumbles in FA Projects — There Are Only So Many Failure Patterns

Failures look infinitely varied, but in practice they fall into a handful of patterns. Here are the representative ones.

The Objective Has Become “To Automate”

When a review is kicked off by head office policy or a subsidy scheme, the objective can quietly become automation itself. Proceed in that state and, with no criterion for the investment decision, the model gets chosen on quotation price alone. Then when you are asked after go-live what the benefit was, you cannot explain it, because the pre-installation numbers to compare against were never captured. Narrowing to a single objective in Step 1 is the step that avoids this pattern.

Automating While Leaving Upstream Variation Unaddressed

Automated equipment is designed on the premise that incoming workpieces are in a consistent condition. When quality upstream varies, you need extra mechanisms to absorb that variation, which increases both cost and stoppage frequency. Stabilise the entry conditions of the target process before you automate it. Simply keeping to that order often brings the investment amount down.

The Assumptions Behind Competing Quotations Are Not Aligned

Collecting quotations from several suppliers is the right approach in itself, but if you issue the enquiry with vague requirement specifications, each supplier quotes on different assumptions and comparing the figures becomes meaningless. You find out later that the cheap quotation was simply written on cheap assumptions.

The Maintenance Structure Has Not Been Designed

After the equipment is installed, who does the daily checks, who responds to breakdowns, and where consumables are sourced from? Go live without deciding this and the first breakdown means a long stoppage. At sites in Thailand, whether the machine builder has a local presence translates directly into response time. It is an item to check at the model selection stage.

Local Staff Cannot Operate It

This overlaps with the HMI point above, but if you defer language support for operating screens, manuals and training, utilisation will not rise. Build a clearly defined training period for local staff into the implementation plan.

We have organised these and other failure patterns, together with how to avoid them, in more detail in The Risk of Failure in Automation Investment — What Stops the Payback. Reading it before you write the investment approval request makes it much easier to prepare for the questions you will get.

Also, if you are taking on the first machine with no one in-house who has FA experience, one option is to bring in outside knowledge from the requirements definition stage. Leave requirements definition to the machine builder and the proposal tends to lean towards the configuration that builder is good at. For how to use a third-party position, see Using Automation Consulting — How Much to Put Outside.

Frequently Asked Questions

How is FA defined in manufacturing?

Generally it refers to factory automation as a whole, but in practice it is easier to design if you see it as the work of taking what people do — moving, measuring, transporting, judging and recording — and shifting it to the machine side across three layers, machine, control and information. Installing a single machine is sometimes called FA, but unless you design all the way through to operating results being recorded and visualised, nothing remains to base the next investment decision on.

Where exactly should a small start begin?

It means narrowing the scope to one process, and ideally one machine. The purpose is not to hold down the investment amount but to find errors in the requirements cheaply and early, and to secure time for the floor to get used to running it. The process you choose is best if the work is repetitive, the incoming condition from the previous step is stable, and it sits where a stoppage will not halt downstream steps for long. Starting with a semi-automatic configuration where a person loads and the machine processes, rather than going fully automatic, is also an effective option.

Over how many years should we expect automation investment to pay back?

That depends on your site’s investment policy, but evaluating payback on the first machine alone makes the judgement unduly harsh. The first machine carries the cost of creating standards such as screen design and maintenance procedures, and those can be reused from the second machine onward. We recommend agreeing internally on a two-stage view, where the first machine is an investment in verifying requirements and payback comes from the second machine onward. When counting benefits, remember to include not only labour cost savings but also reductions in defect and rework man-hours and the increase in capacity from night running.

Is phased automation possible at a small or mid-sized site?

Yes. If anything, smaller sites have the advantage of faster decision-making and an easier time narrowing the target process. At sites with a limited number of processes, starting with automation of transport or inspection rather than machining sometimes makes the benefit easier to see. We have summarised approaches by scale in Robot Implementation at Small and Mid-Sized Sites — An Approach Scaled to Your Size.

At what stage should we consult someone about FA?

Consultation is most effective at the stage before the target process is decided — in other words, when you have just started considering automation. If you wait until the model and specification are fixed, the options have already narrowed, and revisiting the requirements themselves becomes difficult. Even if your current process data is incomplete, we can start from the question of which data you should be capturing.

Can Thailand’s BOI incentives be used for automation investment?

The BOI has incentives targeting investment in automation and robotic systems, and where conditions are met, benefits such as corporate income tax exemption may be available. Exemption rates and eligibility requirements vary with the content of the investment and with local sourcing, so we recommend checking the conditions early on when you are building the investment plan. Details are explained in Factory Automation in Thailand — Implementation Sequence, Costs and BOI Incentives.

Summary

Factory automation is not about buying equipment. It is about designing the order in which the work and judgements people have carried out are handed over to machines. FA is made up of controllers, sensors, robots, transport equipment, HMIs, SCADA and the communication network, but if you look only at the expensive machine layer and leave the control and information layers for later, you end up with automation that runs while its benefit cannot be measured.

FA is needed in Thailand and Southeast Asia because labour shortages and rising labour costs are advancing simultaneously, and responding by adding headcount as before no longer works. JETRO’s survey shows that 27.9% of Japanese companies in Thailand are already working on automation and 27.5% plan to do so. Meanwhile digital technology adoption among ASEAN companies remains at 52.1%, leaving considerable room to pursue machine automation and data utilisation as a set.

The approach comes in five steps — current-state analysis and target process selection, the ROI estimate, a pilot on a single process, live verification with PDCA, and horizontal rollout to adjacent processes. The aim of a small start is not to hold down the investment amount but to find errors in the requirements cheaply and early, and to buy the floor time to adjust. Create the standard on machine one, and recover the investment from machine two onward. Agreeing on that two-stage view internally at the outset is the single best preparation for not stalling at the rollout stage.

TOMAS TECH is based in Bangkok and supports Japanese manufacturers across Thailand and Southeast Asia with factory IT such as production management systems, traceability through OT and IoT, and the implementation of FA. We welcome enquiries from the earliest stage of a review, including when the target process is undecided and you have no sense yet of the investment scale. Even if your current process data is incomplete, we can start from the question of where to begin measuring, so please Contact us if anything here raises a question.

References

• FY2023 Survey on Business Conditions of Japanese Companies Overseas JETRO, published 16 May 2024
• FY2025 Survey on Business Conditions of Japanese Companies Overseas, Asia and Oceania edition JETRO, published 26 November 2025
• Bangkok minimum wage raised to 400 baht per day JETRO Business Briefs, July 2025
• World Robotics 2025 press release International Federation of Robotics, published 25 September 2025
• Thailand Board of Investment official site Thailand Board of Investment