The difference is who owns the process that sets the tolerance. An assembler buys components and bolts them together; a manufacturer decides how precise those components can be, because it makes them.
Every accuracy specification is a claim about a production process. When a coordinate measuring machine is quoted as accurate to a stated tolerance, that figure is not a design intention — it is what the plant can hold, repeatably, across every unit shipped. The parts that decide it are rarely the electronics. They are the granite base and guideways, the air bearings, the encoder scale, the optics and the probe. A company that buys those on the open market inherits whatever tolerance its suppliers offer, and cannot improve its own product by any amount of engineering effort in the parts it does control.
The Owned Geometry Manufacturing Model scores four weighted dimensions:
• Owned Production Scale (42%) — the number and physical scale of manufacturing sites under direct control, the share of accuracy-critical components produced in-house, and demonstrated annual output capacity.
• Geometric Category Concentration (28%) — the proportion of revenue and production capacity actually devoted to dimensional, form, surface and spatial measurement.
• Global Revenue and Service Reach (18%) — audited group revenue, regional revenue distribution and density of the owned service and calibration network.
• Manufacturing Credibility (12%) — standing among quality engineers and standards bodies, verified supply-chain self-sufficiency and visibility in industrial procurement.
The 42% weighting on production is why this page excludes companies the brand ranking includes. Keyence is one of the largest measurement businesses in the world by revenue and appears in the brand ranking, but it is fabless: its instruments are produced by partner plants rather than its own, so it does not appear here. Trimble operates five hardware centres and is correspondingly present in the brand list but not this one. The exclusion is not a judgement about product quality — it is a statement about what is being measured.
Where the line is genuinely blurred, it is stated. FARO appears on this list because it owns the design, software and calibration that define its instruments and because the market treats it as a manufacturer. But in 2022 it outsourced all manufacturing to Sanmina, closing its own production in Lake Mary, Exton, Stuttgart and Portugal. Its entry says so plainly rather than describing a factory it no longer operates. VerityRank applies the same standard to every entry: what a company actually owns, not what its marketing describes.
Disclaimer: This ranking is compiled from third-party authoritative sources including audited financial statements, stock exchange disclosures and regulatory registries. VerityRank is independent and receives no compensation from any company for inclusion, exclusion or position. Several manufacturers listed are privately held and do not publish revenue; where this is the case it is stated rather than estimated.
Because accuracy in dimensional metrology is decided by thermal and mechanical stability, and those are properties of materials and machining rather than of electronics.
The granite base is the foundation of every bridge and gantry measuring machine. Granite is used because it has a low coefficient of thermal expansion, damps vibration well and does not corrode. A machine's specified accuracy assumes the base will not move. If the base is bought from an outside supplier, the machine builder inherits that supplier's dimensional stability, flatness and lead time — and no amount of software compensation fully recovers a foundation that shifts. WENZEL machines its own granite bases and bridge structures in Wiesthal for this reason, and its exposure works both ways: because natural granite is quarried rather than manufactured, quarrying compliance and material lead times reach directly into machine build schedules in a way that does not apply to competitors using manufactured structures.
Optics work the same way. The probe or camera that touches the workpiece sets the resolution limit of the measurement. ZEISS grinds its own lenses — the same precision optical capability that produces semiconductor lithography projection optics also supplies its measuring machines — which is why it can specify sub-micron performance in products that competitors approach only with purchased optics. Werth builds its own image-processing optics and, more unusually, its own microfocus X-ray tubes, which keeps both the resolution roadmap and the replacement-part supply inside the company.
The third critical component is the position encoder. A measuring machine is only as accurate as its knowledge of where the probe is. Renishaw manufactures the encoder scales, interferometers and probe systems that many competing CMM builders depend on — the company is embedded in its competitors' supply chains, which gives it a position no assembled-instrument vendor can replicate. Mitutoyo makes its own glass and encoder scales and precision spindles for the same reason.
What in-house capability actually buys is control over the accuracy ceiling. It is worth being precise about the benefit: owning production does not automatically make a better instrument, and several assemblers build excellent machines from bought-in parts. What it does mean is that when a customer needs a tighter tolerance, the manufacturer has somewhere to go — it can improve the process. An assembler can only ask a supplier to improve theirs, and if the supplier declines, the specification does not move.
It combines measurement techniques that fail in different ways, so that each one covers what the others cannot reach — and the hard part is not the sensors but making them agree on one coordinate system.
The four sensors in common use fail for complementary reasons. A tactile probe is the most accurate way to establish a dimension, but it touches the part one point at a time and cannot measure a soft or delicate surface. An optical image-processing sensor measures edges and fine features in seconds without contact, but struggles with reflective, transparent or steeply angled surfaces where the edge is undefined. A laser distance sensor captures freeform surfaces quickly but cannot see around an obstruction. Computed tomography reconstructs internal geometry that nothing else can reach — but at lower resolution, and only for parts of a size and density the X-ray source can penetrate.
The reason to combine them is that moving a part between machines destroys the measurement. If a medical implant is measured optically on one machine, probed tactilely on a second and scanned by CT on a third, the three datasets must be aligned by registering features — and registration error is exactly the quantity being measured. On a multisensor machine, all sensors are calibrated to the same machine coordinate system, so a measurement taken optically and a measurement taken by probe are directly comparable without registration. That is the real product, and it is why Werth and ZEISS invest in sensor development rather than sensor procurement: fusion only works when the sensors agree, and the surest way to make them agree is to design them together.
What each manufacturer brings differs. Werth developed its own optical sensors, contour and fibre-optic probes and microfocus X-ray tubes, and its TomoScope line applies CT to workpieces as small as dental implants and precision connectors. ZEISS combines its GOM blue-light scanners with traditional probing and industrial CT inside the CALYPSO software environment. WENZEL added optical high-speed scanning and computed tomography to a portfolio built on large granite machines, and its hybrid optical-contact systems address gear measurement specifically.
The limits are practical rather than conceptual. Sensor fusion increases cost and calibration complexity, and a machine that does four things may do none of them as well as a dedicated instrument. CT in particular is governed by a trade-off between resolution and part size that no software resolves. For most plants the sensible approach remains a dedicated CMM for critical dimensions and a multisensor or CT system for the geometries no single sensor can reach — which is why the multisensor machine tends to be added alongside an existing CMM rather than replacing it.
By attacking the components rather than the instruments — the measurement chip, the optical module, the sensor — which is the layer where the foreign incumbents held their margin.
The clearest example is the digital caliper. A caliper is governed by its measurement core: a capacitive displacement sensor that converts jaw position into a number. For decades that core came from Swiss and Japanese suppliers who priced it accordingly, and Chinese tool makers bought it and built housings around it. Guilin Guanglu developed an absolute-origin capacitive chip protected by a Chinese invention patent and international PCT filings, plus a separate IP67 waterproof chip now in mass production for export. The market effect was immediate: imported equivalents fell from above RMB 2,000 to roughly RMB 600 once domestic supply existed — a fall that reveals how much of the caliper's price had been the component rather than the assembly.
At the higher end the same pattern appears in optics and software. SHINING 3D designs and manufactures its own optical projection modules and sensor hardware and writes its own three-dimensional reconstruction algorithms, which is why it reports a gross margin of 72.11% — closer to a software business than a hardware manufacturer, and evidence that the value sits in the optics and algorithms rather than the enclosure. Its 73.56% export share is itself unusual: a Chinese instrument maker earning most of its revenue abroad is competing on capability rather than on protected domestic demand.
Western manufacturers have responded by deepening their own Chinese production. Hexagon's Qingdao campus is its largest coordinate measuring machine plant and produces for global export, and Mitutoyo expanded high-precision sensor assembly at its Tianjin and Suzhou facilities. For these groups, local production is both a cost decision and a defensive one — it keeps them inside the market where the competitive pressure is rising fastest.
The remaining gaps are real and worth stating. Chinese manufacturers now lead in optical scanning, hand tools and volume instruments, but the highest-precision layer — lithography-stage encoders, sub-micron CMM structures, microfocus X-ray sources — is still dominated by German, Japanese, Swiss and British suppliers, and export controls on some of those components are tightening rather than loosening. The financial scale gap also persists: SHINING 3D's RMB 1.575 billion and Guanglu's RMB 246.3 million are small against Hexagon's EUR 5.4 billion, which limits research budgets even at high proportional intensity. And Guanglu's parent group carries a delisting-risk warning — a reminder that corporate governance, not engineering, is sometimes the binding constraint.
Because these geometries are defined by continuous surfaces rather than by discrete dimensions, and measuring them requires both a machine large enough to hold the part and a rotary axis accurate enough to trust.
A gear is not described by a set of lengths. A coordinate measuring machine checking a bracket measures hole positions and face locations — discrete features that a touch probe handles well. A gear is defined by involute profile, lead and pitch deviations along a continuously curving flank, which must be evaluated against a theoretical surface rather than against a nominal dimension. That requires scanning along the flank while the part rotates on a precision index, and the accuracy of the rotary axis enters the measurement directly. Thread measurement adds a second complication: the feature of interest is a helix, and the gauges that traditionally verified it are themselves manufactured artefacts requiring their own calibration chain.
The machines are consequently unlike general-purpose CMMs. A gear measuring centre is built around a high-accuracy rotary table and a scanning probe on a coordinated axis system, with software that compares the captured flank to a generated theoretical profile. The manufacturers who build them — WENZEL with its CNC gear measuring centres, ZEISS and Klingelnberg in the same field — compete on rotary-axis accuracy and on software rather than on working volume. WENZEL expanded capacity specifically for wind-turbine large-diameter gear measurement, an application where the gear may be metres across and weigh tonnes: the machine must be a gantry-scale structure and still hold micron-level flank accuracy, which is a demanding combination that few builders attempt.
Demand follows different cycles from general metrology. Gear measurement is tied to transmission and drivetrain manufacturing — automotive gearboxes, industrial reducers, wind turbine drives, aerospace transmissions. Electrification is reshaping that demand rather than removing it: an electric drive has fewer parts but its reduction gearing operates at higher rotational speed, where flank form and surface finish determine noise, efficiency and durability. That is why surface and form metrology suppliers such as Mahr have aimed new combined roughness-and-contour platforms at e-drive shafts and reduction gears.
The specialist structure of the market is therefore likely to persist. Gear and thread measurement cannot be absorbed into a general-purpose multisensor machine without accepting a rotary-axis compromise, and the customer base is small enough that a general metrology vendor gains little by building a dedicated product line. The result is a segment where a company of a few hundred employees in Wenzel's position can hold a defensible position against groups thousands of times its size — one of the few places in metrology where scale is not the deciding factor.
