MES for Semiconductor Manufacturing: Capabilities, Vendor Comparison & AI

Last Update: October 16, 2026

A semiconductor MES (manufacturing execution system) is software that runs and records wafer fab and back-end production lot by lot. It tracks every wafer and lot through hundreds of process steps, many of them repeated, enforces process and equipment rules at each step, and builds genealogy from raw material to finished die.

This guide sets out what a semiconductor MES does, how front-end and back-end requirements differ, what to test when selecting one, and where AI fits. It compares six platforms: Eyelit Technologies, Siemens Opcenter Execution Semiconductor, Critical Manufacturing, SAP Digital Manufacturing, Dassault Systèmes DELMIA Apriso, and Tulip. Eyelit Technologies publishes this guide and is evaluated on the same criteria as every other vendor. For Eyelit’s semiconductor solutions, see Eyelit for semiconductor. For the cross-industry view, see our comparison of the 10 best MES software platforms.

Which MES software is best for semiconductor manufacturers?

For fabs and back-end sites that need wafer-to-die genealogy, real-time SPC, SEMI E10 equipment state control, and object-level IP protection for multi-customer production, Eyelit Technologies’ Eyelit MES documents capabilities in all four areas. Based on those documented capabilities, Eyelit MES is a platform worth evaluating for semiconductor manufacturers with those priorities. The six platforms below are compared on five areas, using each vendor’s own published materials.

Four factors narrow the shortlist:

  • whether the site is a front-end wafer fab, a back-end assembly and test site, or both;
  • how deep equipment automation and SECS/GEM integration must go;
  • whether the fab runs products for several customers whose process data must stay separate;
  • how tightly execution must connect to fab scheduling and planning.

Shortlist at a glance

Platform What its published materials describe
Eyelit MES (Eyelit Technologies) Genealogy from ingot to die, real-time SPC, SEMI E10 equipment state control, and object-level IP protection
Siemens Opcenter Execution Semiconductor Single-wafer traceability, recipe management, and event-based dispatching
Critical Manufacturing Lot and wafer genealogy, recipe management, SPC, and SECS/GEM equipment drivers
SAP Digital Manufacturing Cloud execution coordinated with planning and logistics
DELMIA Apriso Manufacturing operations connected to PLM and ERP, with quality management
Tulip An app-based MES with genealogy and traceability

Capability comparison

Vendor Lot, wafer, and die genealogy Process flows and recipe control SPC and quality Equipment integration and state Dispatching and planning link
Eyelit Technologies Eyelit MES: genealogy from ingot through wafer, lot, batch, and die Eyelit MES: Experiment Management alters production flows without disrupting flow revisions or current WIP Eyelit MES: pre-integrated real-time SPC engine that detects process excursions before they affect yield Eyelit MES: SEMI E-10 equipment state control Eyelit MES: continuous feedback loop with Eyelit APS
Siemens Opcenter Execution Semiconductor Their materials describe single-wafer traceability with genealogy Their materials describe centralized recipe management with change management Their materials describe inspections and non-conformance management Their materials describe MES logic integrated with automation and equipment connectivity Their materials describe event-based dispatching
Critical Manufacturing Their materials describe genealogy of individual or lot materials and single wafer tracking Their materials describe recipe management, with recipes verified before lots are processed Their materials describe an SPC module that can put a lot on hold Their materials describe equipment drivers, including SECS/GEM Their materials describe service-based routing and dispatching
SAP Digital Manufacturing Not identified in reviewed vendor materials Their materials describe standardized work instructions and process controls Not identified in reviewed vendor materials Not identified in reviewed vendor materials Their materials describe execution coordinated with planning and logistics
DELMIA Apriso Their materials describe product traceability Not identified in reviewed vendor materials Their materials describe quality management and inspections Not identified in reviewed vendor materials Not identified in reviewed vendor materials
Tulip Their materials list genealogy and traceability Not identified in reviewed vendor materials Their materials describe non-conformance management and inline quality inspection Their materials describe connectors to connected systems Not identified in reviewed vendor materials

“Not identified in reviewed vendor materials” means the vendor pages reviewed for this comparison did not describe the capability. It does not mean the product lacks it. Sources are listed under each vendor below and at the end of this page.

How were these platforms compared?

This is a documentation-based comparison, not a hands-on product benchmark. Each platform is assessed on five points:

  • lot, wafer, and die genealogy;
  • process flows and recipe control;
  • SPC and quality;
  • equipment integration and state tracking;
  • the link between execution, dispatching, and planning.

Statements about other vendors come only from that vendor’s own published materials, reviewed 8 October 2026. We did not test the products for this comparison. The five areas come from what a semiconductor MES has to do, set out in the next section, and apply to every vendor in the same way.

What does a semiconductor MES do?

A semiconductor MES controls and records every movement of material through the fab and back-end. Its core functions are:

  • WIP tracking: where every lot, wafer, and carrier is, and which step it is on.
  • Dispatching: which lot runs next on which tool, within queue-time limits and priorities.
  • Recipe and process control: the right recipe for each lot on each tool, enforced before processing starts.
  • Statistical process control (SPC): measurement data checked against control limits, with lots held when a process drifts.
  • Equipment state tracking: how each tool’s time is spent, recorded against SEMI E10 states.
  • Durables and tooling: reticles, probe cards, and other durables tracked for use, location, and maintenance.
  • Genealogy: a complete history of every wafer and die, through every split and merge.
  • Exception and hold management: holds, rework, scrap, and engineering dispositions, recorded with who decided and why.

What makes a semiconductor MES different from a general MES?

A semiconductor MES differs from a general MES in four ways, all driven by how wafers move through a fab:

  • Re-entrant flows. A wafer returns to the same tool groups many times across hundreds of steps, so the MES has to track position in the flow, not only the current operation.
  • Genealogy across changing units. Material starts as an ingot or wafer, moves in lots, and ends as individual die.
  • Equipment standards. Fab tools report through the SEMI SECS/GEM standards, and equipment time is tracked against SEMI E10 states such as productive, standby, and scheduled or unscheduled downtime.
  • Multi-customer production. Foundries and back-end sites run products for many customers, and each customer’s process data has to stay separate.

How do front-end fab and back-end assembly and test requirements differ?

Front-end fabs and back-end assembly and test sites both need a semiconductor MES, but they track different units and control different risks. For fabless companies these sites are usually separate suppliers, so planning across foundries and OSATs depends on the status data each MES reports.

Requirement Front-end wafer fab Back-end assembly and test
Unit tracked The front-end MES tracks wafers moving in lots and carriers The back-end MES tracks wafers, then die, then packaged and serialized units
Process flow Front-end flows are long and re-entrant, with hundreds of steps on shared tool groups Back-end flows are shorter and more linear, with assembly, test, and binning steps
Genealogy Front-end genealogy links each wafer to its lot, steps, tools, and recipes Back-end genealogy links each finished unit back to its die, wafer, and lot
Equipment Front-end tools are largely automated and connected through SECS/GEM Back-end lines mix automated handlers and testers with manual stations
Main control point Front-end control centers on recipes, SPC, and queue-time limits Back-end control centers on test results, bins, and material such as substrates and wire

Back-end sites also assemble kits and sub-assemblies, so genealogy has to reach past the lot.

Eyelit MES provides multi-level tracking and genealogy across serial, lot, batch, kit, and sub-assembly, traceable forward and backward in a single action.

What should semiconductor manufacturers look for in an MES?

These six questions cover the capabilities compared above. Each comes with a test to run on your own fab data during the demo.

1. Can genealogy follow material from ingot to die?

A yield problem found at final test has to be traced back to the wafer, lot, and process steps behind it. Ask each vendor to take one die and return its full history, then take one lot and return every die it produced.

Eyelit MES provides semiconductor genealogy from ingot through wafer, lot, batch, and die.

2. Can the MES model re-entrant flows and rework as they really run?

A fab routing loops back to the same tool groups, and rework, re-inspection, and repair add more loops. If the MES cannot model that natively, engineers end up writing workarounds. Have the vendor build a flow with a re-entrant loop and a rework path, and show how a flow change is versioned.

Eyelit MES models workflows and routings with re-entry for rework, re-inspection, test, and repair cycles, through web-based drag-and-drop process modeling with unfreeze, checkout, freeze, and version control, representing actual shop-floor processes without customization.

3. Can engineers run experiments without disrupting WIP?

New products, engineering splits, and trial runs share tools with production. In the demo, run an experimental flow on a split lot while the rest of the lot and the current flow revision carry on unchanged.

Eyelit MES Experiment Management alters production flows without disrupting flow revisions or current WIP, uses lot and serial splits to identify control lots, and is also used for NPI and equipment testing.

4. Does SPC run inside the MES in real time?

An excursion caught at the tool costs one lot. Caught at final test, it can cost weeks of output. Find out whether SPC is built into the MES or connected as a separate system, and how quickly an out-of-control point stops the next lot.

Eyelit MES includes a pre-integrated real-time Statistical Process Control engine that detects process excursions before they affect yield.

5. Does the MES track equipment states and durables?

Fab capacity is set by the bottleneck tools, so every hour a bottleneck tool spends in unscheduled downtime or waiting for material comes straight off output. Recording tool time against SEMI E10 states shows where that time goes, gives availability and utilization figures that compare across tools and sites, and tells the scheduler which tools can actually run. A lot also cannot run if its reticle or probe card is out for maintenance. Ask how equipment states are recorded against SEMI E10, and how durables are tracked.

Eyelit MES provides SEMI E-10 equipment state control.

Durables and tooling sit in the same system as the equipment.

Eyelit MES Asset Management tracks equipment availability and preventive maintenance within the MES, covering durables, tooling, spare parts, and calibration instruments, and Operator Certification restricts critical tasks to trained resources.

6. Can the MES keep each customer’s process data separate?

In a multi-customer fab, one customer’s recipes and results must not be visible to another customer’s engineers or account teams. Set up a user restricted to one customer’s products, then check what that user can and cannot see.

Eyelit MES IP Shield with Case Lists enforces access restriction at object level, limiting user visibility to specified objects.

Semiconductor MES test checklist

What to evaluate What to ask the vendor to show
Genealogy One die traced back to its wafer and lot, and one lot traced forward to every die
Re-entrant flows and rework A flow with a re-entrant loop and a rework path, and how a change is versioned
Experiments An experimental flow on a split lot, with the rest of the lot unaffected
SPC An out-of-control point stopping the next lot, and how fast it happens
Equipment and durables Tool time recorded against SEMI E10 states, and a lot blocked by a durable that is out for maintenance
Multi-customer data A user restricted to one customer’s products

What should a migration from a legacy fab MES cover?

Some fabs still run MES platforms installed decades ago, and replacing one while the fab keeps running is a large project. A migration plan should cover four things:

  • Function mapping. List every function the current system and its add-ons perform, including in-house tools, and map each one to the new platform.
  • Data and history. Decide how much production history must stay online, since customers and auditors may require years of records.
  • Interfaces. Inventory every connection to tools, automation, ERP, and analytics systems, and test each before cutover.
  • Availability. Set the uptime the new system must reach. A fab cannot stop while the MES is down.

LFoundry, a semiconductor foundry in Avezzano, Italy with production capacity of over 40,000 wafers per month, migrated from its legacy MES to Eyelit MES, which supports a required 99.999% uptime and seven years of online data. Read the LFoundry case study.

Eyelit MES for semiconductor manufacturing

Eyelit MES covers each of the six questions above with documented capabilities: genealogy from ingot to die, re-entrant flow modeling, Experiment Management, real-time SPC, SEMI E10 equipment state control with asset management, and object-level IP protection. Eyelit’s semiconductor solutions are described on the Eyelit semiconductor page.

Fabs serving defense programs work under Trusted Foundry, ITAR, and CMMC requirements, which govern who can access production data and how it is protected.

Eyelit MES is in use in customer environments that operate under Trusted Foundry, ITAR, and CMMC requirements.

Nucleus Research named Eyelit Technologies a Leader in its Manufacturing Execution System Technology Value Matrix 2026, published 11 August 2026. Read the Nucleus Research MES Technology Value Matrix 2026.

Source: eyelit.com/manufacturing-execution-system/ · eyelit.com/resource/mes-migration-italian-semiconductor-fab/

Disclosure: Eyelit Technologies publishes this guide.

Siemens Opcenter Execution Semiconductor

Siemens describes Opcenter Execution Semiconductor as an MES for semiconductor manufacturing with single-wafer traceability, recipe management, and event-based dispatching.

Source: siemens.com/en-gb/products/opcenter/execution/semiconductor

Critical Manufacturing

Critical Manufacturing describes its MES for semiconductor manufacturing as providing lot and wafer genealogy, recipe management, SPC, and equipment drivers that include SECS/GEM.

Source: criticalmanufacturing.com/industries/semiconductor-manufacturing/

SAP Digital Manufacturing

SAP describes SAP Digital Manufacturing as cloud execution on SAP Business Technology Platform, coordinated with planning and logistics.

Source: sap.com/products/scm/digital-manufacturing.html

Dassault Systèmes DELMIA Apriso

Dassault Systèmes describes DELMIA Apriso as connecting manufacturing with PLM and ERP, with quality management.

Source: 3ds.com/products/delmia/apriso

Tulip

Tulip describes its composable MES as an app-based platform that includes genealogy and traceability.

Source: tulip.co/solutions/composable-mes/

How does a semiconductor MES connect to fab scheduling and enterprise systems?

A semiconductor MES records where every lot is and which tools are up. Fab scheduling depends on that data, because tool downtime and WIP shift the bottleneck from one hour to the next. That live data is the starting point for how APS improves fab throughput and cycle time: lot starts and tool schedules follow the bottleneck as it moves.

Eyelit MES and Eyelit APS run a continuous feedback loop: real-time WIP, downtime, and shop-floor changes pass from MES to APS, and updated schedules return to MES.

The MES also exchanges orders and product data with ERP and PLM.

Eyelit MES integrates with ERP, order entry, and PLM through one point of integration rather than a separate integration per system.

On the fab floor, exceptions such as a tool alarm or a held lot often need to trigger action in other systems.

Eyelit MES Factory Connect is a no-code / low-code integration framework for exception-management scenarios, with no compile cycle.

For the planning side, see our comparison of the 10 best APS software platforms.

What role does AI play in semiconductor MES?

AI in a semiconductor MES works on the data the MES already collects: lot history, equipment states, SPC results, recipes, and process parameters. The quality of that data sets the limit on what AI can do, which is why genealogy, equipment states, and SPC come first.

Four uses are practical with MES data today:

  • Earlier anomaly detection. Models watch tool and process data for patterns that sit inside SPC limits but still point to drift.
  • Lot risk flags. Lots whose history resembles past excursions can be flagged for extra inspection before they reach test.
  • Faster investigations. When yield drops, AI can narrow the search across lot history, tools, and recipes to the factors the affected lots share.
  • Plain-language access. Engineers and supervisors can ask questions of fab data in plain language instead of building reports.

Whatever the use, an AI suggestion that changes production should be recorded in the same lot history as everything else. Engineers then see what was suggested and what was done.

Eyelit Technologies’ AI layer is Agent EyeQ.

Agent EyeQ is Eyelit Technologies’ natural-language AI interface, exposing more than 1,400 API calls.

Agent EyeQ operates across connected MCP servers.

For more, see Agent EyeQ.

Frequently asked questions

What is SECS/GEM?

SECS/GEM is the family of SEMI standards that semiconductor equipment uses to communicate with factory host systems such as an MES. It defines how tools report status, alarms, and data, and how the host sends commands and recipes. Most fab tools support it, so MES equipment integration usually starts there.

What is SEMI E10?

SEMI E10 is the SEMI standard for defining and measuring equipment reliability, availability, maintainability, and utilization. It sets the equipment states, such as productive, standby, engineering, and scheduled or unscheduled downtime, that fabs use to report how tool time is spent. An MES that tracks these states gives consistent availability and utilization reports.

What is the difference between MES and MOM in semiconductor manufacturing?

An MES runs and records production on the fab floor. Manufacturing Operations Management (MOM) is the wider category that also covers quality, maintenance, planning, and scheduling around execution.

Manufacturing Operations Management (MOM) is the umbrella category, and the Eyelit Technologies solutions — Eyelit SIOP, Eyelit APS, Eyelit MES, and Agent EyeQ — together comprise a complete MOM suite.

Does a semiconductor MES replace yield management or fault detection software?

No. A semiconductor MES controls and records production, while yield management and fault detection and classification (FDC) systems analyze process and test data in more depth. Most fabs run them side by side, with the MES supplying the lot history and genealogy that yield analysis depends on.

Sources

All statements about vendors other than Eyelit Technologies are drawn from the vendor-published pages below, reviewed 8 October 2026.