Guide

The toolmakers: who builds the machines that build chips

Published 25 August 2026. Facts verified to 25 August 2026 unless dated otherwise.

A fab is mostly an equipment order with a building around it, which makes the tool industry the layer that decides whether a plant runs at all.

What an equipment maker actually sells

A semiconductor equipment manufacturer sells one step of a process flow that runs to several hundred steps. It sells a deposition chamber, or a scanner, or an implanter, or an inspection system, and the buyer treats that machine as a single node in a sequence where every other node has to agree with it. That is the first thing to understand about this industry: the product is not really the hardware. The product is a qualified, repeatable result on a specific film stack, on a specific wafer size, at a specific throughput, supported by people who will fly in when the result drifts.

The commercial consequence is that a tool sale is the beginning of a relationship rather than the end of a negotiation. A fab buying a new tool class runs demonstration wafers, characterises the process window, argues about particle counts, and only then releases the tool to production. That cycle costs both sides months of engineering time. Once it is done, the fab has a strong reason never to repeat it with a different vendor for the same layer, and the vendor has an installed base that generates spares, service contracts, upgrades and the next order. Equipment is sold once and monetised for a decade.

This also explains why the headline investment number attached to any fab is misleading unless you know its structure. In a modern wafer plant the tools account for far more of the total capital outlay than the structure built to house them, a split this desk sets out in full on the fab capex breakdown. When a project announces a very large figure, most of that money is an order book for a small number of foreign companies.

The tool classes, and how many credible suppliers each one has

The industry is often described as five companies. It is more accurate to describe it as a set of separate markets, each with its own competitive shape, that happen to sell to the same customers. The map below is standard industry structure rather than a ranking, and this record prints no market-share figures on this page because it has not verified any to its own standard.

Read the table sideways and the pattern appears. Where a step can be traded against time or chemistry, two or three vendors survive and compete on cost of ownership. Where a step sets a physical limit that nothing later in the flow can recover, supply narrows towards one name. Lithography is the extreme case, because a pattern printed too coarsely or slightly out of position cannot be repaired by anything that happens later in the flow, which is why our ASML company file treats a single toolmaker as a strategic actor rather than a vendor.

Why this industry consolidated, and why it stays consolidated

Three forces did the work. The first is qualification friction, described above: the customer, not the supplier, bears the cost and risk of trying something new, so incumbency compounds. The second is research and development scale. A tool class serves a global market of a few hundred serious buyers, and staying current in it means funding physics, software, precision mechanics and materials science simultaneously. Below a certain revenue base that spending is not survivable, and firms either specialise into a niche or exit.

The third is co-development. Tool roadmaps are written with lead customers years before the tools ship, because the fab needs to know what it will be able to build and the vendor needs to know what to design for. A newcomer is not merely late to a market. It is absent from the conversation in which the next generation of the market is defined. That is the structural reason a new national entrant cannot buy its way into front-end tool manufacturing on a five-year plan, and the honest reason this record treats localisation talk in this layer more sceptically than in any other.

The service annuity nobody photographs

A tool installed today will consume parts, chamber kits, software updates and field engineering for as long as it runs, and a fab's production tools run for many years past their depreciation schedule. That service and spares business is the steadier half of an equipment maker's income, it is less exposed to the capital cycle than new-tool sales, and it is also the part of the industry that genuinely can localise. Field service, spares warehousing, refurbishment, chamber cleaning and precision cleaning services all sit close to the customer by necessity.

For India this matters more than it sounds. The first practical question a fab operator asks a vendor is not about the machine, it is about response time when the machine stops. Every hour a bottleneck tool is down is lost output on an asset that depreciates whether or not it is running. A country building its first commercial fab is therefore also, quietly, building an argument for regional spares depots and resident engineering teams, and those are the first pieces of an equipment industry that can actually take root.

Export controls: the political layer sitting on a capital-goods market

Semiconductor equipment is one of the most heavily controlled categories in international trade, and the controls are not uniform. They bite hardest at the leading edge, where extreme-ultraviolet systems and the most advanced immersion scanners sit inside allied licensing regimes, and they thin out considerably as you move into mature-node tooling. That gradient has a direct bearing on India's position. A fab built to run 28 to 110nm processes is buying from the part of the market where licensing is routine rather than contested.

This is a structural convenience rather than a policy achievement, and it should be described as such. India's mature-node lane avoids the hardest procurement questions in the industry not because of any negotiation, but because it is not attempting the node range where those questions arise. Our page on why a country builds its own fab makes the wider strategic argument; the narrow equipment point is simply that the toolset for this fab is buyable.

What India buys today, and what it does not make

India has no domestic manufacturer of a front-end production tool class. Scanners, etchers, deposition systems, implanters, polishers, inspection systems and automated test equipment are imported, and will be imported for the foreseeable life of every plant currently approved. The India Semiconductor Mission's own status, published through the Press Information Bureau and current to July 2026, counts twelve approved manufacturing units carrying cumulative investment of over Rs 1.64 lakh crore, made up of one silicon fab, two compound-semiconductor fabs and nine packaging and test units DURABLE. None of those is an equipment plant. Any claim that India is about to make its own chipmaking machines should be checked against that list first.

What does exist is a components and sub-systems opportunity, and it is real rather than rhetorical. A production tool is an assembly of vacuum pumps, valves, mass flow controllers, gas panels, RF generators, chambers, robotics, precision-machined parts, seals and control electronics, much of it bought in by the tool maker rather than built in-house. Indian precision engineering already supplies adjacent industries to demanding specifications. The gap is qualification into a semiconductor equipment supply chain, which is its own multi-year process with its own auditors.

The Indian moves that are on the record

Two of the largest American toolmakers have made India commitments that are worth reading precisely, because both are engineering and training investments rather than factories that will produce tools.

• Applied Materials. The company announced a collaborative engineering centre in Bengaluru on 22 June 2023 through its own newsroom, with 400 million dollars of investment planned over four years, stated to support more than 2 billion dollars of planned investments in its first five years of operation and to create at least 500 advanced engineering jobs, with the centre's stated purpose being the development of equipment sub-systems and components alongside global and domestic suppliers REPORTED.
• Lam Research. TrendForce reported on 24 February 2025 that Lam Research would invest about 1.151 billion dollars in Karnataka, a figure carried in Indian coverage as about Rs 10,000 crore, tied to tool engineering, training and ecosystem development, with a training programme reported in the same round at 60,000 engineers delivered through the company's SEMulator3D process-simulation platform and funded through the India Semiconductor Mission and the electronics ministry REPORTED.

Both belong in the REPORTED tier until buildings, hires and qualified suppliers can be counted. Neither converts India into an equipment producer. What they do is put the two firms' engineering organisations inside the country, which is how a components supply chain gets audited into existence, and it is a more plausible route than any announcement about building scanners.

Dholera's equipment question, and what is not public

The plant under construction at Dholera is a 300mm fab with an approved investment of Rs 91,000 crore, the fiscal support agreement of 5 March 2025 citing Rs 91,526 crore, and a stated capacity of up to 50,000 wafers per month DURABLE. On the equipment side, one thing is on the record and a great deal is not. ASML and Tata Electronics announced a lithography memorandum of understanding on 16 May 2026 through both newsrooms, naming nodes of 28, 40, 55, 90 and 110nm REPORTED. That is a memorandum, not a purchase order.

No tool orders, delivery slots, configurations or financial terms for the Dholera fab are in the public domain, from ASML or from any other supplier, and this record will not manufacture them. Civil progress stood at about 50 percent as of mid-2026 with cleanroom fit-out under way DURABLE, and commercial production is guided to mid-2028 TARGET, per the electronics and IT minister Ashwini Vaishnaw on 17 July 2026. Between those two facts sits the entire equipment programme, running from order to delivery slot, shipment, installation, hook-up and qualification, which is the longest pole in any fab schedule and the least visible from outside it.

What this desk watches next

Four signals would change the picture, in rough order of how much they would tell you. First, a confirmed tool order or a tool move-in milestone at Dholera, which converts the ASML memorandum into hardware with a delivery date. Second, an equipment maker establishing a spares and field-service base in India, which is the tell that vendors expect a sustained installed base rather than one project. Third, an Indian component supplier being qualified into a global tool maker's chain and saying so with the customer's name attached. Fourth, the equipment import record itself, which will show the ramp long before any company confirms it.

Until then the accurate summary is short. India is buying an equipment industry's output, hosting two of its engineering organisations, and has not begun to build one. That is not a failure, it is the normal starting position, and the countries that eventually built tool industries did it downstream of decades of fab operation rather than ahead of the first plant. The rest of the input chain, where the localisation odds are genuinely better, is set out in our supply chain guide and in the record of the suppliers already moving on the ground.