Anonymous analytics help us improve. No advertising. Privacy.
Semiconductor Companies in Singapore: Buyer's Guide (2026)
What engaging a foundry, OSAT, or design partner gives you: access to manufacturing that costs billions to build and which you will never own. And what it quietly takes back: a design welded to one company's process, and a place in a queue you do not control.
Authorship and review
Ranked list
Ranked list — directory records ordered by the published profile-signal methodology; paid modules are separate.
June CEditorial label; identity and credentials unpublished
Last updated
This byline is not proof of the operator, a legal identity, or reviewer credentials. Editorial standards
A semiconductor partner in Singapore occupies one link of a long chain: foundries that fabricate wafers, design houses that create the circuits, assembly and test providers who package and verify the finished parts, design-automation vendors, and the equipment makers who supply the machines that do all of it, delivered to organisations in Singapore.
Singapore is unusually significant in this chain. It is widely reported to produce around a tenth of the world's semiconductors and roughly a fifth of global semiconductor equipment output, and the Economic Development Board has reported over S$30 billion of semiconductor investment between 2022 and 2025. Local strength sits in mature and specialty processes, in packaging and test, and in equipment, rather than in the leading-edge logic nodes that dominate the headlines. For most buyers, that is precisely the part of the chain they actually need.
The trade here is the starkest in this directory. A modern fabrication plant costs billions of dollars and is obsolete on a schedule nobody controls, so you rent access to one. In exchange, your product becomes dependent on a specific company's process, in a way that is considerably harder to reverse than any software contract, because the dependency is not in the paperwork. It is etched into the design itself.
The list below groups semiconductor companies with a recorded Singapore-presence signal. It is unranked: ordered by profile signal score, then company name, with inclusion reflecting recorded profile signals rather than endorsement. The buyer's guide beneath it names no foundries and no manufacturers, because the argument it makes applies across the chain. What engaging a semiconductor partner is genuinely worth, what it costs you later, and what to verify before you commit a design.
Notable semiconductor providers
Unranked — ordered by profile signal score, then company name. Inclusion reflects a recorded Singapore-presence signal, not endorsement.
Listing order reflects recorded profile signals and is not affected by payment. Sponsored placements, if any, are labelled separately and never reorder this list.
Arcadia Engineering Services supplies refurbished semiconductor manufacturing equipment, spare parts and engineering support, concentrating on wet chemical etch and high purity chemical applications.
Deston Plastics manufactures precision plastic engineering parts and the dies, moulds, tools, jigs and fixtures used to produce them, operating from Mega@Woodlands.
I.C.E. Aluminium provides precision aluminium parts manufacturing, running a one-stop service from extrusion through CNC machining to surface treatment.
Samsung Electronics is a South Korean multinational and one of the world's largest technology companies. It is a dominant producer of memory and storage semiconductors, including DRAM and NAND flash, and operates a major contract chip-foundry business.
Camtek is a global manufacturer of metrology and inspection equipment for the semiconductor industry, with operations in Singapore serving the Asia-Pacific market.
Capcon Singapore designs and builds die bonders for advanced integrated-circuit packaging, with its main research and development base in Singapore and additional teams in Suzhou, Beijing and Taiwan.
Exyte Singapore is the local operation of the German engineering group that designs and builds controlled-environment facilities, including semiconductor fabrication plants, cleanrooms and data centres.
Feinmetall Singapore is the regional arm of German test-equipment maker Feinmetall, designing, manufacturing and selling vertical probe cards and related accessories.
Innovative Global Solutions & Services (IGSS) is a Singapore-based company specialising in the manufacture and repair of semiconductor foundry equipment, alongside broader industrial trading and services.
JLK Technology is a Singapore-headquartered contract development and manufacturing organisation (CDMO) with over 30 years of expertise, serving the medtech, consumer-electronics and automotive sectors.
Marvell Technology is a semiconductor company providing essential data infrastructure solutions. They design custom silicon, storage controllers, Ethernet networking chips, and optical interconnects.
Radiant Opto-Electronics is a Taiwan-based semiconductor company specializing in backlight modules and LED lighting solutions for LCD displays. The company develops high-performance optical films, light guide plates, and backlight assemblies.
FA Systems Automation builds factory automation equipment and precision engineering solutions from its Changi South facility, serving semiconductor and electronics manufacturers across the region.
Hybrionic designs and manufactures thick film microcircuits and hybrid circuit assemblies from its Ang Mo Kio facility, with additional operations in Malaysia.
Lewe Engineering is a Singapore engineering, procurement and construction management contractor specialising in mechanical, electrical and plumbing design for industrial facilities.
Advanced Micro-Fabrication Equipment, known as AMEC, develops plasma etch and deposition equipment and is the leading domestic semiconductor equipment maker in China.
Advanced Semiconductor Equipment Technology sells, assembles and services semiconductor production equipment, operating under licence agreements with original equipment manufacturers rather than designing its own tools.
Ardentec provides independent semiconductor test services, specialising in wafer probing and final test for chipmakers that outsource back-end verification.
Bitsilica provides semiconductor design and verification services, covering register transfer level design, intellectual property and system-on-chip verification, physical design and static timing analysis, design for testability, and field programmable gate array work.
Cadence Design Systems develops electronic design automation software used to design and verify integrated circuits, systems-on-chip and printed circuit boards.
Caliber Interconnects provides engineering design services for semiconductor test and packaging, covering ATE loadboards, vertical probe card substrates, IC package and substrate design, and high-density interconnect layout and simulation.
Elmos Semiconductor designs mixed-signal integrated circuits for vehicles, covering the sensing, motor control and interface functions that sit behind everyday car features.
Grand Venture Technology manufactures precision components and complex mechanical assemblies for capital-equipment makers, covering precision machining, sheet-metal fabrication, ceramic and quartz machining, vacuum-compatible parts, and cleanroom assembly and testing.
Gyromatic Innovation develops mobile robots and material handling systems for semiconductor fabs, covering both the vehicles and the software that dispatches them.
Hirata FA Engineering designs and builds factory automation systems, working as a system integrator that takes projects from concept through design, manufacture, installation and maintenance.
Kromax South Asia distributes semiconductor production equipment and materials and provides the engineering services around them, working as a channel for overseas equipment makers entering Southeast Asia.
MEDs Technologies is a Singapore-headquartered engineering and distribution company serving radio-frequency, microwave, and electronics markets across Asia-Pacific.
Mi Material is a semiconductor-materials business specialising in solder balls, solder powder, and related products used in advanced electronic packaging.
Micro-Mechanics designs, manufactures and markets high-precision consumable tools and parts used in process-critical steps of semiconductor wafer fabrication, assembly and test.
Mieux designs and manufactures test hardware for semiconductor final test, centred on test sockets and the consumable parts around them. Its catalogue covers test sockets, spring probes, test fingers, change kits, fixtures and stiffeners, alongside test board design and assembly.
National Instruments builds modular test and measurement hardware and software, centred on the PXI platform where instrument functions come as interchangeable cards in a common chassis with shared timing and synchronisation.
Odyssey Technical Solutions repairs radio-frequency, direct current and microwave power delivery equipment for semiconductor and industrial users, a niche it claims as the largest dedicated specialist in.
Samsung Display specializes in the manufacturing of advanced display panels, with a focus on OLED technology. The company produces flexible, foldable, and transparent displays for various applications, including smartphones, tablets, wearable devices, and automotive displays.
STATS ChipPAC is an outsourced semiconductor assembly and test provider headquartered in Singapore, offering turnkey back-end manufacturing from package design and characterisation through wafer probing, wafer bumping, assembly, final test and drop shipment.
STMicroelectronics is a semiconductor manufacturer that designs and produces chips for various markets, including automotive, industrial, personal electronics, and communications.
Sunyu Photonics builds automated test systems for silicon photonics, targeting the slow wafer-level testing that constrains photonic integrated circuit production.
TAG Industrial supplies in-plant material handling equipment to Singapore manufacturers and facilities, ranging from power-assisted lifting and movement aids through to fully autonomous vehicles and cleaning robots.
Western Pneumatic supplies vacuum and pneumatic equipment to manufacturers in Singapore and provides the installation, maintenance and repair services that keep it running.
XG Tech develops computing platforms for vehicles, working across the smart cockpit, autonomous driving functions and the full-stack integration that ties them together.
How to choose a semiconductor partner in Singapore in 2026: the advantages, the pain points, and the checks
What you are actually buying
You are buying access to a process, and a place in a queue. Both matter more than price, and buyers who negotiate hardest on the per-wafer figure frequently lose far more on the other two.
The process is the deeper commitment. A chip is not designed in the abstract and then manufactured wherever is cheapest. It is designed against a specific foundry's process design kit: the models, rules, and libraries that describe what that particular fab can physically make. That kit is proprietary, and the design that comes out of it is shaped around it. Moving the same product to another foundry is not a procurement exercise; it generally means redesign, a new mask set, and a full requalification, which is why second-sourcing is a strategic decision made at the beginning of a product's life or, usually, never made at all.
The queue is the commercial reality. Foundry capacity is allocated rather than simply sold, and allocation follows volume, relationship, and strategic importance. A large customer with a long history is not standing in the same queue you are. Understanding where you sit in that hierarchy, before you depend on it, is the single most useful thing a smaller buyer can do.
The advantages that justify using a foundry or OSAT
You do not build a fab, and you should be extremely glad about that. The capital, the process engineering, the yield learning, and the obsolescence are somebody else's. This is the least controversial buy-versus-build decision in all of technology, and it is not close.
Process engineering that took decades to accumulate. Yield, reliability, and process control at scale are the product of enormous sustained investment and a great deal of institutional knowledge that cannot be bought or hurried. You are renting the accumulated learning, which is worth more than the equipment.
Mature and specialty processes are where most real products live. The leading edge gets the attention; the majority of shipping products need something proven, well-characterised, and long-lived, and that is exactly where Singapore's strength lies. Buying a mature process is frequently the smart decision rather than the compromise.
Packaging and test as a discipline in its own right. Advanced packaging increasingly determines what a product can do, and the assembly and test ecosystem here is deep. This part of the chain has become a source of genuine differentiation rather than a commodity finishing step.
Qualification and quality systems you could not construct. Automotive, aerospace, medical, and industrial customers demand quality regimes that take years to establish. A partner who already holds them is selling you access to markets as much as manufacturing.
Ecosystem and proximity. Design services, equipment vendors, materials, test houses, and the specialist engineering labour to connect them are concentrated here. That density genuinely shortens cycles, and it is not easily replicated.
National support is real. Singapore actively supports the sector, and grants and co-funding exist for capability building and equipment. Terms change, so verify current programmes with the relevant agency rather than the vendor, but this is substantial and frequently underused.
The pain points buyers consistently underestimate
The design kit is the lock-in, and it is not a contract term. Your design is built against one foundry's process design kit. It is not portable. Moving means redesign, a new mask set, and requalification, at a cost and a delay that frequently exceeds the value of whatever pricing advantage prompted the move. Nobody is trapped by the supply agreement. They are trapped by the physics.
Mask costs are sunk on the first tape-out. The mask set is a large, non-recurring expense, tied to a specific process, and it does not travel. A design error discovered after tape-out is not a patch; it is a new mask set and months of schedule.
Capacity is allocated, not simply bought, and you are not first in the queue. In a tight market, allocation goes to the largest and most strategic customers. A smaller buyer discovering this during a shortage discovers it at exactly the moment it cannot be fixed. Get capacity commitments in writing, and understand what they are actually worth when the market turns.
Process nodes are eventually discontinued, and your product dies with them. A fab will at some point end-of-life a process. If your product depends on it, you face a redesign onto a new node on the foundry's timetable, not yours. Ask about the expected longevity of the process before you commit a design to it. This question is rarely asked and is one of the most consequential in the entire engagement.
Second-sourcing sounds prudent and is extremely expensive. Because designs are process-specific, maintaining a genuine second source means maintaining a second design, a second mask set, and a second qualification. Many organisations claim to have a second source and, on inspection, have a plan to create one, which is a different thing entirely.
Lead times are long, and they are physical. Wafer fabrication, packaging, test, and qualification take the time they take. No escalation shortens a process step, and buyers who compress their schedule against these realities simply discover them later, more expensively.
Minimum volumes make small products uneconomic. Foundries and packaging houses are optimised for volume, and a low-volume product may face minimum order quantities that dominate its economics or exclude it entirely.
Intellectual-property blocks come with their own licensing chains. Reused design blocks carry licence terms, royalties, and restrictions of their own, and their compatibility is often process-specific too. The dependency stack is deeper than most buyers map.
Export controls and geopolitics now sit inside a technical decision. Where a chip is made, what it can do, and who it may be sold to have become questions with a legal dimension that changes with policy rather than with engineering. This is a supply-chain risk that no amount of good design mitigates, and it must be assessed rather than assumed away.
What changed in 2026
Capacity moved from commodity to strategic asset, and buyers are treated accordingly. The years of shortage taught the industry that allocation is a lever, and that lesson has not been unlearned. Foundries now think in terms of strategic customers and long-term commitments, which is excellent if you are one and uncomfortable if you are not. For a smaller buyer, the practical consequence is that the relationship, the forecast you provide, and the commitment you are willing to make are now genuinely part of the negotiation rather than administrative detail.
Advanced packaging became a source of differentiation rather than a finishing step. As gains from shrinking transistors get harder, more of the performance is coming from how chips are packaged, stacked, and connected. That elevates the assembly and test part of the chain, where Singapore is strong, from a cost line into a design decision, and it deserves attention far earlier in a product's life than it traditionally received.
AI demand distorted the whole chain, including the parts that do not serve it. The enormous pull of accelerated computing has consumed capacity, equipment, materials, and skilled people, and the effects reach products with no connection to AI at all, through longer lead times, tighter allocation, and competition for the same engineers. Plan on the assumption that the chain is under strain, because it is, and it is not obviously about to stop.
Geopolitics became a design input. Export controls, national industrial policy, and supply-chain security requirements now shape where things can be made and who they can be sold to, and they change on political rather than engineering timescales. A partner who cannot discuss this coherently is not managing a risk that is genuinely present in your product.
The diligence that actually separates partners
Ask how long the process will exist before you commit a design to it. Expected longevity, end-of-life policy, notice period, and what a migration path would look like. This single question is worth more than a price negotiation, and it is the one buyers most reliably forget.
Establish where you sit in the allocation hierarchy. What a capacity commitment actually obliges the partner to, what happens in a shortage, and how customers like you were treated the last time the market tightened. Ask the question plainly, because the answer is knowable and the consequences are severe.
Price the second source honestly, at the start. What redesign, a new mask set, and requalification would genuinely cost, and how long it would take. Then decide, deliberately, whether you are buying a second source or merely intending to want one later.
Map the full dependency stack, not just the fab. Design kit, intellectual-property blocks and their licences, packaging, test, and the equipment behind each. The constraint that stops your product is frequently three links away from the one you negotiated.
Verify the quality regime against your end market. Automotive, aerospace, medical, and industrial each demand specific systems, and holding the certificate is not the same as having shipped under it. Ask for evidence of parts qualified and shipped into your sector.
Model the schedule against physical reality. Fabrication, packaging, test, and qualification take what they take. Build the plan around that, with genuine contingency, because escalation does not accelerate a furnace.
Assess the geopolitical exposure explicitly. Where the parts are made, which controls apply, and what would happen to your product if the policy changed. This belongs in the risk register alongside supply and yield.
Verify the Singapore presence and the actual capability. Match the registered name and UEN against ACRA, and confirm that the specific process, package, or test capability you need exists here rather than in a group facility elsewhere.
Red flags worth walking away from
No clear answer on the expected longevity of the process you are designing into.
Capacity described as available without any written commitment behind it.
A second source claimed but never costed, designed, or qualified.
Quality certifications held but no evidence of parts shipped in your end market.
Lead times that shorten conveniently to fit your schedule.
Silence, or evident discomfort, on export controls and supply-chain policy.
Intellectual-property blocks proposed with no clarity on licence terms or royalties.
When a semiconductor partner is the wrong answer
Use a foundry and an assembly partner for essentially all manufacturing. Owning a fab is not a decision available to you, and the mature and specialty processes available here are where the majority of real products belong. This part of the decision is not difficult.
The harder question is whether you should be designing a custom chip at all. A great many products reaching for one would be better served by an existing part, a programmable device, or a system-level solution, at a fraction of the cost and with none of the mask, qualification, and process-longevity commitments. Custom silicon is justified by volume, by power or performance constraints that nothing else meets, or by a genuine competitive advantage in the silicon itself, and it is a many-year commitment to a specific manufacturing process on somebody else's timetable. Think hardest of all before committing a long-lived product to a process nobody will confirm will still exist when the product is mature, because the redesign will arrive whether or not you have budgeted for it.
Frequently asked questions
Why can I not just move my chip design to another foundry?
Because a design is built against one foundry's proprietary process design kit and is shaped around what that fab can physically make. Moving generally means redesign, a new mask set, and full requalification. The lock-in is etched into the design, not written into the supply agreement.
What is the difference between a foundry, an OSAT, and an IDM?
A foundry fabricates wafers for other companies' designs. An OSAT packages and tests those wafers into finished parts. An integrated device manufacturer does everything itself, including its own designs. Most buyers today engage a foundry and an OSAT separately, and increasingly the packaging is where differentiation lives.
How long will my chosen process node stay available?
Ask, in writing, before you commit a design to it. Foundries eventually end-of-life processes, and if your product depends on one, you face redesign onto a new node on their timetable rather than yours. This question is rarely asked and is among the most consequential in the engagement.
Can I get guaranteed foundry capacity?
Capacity is allocated rather than simply sold, and allocation follows volume, relationship, and strategic importance. Smaller buyers discover their position in that hierarchy during a shortage, when it cannot be fixed. Get commitments in writing and ask how customers like you fared the last time the market tightened.
Is a second source worth maintaining?
It is genuinely expensive, because designs are process-specific: a real second source means a second design, a second mask set, and a second qualification. Many organisations believe they have one and actually have an intention to create one. Cost it honestly at the start, then decide deliberately.
Why is Singapore significant in semiconductors?
It is widely reported to produce around a tenth of the world's semiconductors and roughly a fifth of global semiconductor equipment output, with EDB reporting over S$30 billion of investment from 2022 to 2025. Local strength lies in mature and specialty processes, packaging and test, and equipment.
Should I design a custom chip at all?
Often not. Many products reaching for custom silicon are better served by an existing part, a programmable device, or a system-level solution, without the mask costs, qualification burden, or process-longevity risk. Custom silicon is justified by volume, by hard constraints nothing else meets, or by real advantage in the silicon.