Category insight
Wafers & Foundry Services
Week 5
Foundry Wafer Purchases
Processed wafers bought from foundries, priced per wafer by node and options. Leading-node supply is concentrated and 2027 price increases are set.
Why now
October 27, 2026
TSMC is reported to raise wafer prices 3% to 6% from January 2027, most at 2nm and 3nm, and to be weighing increases of up to 20% on CoWoS packaging, TrendForce reported on 27 October, citing Taiwanese press.
TrendForce (citing Liberty Times and Commercial Times)
Read the sourceTypical savings range
Not established for the category; 5% to 9% proxy on should-cost negotiation
From the category playbook
Spend share
Not published at buyer level; usually the largest direct cost line for fabless and fab-lite buyers (as stored)
Criticality
High
Playbook scope:
Manufacturing, Semiconductors (playbook written for fabless and fab-lite buyers of processed wafers; Global)
Point-in-time page: the why-now note reflects reporting on the date shown, and the benchmarks reflect the playbook at publication. This page is not updated as events move on.
At a glance
Processed wafers bought from pure-play or integrated foundries, priced per wafer by process node, mask-layer count and process options, plus multi-project wafer (MPW) shuttles for prototypes and mask NRE. For fabless and fab-lite companies this is usually the largest direct cost line, and its terms are set in long-term agreements that combine volume commitments, prepayments and capacity reservation.
How it is contracted
- Long-term agreements: per-wafer price by node and option, tiered by volume, with annual price adjustment and capacity reservation terms
- Take-or-pay and prepayment: committed wafer starts with shortfall penalties or prepaid capacity
- MPW shuttles: shared mask sets for prototypes on scheduled runs
- NRE: mask and engineering charges billed per tape-out
Benchmarks
- Price index: none published for foundry wafers; quarterly foundry revenue and utilisation commentary is a market signal only
- Market concentration: top-10 foundry revenue US$53.49B in the second quarter of 2026, with TSMC at a 72.5% share (TrendForce, as stored)
- Supplier market: oligopoly, low competitiveness at leading nodes
- Pricing structure: per-wafer price by node and option, volume tiers, NRE for masks
- Should-cost negotiation: 5% to 9% savings reported for category-wide should-cost work in semiconductor procurement (McKinsey, August 2026; a proxy, not validated for this category)
Top five value levers
Most relevant to this disruption
- Should-cost based wafer price negotiation: with 3% to 6% increases reported for January 2027, a node-level cost model is the basis for challenging the increase letter
- Capacity commitment restructuring: commitments set now carry into a tighter 2027, so volume, prepayment and allocation terms matter more than usual
- Mature-node dual sourcing: mature-node prices are also rising as foundries cut 8-inch and 12-inch capacity, so a qualified second source restores leverage
commercial
Rank 1
Should-cost based wafer price negotiation
Use node-level cost models (equipment depreciation, materials, utilisation) to challenge annual wafer price increases. The dominant leading-node foundry rarely moves list price, so the focus is mature nodes and options. Timeline: About 6 months.
Typical impact
5% to 9% (proxy)
supplier
Rank 2
Mature-node dual sourcing
Qualify a second foundry for mature-node (28nm and above) products with multi-year life remaining, to create real price tension. Porting, new masks and 6 to 18 months of requalification can erase savings on short-life parts. Timeline: About 18 months.
Typical impact
Not established; size from client data
commercial
Rank 3
Capacity commitment restructuring
Rebalance take-or-pay volumes, prepayments and reservation fees in long-term agreements to match the demand forecast, shifting committed volume across nodes or extending the period instead of paying shortfall. Released capacity can be hard to regain in an upturn. Timeline: About 6 months.
Typical impact
Not established; size from client data
process
Rank 4
Yield-linked pricing terms
Negotiate known-good-die or yield-baseline clauses so the buyer does not pay full price for low-yield lots, supported by shared sort and inline yield data. Timeline: About 9 months.
Typical impact
Not established; size from client data
specification
Rank 5
Mask-layer and option rationalisation
Remove or consolidate process options, metal layers and special implants that product performance does not need. The benefit lands mostly on new tape-outs. Timeline: About 12 months.
Typical impact
Not established; size from client data
Quick wins
- Reconcile long-term agreement committed volumes against actual wafer starts and flag shortfall exposure (low effort, 4 weeks)
- Audit wafer invoices for adders and options that do not match the latest quote (low effort, 4 weeks)
- Bundle next year's prototype tape-outs into scheduled MPW shuttles (low effort, 6 weeks)
Risk patterns
- Single-foundry concentration at critical nodes: more than 90% of a node's wafer spend sits with one foundry and one fab. Mitigation: map dual-source feasibility and hold a die bank for critical parts.
- Capacity allocation cuts in upturns: purchase order lead times lengthen and allocation letters appear. Mitigation: secure allocation clauses for critical products before the cycle tightens.
- Export-control exposure: wafers bought from entities or regions on BIS lists or for restricted end uses. Mitigation: screen foundry entities against the Entity List and document end-use compliance.
Questions to ask your team
- Which nodes have only one qualified foundry, and what would a second source cost?
- How do committed wafer volumes compare with the current demand forecast?
- Which process options are applied by default rather than by design need?
Data to pull
The files to request
- Wafer purchase history by part, node, foundry, fab site and price for three years
- Long-term agreement terms including committed volume, prepayments and penalties
- Wafer sort yield by lot and product
- MPW and mask NRE invoices by project
- Product roadmap with remaining life by part
Have this data? Run it through Procurement Insights
Upload categorized spend and get savings analysis, a supplier review and negotiation playbooks for this category.
Upload your spendPlaybook confidence
Medium
Benchmark confidence
Low
Levers in playbook
6
