The Fabless Model and Its ERP Requirements
Fabless semiconductor companies design chips but outsource manufacturing to foundries and OSAT providers. This model --- pioneered by companies like Qualcomm, Broadcom, and NVIDIA --- eliminates the $10-20 billion capital cost of building a fab, as documented in SIA's annual factbook . But it creates a different set of operational challenges that generic ERP cannot address.
A fabless company's core operations span design handoff, foundry management, supply chain coordination, and customer delivery --- all without owning a single manufacturing tool. The ERP must orchestrate these external relationships while maintaining visibility into processes the company does not directly control.
Fabless-Specific ERP Requirements
Design-to-Production Handoff
The transition from chip design to volume production is the most critical process for fabless companies. The ERP must manage:
- Tapeout tracking — managing GDSII file submissions to foundries with revision control
- Mask set management — tracking mask orders, costs, and revisions across photomask suppliers
- PDK compliance — ensuring designs meet foundry process design kit requirements
- NPI milestones — tracking engineering lots, qualification lots, and production ramp
Each new product goes through multiple silicon iterations (A0, B0, C0) before reaching production. The ERP tracks each iteration's test results, design changes, and qualification status.
Foundry and OSAT Management
Fabless companies typically work with multiple foundries and OSAT providers. The ERP must coordinate:
- Multi-foundry strategies — running the same product at two or more fabs for supply security
- Work-in-progress visibility — tracking lot status at external facilities in near real-time
- Yield monitoring — comparing yield performance across foundries
- Quality management — managing outgoing quality metrics and customer returns by foundry source
FlowSense Semiconductor provides a supplier portal that gives fabless companies real-time visibility into foundry operations without requiring access to the foundry's internal MES.
Intellectual Property Protection
Fabless companies' primary assets are their chip designs. As SEMI's IP protection guidelines recommend, the ERP must enforce:
- Technology control plans — restricting design file access based on ITAR/EAR classification
- Need-to-know access — limiting visibility to specific product programs
- Audit trails — tracking every access to sensitive design and test data
- Export control screening — verifying that design transfers comply with regulations
Revenue Recognition and Pricing
Fabless semiconductor pricing is complex:
- Wafer pricing — cost per wafer die plus test, packaging, and logistics
- Die pricing — selling individual die to customers who handle their own packaging
- Consignment models — customer-owned wafers processed at contracted facilities
- Volume-based tiering — pricing that changes with quarterly volume commitments
- NRE (Non-Recurring Engineering) — one-time charges for mask sets, IP licenses, and design services
Generic ERP pricing modules cannot model these semiconductor-specific structures without extensive customization.
Key Selection Criteria for Fabless ERP
1. Multi-Source Visibility
The ERP must provide a unified view across all external manufacturing partners. A single dashboard showing WIP status at TSMC, assembly progress at ASE, and test throughput at Amkor is essential for fabless operations management.
Beyond basic lot status, fabless operations teams track specific metrics across their manufacturing partners to manage performance and identify issues early. Foundry-level metrics include wafer-out yield by product and technology node, cycle time from wafer start to ship (target versus actual), on-time delivery rate against committed ship dates, and line yield at critical process steps such as lithography and etch. At the OSAT level, the key metrics are assembly first-pass yield, test yield by insertion (wafer sort, final test, QA), DPPM (defective parts per million) trends by package type, and unit-level test time which directly affects cost per unit. The ERP aggregates these into a supplier scorecard that operations reviews weekly, comparing performance across foundries running the same product and across OSATs handling the same package type. Trend alerts flag when any metric degrades beyond a configurable threshold --- for example, when a foundry's wafer-out cycle time exceeds the rolling 12-week average by more than 10%, or when an OSAT's test yield drops 0.5% below the qualified baseline. This data also feeds into the demand planning system, as described in our semiconductor supply chain management guide.
2. NPI Program Management
New product introduction is the lifeblood of fabless companies. The ERP should track silicon iterations, qualification milestones, and production ramp timelines with gate-based approval workflows.
3. Demand Planning with Foundry Lead Times
Fabless demand planning must account for 16-26 week total lead times, capacity booking deadlines at foundries, and the lag between placing wafer start orders and receiving finished goods.
4. Test Data Management
Even without owning test equipment, fabless companies receive vast amounts of electrical test data from OSAT providers. The ERP must ingest, store, and analyze this data for yield trending, product qualification, and customer quality documentation.
5. Cost Management
With manufacturing outsourced, cost management focuses on:
- Wafer cost optimization across foundries
- Assembly and test cost negotiation
- Logistics and tariff management for global supply chains
- NRE amortization across product volumes
Engineering Change Order (ECO) Management
Design revisions are a constant in fabless semiconductor development. A timing closure failure discovered during physical verification, a reliability issue found in qualification testing, or a customer-requested feature addition all generate engineering change orders that must propagate across the fabless company's network of foundry and OSAT partners. The ERP manages this ECO lifecycle end to end: the design team initiates the change request with a description, affected mask layers, and impact assessment; the operations team evaluates the WIP impact, identifying lots already in process that are affected; and the supply chain team coordinates with the foundry on mask revisions and revised tapeout schedules.
The critical complexity is version control across distributed partners. When an ECO advances the design from revision B0 to B1, the ERP must track which foundry lots are running B0 silicon, which mask sets correspond to each revision, and which OSAT test programs need updating for the new revision. Lots already past the affected process layers continue on B0 test flows, while new wafer starts use the B1 mask set and B1 test program. The ERP prevents cross-contamination by enforcing revision-level constraints at every handoff point --- a B1 wafer cannot be tested with a B0 test program, and a B0 die cannot be assembled into a package qualified only for B1. This revision integrity is especially critical for safety-regulated products under IATF 16949 or AS9100, where mixing silicon revisions within a single shipment constitutes a quality escape.
Customer Demand Visibility
Fabless semiconductor companies serve customers whose demand signals arrive through multiple channels --- direct purchase orders, distributor forecasts, design-win pipeline projections, and long-term supply agreements. The ERP aggregates these signals into a unified demand picture that drives foundry capacity booking decisions. Without this aggregation, operations teams piece together demand from email threads, CRM notes, and sales spreadsheets, leading to either over-commitment (booking more foundry capacity than actual demand supports) or under-commitment (missing allocation windows and facing supply shortfalls).
The demand aggregation engine normalizes customer forecasts into a common weekly or monthly bucket structure, applies confidence weights based on signal source (firm POs carry higher weight than distributor forecasts), and reconciles against existing backlog and inventory positions. A demand waterfall view shows the progression from design-win pipeline to qualified demand to firm orders, giving the planning team visibility into how much of the forecasted volume has actually converted. The system also detects demand anomalies --- a customer whose orders suddenly double may be building safety stock rather than reflecting genuine end-market demand, a signal the operations team needs to validate before committing additional foundry capacity. This demand intelligence directly feeds the supply chain planning functions that determine wafer start quantities and allocation strategy.
Quality Monitoring Without Fab Access
Fabless companies face a fundamental quality challenge: they are accountable for the quality of their products but do not have direct access to the manufacturing floor where quality is built in. The ERP addresses this through systematic remote quality monitoring. Foundries and OSAT partners transmit standardized quality data --- SPC charts, defect pareto summaries, lot acceptance test results, and outgoing quality audit samples --- through automated data feeds that the ERP ingests and analyzes without requiring physical presence at the partner facility.
The quality monitoring module applies the fabless company's own control limits and acceptance criteria to incoming data, which may be tighter than the foundry's standard limits. When a process parameter at the foundry drifts toward the fabless company's internal spec limit (even if it remains within the foundry's broader limits), the system alerts the fabless quality engineer to engage with the foundry's process team before the issue affects product. For incoming material from OSAT partners, the ERP manages skip-lot inspection schemes based on supplier quality history --- partners with a consistent track record undergo reduced incoming inspection, while new partners or those with recent quality events face tighter scrutiny. This data-driven approach to remote quality management replaces the ad-hoc audit visits and reactive firefighting that characterize fabless quality programs built on generic ERP or spreadsheet tracking. For details on how lot-level traceability supports these quality workflows, see our wafer lot tracking guide.
Implementation for Fabless Companies
Fabless ERP implementation differs from fab-based deployment:
- 1Start with supply chain — foundry and OSAT coordination is the highest priority
- 2Integrate test data feeds — connect to OSAT data portals for automated yield reporting
- 3Set up NPI tracking — establish milestone workflows for silicon programs
- 4Configure demand planning — model your specific lead time structure and booking cadence
- 5Enable compliance modules — ITAR/EAR controls are especially critical for fabless defense suppliers
Built for fabless operations. See how FlowSense Semiconductor supports your model.
