Wafer Demand Forecast Console
Unit demand becomes wafer demand through yield and die size: wafers = units ÷ (dies per wafer × yield). Convert your shipment forecast into wafers per year, project it with a growth rate, and read the peak wafer-start capacity (wspm) you must reserve — the number foundries allocate in.
Unit demand, die area & yield → wafers and peak capacity.
Multi-year wafer ramp console
Reserve to the peak year (14,574 wspm), not year one — foundry capacity is booked years ahead and must cover your busiest period.
Your demand needs about 14,574 wspm at peak. Each wafer yields 530 good dies at 85% yield; at 100mm² a wafer holds 623 dies.
Double the die area and you'd need roughly twice the wafers for the same units — die size drives capacity as much as demand. Forecast per node; capacity is not interchangeable.
Sharpen the inputs in the Die Per Wafer and Yield consoles.
Why wafer forecasting matters
A million chips isn't a million wafers' worth — it's units ÷ (dies per wafer × yield). A big, low-yielding AI die needs far more wafers per shipped unit than a tiny high-yielding MCU, which is why die size drives capacity planning as much as demand does.
Leading-edge wafer capacity is allocated long before products ship, so forecasting wafer demand accurately — and early — is how companies secure supply. Underforecast and you're short; overforecast and you've prepaid for idle capacity.
Small changes in end demand magnify up the supply chain into large swings in wafer orders, which is how the industry lurches between shortage and glut. A disciplined forecast tied to real unit demand and yield is the antidote.
Surging demand for large AI accelerator dies — low-yielding and wafer-hungry — has strained leading-edge and advanced-packaging capacity disproportionately, because each unit consumes far more wafer area than a conventional chip.
From shipment forecast to wafer starts
Capacity planning in semiconductors starts with a deceptively simple conversion that teams routinely get wrong: turning a forecast of how many chips you'll sell into how many wafers you must buy. The two numbers are not proportional, because what sits between them — die size and yield — varies enormously by product, and that variance is exactly what makes a small AI accelerator program consume more wafer capacity than a giant microcontroller program.
The conversion is wafers = units ÷ (dies per wafer × yield). Dies per wafer falls with die area, so a large die yields few units per wafer; yield is the working fraction, so a low-yielding part wastes wafers. Multiply the two effects and a 700mm² AI die at 60% yield can need tens of times more wafers per shipped unit than a 12mm² automotive MCU at 92% yield. This is why the AI build-out strained leading-edge and packaging capacity far beyond what its unit volumes alone would suggest — each unit is simply wafer-hungry.
Because fabs take years to build and capacity is reserved in advance, the forecast has to look forward, and growth compounds. A high-growth product can triple its volume over a few years, so the peak-year wafer demand — and the wafer-start capacity you must secure — sits well above year one. Reserving to year one and hoping to add capacity later is how companies end up unable to supply the demand they created.
Use this alongside the rest of the suite: the Die Per Wafer and Yield calculators sharpen the two inputs that drive the conversion, and the Fab ROI console turns the capacity you need into the investment to build it.
Trusted by Supply Planning & Capacity Teams
“The die-size-to-wafer leverage is the insight planners miss — this makes it obvious why our AI part consumes 20× the wafers per unit of our MCU. I use the peak-year wspm directly in capacity reservation conversations with our foundry.”
“Clean unit-to-wafer conversion with growth. I run it per node to spot where customers will be constrained. The bullwhip reality anchor is exactly the conversation we have every cycle.”
“Helped us reserve the right capacity for a high-growth product before the shortage hit. Would love built-in yield-ramp modeling, but projecting to the peak year and reading wspm is exactly what I needed.”
“For long-life automotive parts the multi-year view is essential. Small die, high yield, huge volume — and this shows the steady wafer demand clearly. The per-node guidance saved us from an aggregate-forecast mistake.”
Love using our calculator?
Related tools
Similar Calculators
More tools in the same category
Wafer Cost Calculator
Model total chip manufacturing cost with precision, integrating wafer procurement, die yield curves, packaging tiers, and test economics. Compare multi-node production scenarios and profitability trajectories across fabless, IDM, and foundry business models with real-time fab pricing benchmarks.
Die Per Wafer Calculator
Estimate usable dies per wafer using advanced edge-exclusion algorithms, scribe-line optimization, and die-size scaling across 200mm to 450mm wafers. Analyze manufacturing efficiency, reticle utilization, and output potential with support for rectangular dies, multi-project wafers, and defect-cluster-aware yield mapping.
Yield Calculator
Calculate expected wafer yield from defect density (D0), die area, and critical area models using Poisson, Bose-Einstein, and Murphy yield models. Visualize yield curves, process-window sensitivity, and production impact across mature and advanced nodes with Monte Carlo statistical simulation.
Chip Profitability Calculator
Estimate profit margins, break-even volumes, and manufacturing economics with dynamic ASP modeling, volume-discount curves, and lifecycle revenue forecasting. Supports NRE amortization, mask-set cost recovery, and multi-year ramp planning for semiconductor product managers and investors.
Node Migration Calculator
Compare cost, power, performance, and area (PPAC) trade-offs when migrating between fabrication nodes from 28nm to 2nm and beyond. Incorporate EUV lithography cost premiums, design-rule complexity, and backside power delivery impact for data-driven node-selection decisions.
Fab ROI Calculator
Estimate semiconductor fab investment returns with dynamic capacity planning, operating cost modeling, and long-term profitability analysis. Supports greenfield vs. brownfield scenarios, government incentive modeling (CHIPS Act, EU Chips Act), and multi-decade depreciation schedules.
Often Used Together
Complementary tools for complete analysis
Related Articles
Dive deeper with our expert guides and tutorials related to Wafer Demand Forecast
wafers = units ÷ (dies per wafer × yield) · forecast per node · Last reviewed: 2026-06