OperationsAdvanced8 min read

Capacity Strategy

Capacity strategy decides HOW MUCH capacity to build, WHEN to add it, and WHERE to place it across a 5-15 year horizon. The 3 dominant timing patterns: (1) Lead — build ahead of demand (cheap if demand arrives, catastrophic if it doesn't), (2) Lag — wait for demand to materialize (loses share but lowers risk), (3) Match — incremental additions tracking demand. Each unit of capacity is a sticky $-bet: a chip fab is $10-20B and depreciates over 7-10 years; an auto plant is $1-3B over 25 years. Volkswagen's Skoda capacity expansion in 2018 added 250K units before EV demand softened — half the lines now run in dual-shift instead of triple. KnowMBA POV: capacity strategy that doesn't include downside scenarios becomes structural waste.

Also known asCapacity Planning StrategyCapacity ExpansionLong-Range CapacityCapacity Investment Strategy

Challenge a friend Browse library

The Trap

The trap is single-point demand forecasts. Forecasts at the 5-year horizon are wrong by 20-50% in 60%+ of cases (PwC Capital Projects survey). Building to a single forecast number — even an 'aggressive but achievable' one — produces stranded capacity in any downturn. The second trap: confusing nameplate capacity with effective capacity. Nameplate is the brochure; effective is what you'll actually run after OEE losses (usually 65-80% of nameplate). Many companies plan as if 1.0M nameplate = 1.0M sellable; the gap shows up as missed shipments at peak demand.

What to Do

Build capacity using the 'staircase' framework: (1) Forecast demand in 3 bands (low/base/high). (2) Set capacity additions in modular increments, each a separate go/no-go decision tied to demand triggers. (3) For each increment, model NPV under all 3 demand scenarios. (4) Add only what is NPV-positive in 2 of 3 scenarios. (5) Pre-purchase land/permits but defer construction — cheap optionality vs. expensive irreversibility.

Formula

Capacity Cushion = (Effective Capacity − Average Demand) / Effective Capacity × 100. Healthy: 10-25% for stable industries; 25-40% for volatile demand.

Pro Tips

01
Pre-permit and pre-design 2x your committed capacity. Permits, EIAs, and zoning are the slowest part of capacity additions (12-36 months). Having shovel-ready options lets you ramp in 6-9 months instead of 24-36 if demand surprises up.
02
Capacity decisions should match the volatility of the END market, not the AVERAGE of the end market. Cyclical industries (semis, autos, chemicals) need more cushion because 'average' demand never actually happens — you're either above trend or below.
03
Track 'capacity utilization at design margin,' not just utilization. A plant running at 92% utilization but at 60% of design margin (because mix shifted to lower-margin SKUs) is in worse shape than a plant at 75% utilization on its premium mix.

Myth vs Reality

Myth

“Building capacity ahead of demand always wins share”

Reality

Lead-strategy works when (a) the market grows roughly as forecast and (b) competitors don't also build. When BOTH conditions break, lead-strategy creates industry-wide overcapacity that crashes prices for years (solar 2012-2015, LCD panels 2008-2012, lithium 2023-2024). Always model what happens if every competitor reads the same growth report you did.

Myth

“Adding capacity is reversible if demand misses”

Reality

Capacity is among the LEAST reversible decisions a company makes. Idle plants still incur depreciation, property tax, minimum staffing for safety, and demolition liability. Selling specialized assets recovers 20-40% of book value at best. The optionality value of NOT building is typically underweighted by 3-5x in capex models.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Knowledge Check

Skoda announced ~250K units of additional EV capacity at its Mladá Boleslav plant before 2024 EV demand growth slowed in Europe. The lines now run dual-shift instead of triple. Which capacity-strategy lesson does this best illustrate?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Capacity Cushion (Stable Industries)

Process industries (chemicals, paper, steel)

Optimal

10-20%

Conservative

20-30%

Tight (Service Risk)

5-10%

Stranded Capacity

> 35%

Source: APQC / Boston Consulting Group benchmarks

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

🔋

Tesla Gigafactory Strategy

2017-2024

mixed

Tesla pursued an aggressive lead-capacity strategy with Gigafactory Nevada (cells), Shanghai (vehicles, 2019), Berlin and Austin (2022), and announced Mexico (delayed). The strategy paid off through 2022 when EV demand exceeded supply and Tesla captured premium pricing and 30%+ gross margins. By 2024, EV demand growth in major markets slowed and global capacity (Tesla + BYD + legacy OEMs) outran demand. Tesla idled lines, cut prices to maintain volume (gross margin fell from ~30% to ~17%), and paused or rescoped Mexico. The lead strategy worked in the 2018-2022 demand surge and stranded capacity during the 2023-2024 plateau.

Gross margin peak (2022)

~30%

Gross margin 2024

~17%

Mexico Gigafactory status

Paused/rescoped

Lead capacity is high-beta: it amplifies upside in demand surges and amplifies pain in plateaus. The right metric is not 'will we run out of capacity?' — it is 'what happens to unit margin if demand grows 30% slower than the forecast?'

Source ↗

🚙

Skoda Auto

2018-2024

mixed

Skoda expanded capacity at Mladá Boleslav and Kvasiny by ~250K units to support both ICE and EV (Enyaq) volume, on a base of robust European auto demand growth assumptions. Post-2022, European EV demand growth slowed and ICE demand softened in key markets. Lines that were planned for triple-shift now run dual-shift, raising unit cost on the additional capacity by 8-12%.

Capacity added

~250K units

Shift utilization

Triple → Dual

Unit cost penalty on added capacity

+8-12%

Volume-based justification of capex is fragile when end-market growth slows. Modular, demand-triggered additions would have captured the upside while limiting downside.

Source ↗

Decision scenario

The $850M Capacity Bet

You are CFO at a specialty chemicals company. The CEO is pushing a single $850M plant decision. Base-case 5-year IRR is 6%. Downside (-15% price, -20% volume) is -3% IRR. Upside is 14%.

Capex

$850M

Base IRR

Downside IRR

-3%

Upside IRR

14%

Demand uncertainty

±30%

Decision 1

The CEO wants a single up-or-down vote on $850M. The board is split.

Approve the full $850M plant — base IRR is acceptable and the team has run the numbers carefullyReveal

When 3 competitors announce similar capacity 12 months later, the industry overbuilds. Pricing falls 18%. Your plant operates at 64% utilization. Lifetime IRR comes in at 1.5%. The $850M would have been better deployed in incremental plant upgrades or M&A.

Realized IRR: 6% expected → 1.5% actualStranded capex: ~$300M effectively idle

Counter-propose a staged build: Phase 1 ($400M, 60% capacity) with Phase 2 ($450M) gated on price + volume triggers in month 18Reveal

Correct. Phase 1 captures the most NPV-rich portion of demand. Triggers act as a forcing function for honest market re-assessment. When pricing softens in year 2, you defer Phase 2 by 18 months at minimal carrying cost. You preserve $450M of optionality and can redeploy it to higher-IRR opportunities. The board respects the discipline.

Capex committed (year 1): $850M → $400MOptionality preserved: $450M deferable / redeployableRealized IRR Phase 1: 9-11%

Related concepts