The Manufacturing cost per Good die (MCGD) is a manufacturing-specific key performance indicator (KPI) for wafer manufacturing. It assigns historically valued manufacturing costs of a released final quantity to the number of quality-verified good dies at a technically defined internal Completion Point. It is not a general profitability metric, but a control metric tailored to the actual cost generation in internal wafer manufacturing.

Why is a simple "cost per chip" calculation not sufficient?

In semiconductor manufacturing, there is no simple 1:1 relationship between cost accrual and released output. Manufacturing spans many process steps and multiple periods; actual wafer routes deviate from planned routes; yield losses reduce the released output; and a substantial portion of costs follows time-, state-, or pool-based logics. A naive division of monthly costs by an arbitrary output quantity would systematically lead to distortions.

What costs are included in the MCGD – and what is excluded?

The MCGD covers four operational cost classes: DPM (direct process-related materials), EUM (energy, utilities, and media), COO (equipment-related ownership costs), and MOH (indirect production-related manufacturing overhead). Excluded are one-off design and development efforts, mask sets, tape-out, packaging, final test, as well as sales, marketing, and corporate overhead costs.

How does the MCGD treat yield?

Yield is not an additive cost class on top of already incurred manufacturing costs. Instead, yield operates on two levels: first, as a denominator effect, because losses reduce the number of quality-verified good dies; and second, as an analytical reporting view that partitions already incurred costs into good-output and yield-loss shares without creating a second cost incidence or double counting.

What is the "Completion Point" and why does it matter?

The Completion Point (CP) is the technically defined internal state at which wafer manufacturing is considered complete within the chosen scope. It serves as the reference point where good-die output and total die quantities are determined. A stable and quality-approved CP is essential for the metric to be meaningful and comparable.

What does "valuation at the event time" mean?

A released completion quantity typically contains pass events from several earlier periods. Valuation is performed at the time each pass event actually occurred — using the prices, tariffs, or burden rates that were valid at that moment — rather than at the rates of the reporting or release month. This prevents retroactive distortions when rates change over time.

What role does genealogy play in the MCGD framework?

Wafer-level genealogy — covering split, merge, rework, partial scrap, hold, and special wafer types — is critical for robust cost allocation and reliable yield attribution. The framework requires a stable, historized wafer identity as the primary reference for pass events, quantities, and genealogical operations. Without this traceability, consistent cost assignment is not achievable.

What is the "Industrial Data Platform" in this context?

The Industrial Data Platform is the proposed central integration and evaluation layer for computing the MCGD. It brings together data from ERP, PLM, MES, equipment-level systems, yield and quality systems, and Finance/Controlling. The source systems remain authoritative for their respective data domains; the platform performs the cost-relevant linking, valuation, snapshot creation, and release of the KPI.

How does the MCGD relate to existing standards like SEMI E35 or IEC 62264-1?

No single standard provides the complete MCGD model. Instead, the framework integrates complementary building blocks from multiple standards: SEMI E35 for Cost-of-Ownership and yield-loss perspective, IEC 62264-1 for resource and routing modeling, SEMI E10 for equipment state semantics, SEMI S23 for energy and utility consumption, ISO 10303-44 for product structure, and NASA-HDBK-0008 for Bill of Materials. The MCGD closes the integration gap that none of these standards individually addresses.

What are the main limitations of the MCGD approach?

Key limitations include: the Completion Point must be technically stably defined; lot-level fallbacks (instead of wafer-level genealogy) reduce cost and yield allocation precision; high shares of burden-based MOH allocation shift the metric away from process-near evaluation; comparability between fabs is limited if scope rules, burden bases, or MOH mappings differ; and if no good dies are released in a period, the metric is undefined. The MCGD is presented as an operationalizable reference model — empirical fab validation is identified as a topic for further research.

Close-up of semiconductor packaging process. Computer chips are being extracted by a pick and place machine from wafer and attached to substrate. Computer chip manufacturing at factory.

Article

Manufacturing cost per good die (MCGD) One KPI. Full cost transparency.

Get the full MCGD paper

Move beyond simplified cost views and understand what drives your manufacturing performance

In semiconductor manufacturing, determining reliable costs is inherently complex. Costs accumulate over multiple process steps and time periods, while the relevant output, quality-approved good dies, is only known at the end of production.

Traditional cost metrics often fail to reflect this complexity.

Manufacturing cost per good die (MCGD) addresses this gap by providing a consistent, manufacturing-near KPI that connects cost, yield, and financial logic into a single, operational framework.

Why existing cost approaches are not sufficient
Common approaches such as unit cost or monthly averages introduce systematic distortions:

Costs and output are not aligned in time
Yield losses are only partially reflected
Actual routing and rework are not fully considered
Indirect costs are aggregated without clear causation

As a result, cost figures often lack the precision required for operational decision-making.
How MCGD solves the problem
MCGD establishes a standards-consistent, manufacturing-driven cost framework and ensures that:

Costs and quantities refer to the same released completion quantity
All cost components are evaluated at the time of each manufacturing event (event-time valuation)
Yield acts as a denominator effect, not as an additional cost
What MCGD adds to your cost perspective
MCGD links historically valued manufacturing costs directly to the released good-die output at a defined completion point. This enables:

Transparent allocation of manufacturing costs to actual output
Consistent integration of yield effects without double counting
Alignment of production data with financial reporting
Reliable comparability across products, process segments, and time periods
How can MCGD be operationalized
MCGD is implemented within an Industrial Data Platform that integrates data from MES, ERP, PLM, and equipment systems. This enables the following additional use cases:

Wafer-level traceability
Event-based cost evaluation
Governed KPI calculation
Reconciliation with financial reporting

Where does your organization stand in terms of cost transparency?

Manufacturing environments differ – especially when integrating cost, yield, and production data. Our pulse check aims to assess the current state across organizations, from data availability and KPI definitions to operational use.

Complete the Industry Pulse Check

to receive the full MCGD cost framework.

After completion, we also share the aggregated and anonymized survey results as soon as they are available, so you can benchmark your position. The survey takes only 2–3 minutes.

Form is loading...

Next Step:

Question 1
Question 2
Question 3
Personal data

How relevant do you consider wafer- or chip-level manufacturing cost calculation as an alternative to periodic overhead costing?

1 = Not relevant
5 = Extremely relevant

Where do you currently see the biggest challenges in determining wafer- or chip-level manufacturing costs?

1 = Not a challenge
5 = Critical challenge

The data collected to calculate manufacturing costs with MCGD also forms the basis for other use cases. How relevant are the following additional use cases for you?

1 = Not relevant
5 = Extremely relevant

If you want to receive relevant updates on MCGD and a summary of the survey results, please make sure to activate the checkbox below and confirm your consent when you receive our registration confirmation via email.

If you want to have more information on data processing at ZEISS, please refer to our data protection notice.

The Manufacturing cost per good die (MCGD) is a manufacturing-specific key performance indicator (KPI) for wafer manufacturing. It assigns historically valued manufacturing costs of a released final quantity to the number of quality-verified good dies at a technically defined internal Completion Point. It is not a general profitability metric, but a control metric tailored to the actual cost generation in internal wafer manufacturing.
In semiconductor manufacturing, there is no simple 1:1 relationship between cost accrual and released output. Manufacturing spans many process steps and multiple periods; actual wafer routes deviate from planned routes; yield losses reduce the released output; and a substantial portion of costs follows time-, state-, or pool-based logics. A naive division of monthly costs by an arbitrary output quantity would systematically lead to distortions.
The MCGD covers four operational cost classes: DPM (direct process-related materials), EUM (energy, utilities, and media), COO (equipment-related ownership costs), and MOH (indirect production-related manufacturing overhead). Excluded are one-off design and development efforts, mask sets, tape-out, packaging, final test, as well as sales, marketing, and corporate overhead costs.
Yield is not an additive cost class on top of already incurred manufacturing costs. Instead, yield operates on two levels: first, as a denominator effect, because losses reduce the number of quality-verified good dies; and second, as an analytical reporting view that partitions already incurred costs into good-output and yield-loss shares without creating a second cost incidence or double counting.
The Completion Point (CP) is the technically defined internal state at which wafer manufacturing is considered complete within the chosen scope. It serves as the reference point where good-die output and total die quantities are determined. A stable and quality-approved CP is essential for the metric to be meaningful and comparable.
A released completion quantity typically contains pass events from several earlier periods. Valuation is performed at the time each pass event actually occurred — using the prices, tariffs, or burden rates that were valid at that moment — rather than at the rates of the reporting or release month. This prevents retroactive distortions when rates change over time.
Wafer-level genealogy — covering split, merge, rework, partial scrap, hold, and special wafer types — is critical for robust cost allocation and reliable yield attribution. The framework requires a stable, historized wafer identity as the primary reference for pass events, quantities, and genealogical operations. Without this traceability, consistent cost assignment is not achievable.
The Industrial Data Platform is the proposed central integration and evaluation layer for computing the MCGD. It brings together data from ERP, PLM, MES, equipment-level systems, yield and quality systems, and Finance/Controlling. The source systems remain authoritative for their respective data domains; the platform performs the cost-relevant linking, valuation, snapshot creation, and release of the KPI.
No single standard provides the complete MCGD model. Instead, the framework integrates complementary building blocks from multiple standards: SEMI E35 for Cost-of-Ownership and yield-loss perspective, IEC 62264-1 for resource and routing modeling, SEMI E10 for equipment state semantics, SEMI S23 for energy and utility consumption, ISO 10303-44 for product structure, and NASA-HDBK-0008 for Bill of Materials. The MCGD closes the integration gap that none of these standards individually addresses.
Key limitations include: the Completion Point must be technically stably defined; lot-level fallbacks (instead of wafer-level genealogy) reduce cost and yield allocation precision; high shares of burden-based MOH allocation shift the metric away from process-near evaluation; comparability between fabs is limited if scope rules, burden bases, or MOH mappings differ; and if no good dies are released in a period, the metric is undefined. The MCGD is presented as an operationalizable reference model — empirical fab validation is identified as a topic for further research.