How Town Planning Works | TPW-0233 — The Semiconductor Fab Capacity Test: How Ultra-Pure Water, Reliable Power, Chemicals, Wastewater, Suppliers and Workforce Decide Whether a Region Can Host a Chip Plant

A semiconductor fab can look like a single industrial project.

It is not.

A modern fabrication facility is a concentrated demand for systems that normally sit in separate planning departments:

• large and extremely reliable electrical supply;
• high-quality raw water and ultra-pure process water;
• wastewater collection, segregation, treatment and reuse;
• hazardous and high-purity chemicals;
• bulk gases;
• emergency power;
• cleanrooms and controlled environments;
• heavy construction logistics;
• specialized supplier space;
• secure transport and global logistics;
• engineering, technical and scientific labour;
• housing and public services for a growing workforce.

This matters because semiconductor policy has moved from a narrow industrial question to a regional capacity question. In 2026 the OECD published dedicated semiconductor ecosystem reports for Mexico, Costa Rica and Panama. The Panama report, released on 29 July 2026, identifies reliable utilities infrastructure, skills, suppliers and customers as core conditions for ecosystem development. The Costa Rica report goes deeper into energy, water, transport and connectivity, stressing that semiconductor manufacturing is highly sensitive to power reliability.

The U.S. CHIPS programme provides a second signal. NIST’s programme-level environmental work treats utilities, water quality, hazardous materials, waste, environmental justice and cumulative effects as material planning concerns for semiconductor fabrication. Its environmental assessment notes that semiconductor fabs can consume millions of gallons of water a day; the median ultra-pure-water use reported across 29 facilities for 2021 was 1.16 million gallons per day, and the assessment emphasises that expansions can strain regional water and power resources if capacity is not planned. Current CHIPS R&D awards in 2026 also show the industry working on PFAS substitutes, advanced catalysts, rare-earth alternatives and backup-power chemistry because the manufacturing ecosystem depends on far more than silicon wafers alone.

The search and policy signal is therefore not simply “countries want chip factories.” The harder planning question is whether a proposed host region can support one without quietly displacing other water, energy, housing and industrial needs.

The reader job for this article is:

How should planners test whether a site and its wider region can genuinely host a semiconductor fab or fab cluster—before incentives, land allocations and political commitments outrun the capacity of water, power, wastewater, chemicals, suppliers, transport and workforce systems?

This article owns that question.

It does not replace existing owners for employment land, data centres, drought capacity, transmission corridors, housing needs, hazardous industry, environmental justice, critical minerals, industrial logistics or regional economic development. Those systems remain canonical. The Semiconductor Fab Capacity Test owns the integrated pre-commitment test for a uniquely infrastructure-sensitive manufacturing use.

1. Start by identifying which semiconductor activity is proposed

“Semiconductor facility” can mean:

• wafer fabrication;
• advanced packaging;
• assembly and test;
• compound semiconductor production;
• power semiconductor manufacturing;
• R&D or pilot line;
• materials or equipment production.

These activities have different land, water, power, chemical and workforce profiles.

A planning authority should never assess a generic “chip plant” without understanding the actual stage of the value chain.

2. Front-end fabs are usually the most infrastructure-intensive

Wafer fabrication involves repeated deposition, lithography, etching, cleaning, implantation and polishing steps inside controlled cleanrooms.

That normally means high utility intensity, large cleanroom systems, ultra-pure water and complex chemical handling.

Back-end assembly or packaging may be materially different.

Use the actual process when deciding whether the region has capacity.

3. The capacity test should happen before incentives are final

Governments often compete for semiconductor investment with:

• tax incentives;
• land;
• infrastructure grants;
• training support.

If those commitments are made before utility feasibility is understood, public agencies can become locked into expensive upgrades.

The planning sequence should be: capacity first, commitment second.

4. Site area understates the real geography

The fab parcel may be large, but its functional footprint includes:

• substations;
• transmission upgrades;
• water pipelines;
• wastewater treatment;
• gas and chemical yards;
• supplier sites;
• worker housing;
• transport corridors.

Map the regional system, not only the development boundary.

5. Electricity reliability matters as much as electricity quantity

Semiconductor tools can be highly sensitive to interruptions and power-quality disturbances.

OECD’s 2026 Costa Rica analysis explicitly identifies reliable electricity as a semiconductor ecosystem requirement.

The region therefore needs to test:

• available megawatts;
• redundancy;
• voltage quality;
• outage performance;
• restoration time.

A grid that can technically supply the load may still be commercially unsuitable if interruptions are frequent.

6. Peak power should be modelled by phase

A multi-fab campus may build over many years.

Do not allocate full future grid capacity on day one without a phasing plan.

Model:

• construction load;
• first-fab load;
• later-fab load;
• supplier load;
• wastewater and water-treatment load.

Then align transmission upgrades with actual build-out gates.

7. Grid connection can become the critical path

A fab may be ready to build before a transmission upgrade is ready.

This can create pressure for temporary generation, rushed substations or expensive workarounds.

The Transmission Corridor Map remains the grid owner. The fab capacity test should require a credible energisation schedule that matches factory commissioning.

8. Backup power is not optional contingency thinking

NIST’s semiconductor work highlights backup power because sudden interruptions can damage expensive equipment or product in process.

A fab may use:

• diesel generators;
• gas generators;
• batteries;
• other resilience systems.

Those systems have land-use, air-quality, fuel-storage and noise effects.

The emergency system should be included in the original application, not discovered after approval.

9. Backup generators can create a local air-quality burden

A plant that relies on dozens of large generators for testing and emergencies may create emissions near surrounding communities.

The Airshed owner remains canonical. The fab application should provide generator count, fuel, testing schedule and expected emergency operation.

Do not assess only normal grid-powered operation.

10. Water demand should be treated as a regional allocation question

NIST’s programme assessment states that fabs consume millions of gallons of water daily and that increased water demand can strain regional surface and groundwater resources, especially in arid or drought-prone locations.

This is not a small industrial connection.

The water utility should model the project against existing customers, planned growth and drought scenarios.

11. Ultra-pure water changes the infrastructure chain

Chip manufacturing requires water far purer than drinking water.

Municipal water is therefore an input, not necessarily the final process water.

The site may need large treatment trains, storage and recycling systems.

The capacity test should distinguish:

• raw or municipal supply;
• ultra-pure-water production;
• process use;
• reclaimed loops;
• final discharge.

12. Water efficiency should be included in the baseline scenario

A capacity test should not assume either worst-practice water use or perfect future technology.

Require the applicant to state:

• expected water intensity;
• recycling rate;
• source mix;
• future reduction commitments.

Then test a conservative operating scenario.

Planning is strongest when the utility assumptions are explicit.

13. Reclaimed water can reduce drinking-water competition

Where quality requirements permit, reclaimed water can support cooling, utility systems or be further treated.

Some fabs invest heavily in internal recycling.

The Water Reuse District owner provides the broader system. The fab test asks whether reclaimed supply is physically and contractually available at the required scale.

14. Reuse claims should not be double-counted

A company may say it recycles 70 percent of water.

The planner still needs to know gross intake, net consumptive use, discharge and the number of times water circulates internally.

Percentage language can obscure actual basin demand.

Use a water balance in physical units.

15. Drought conditions should be tested explicitly

A water contract that works in an average year may fail in a drought.

The Drought Capacity Map remains the citywide owner. The fab capacity test should ask what happens when municipal restrictions or source reductions occur.

Is industrial supply firm? Does the plant curtail? Does another user lose water?

These questions belong before approval.

16. The fab should not silently outrank household reliability

Economic-development value can be substantial.

That does not mean the plant automatically receives priority over existing residents during scarcity.

Water-allocation law differs by jurisdiction, but the regional plan should make priority rules visible rather than leaving them to crisis.

17. Wastewater volume can rise with production

More water in usually means more water to treat unless recycling increases proportionately.

The sewer and treatment system needs to understand:

• quantity;
• temperature;
• pH;
• fluoride;
• metals;
• solvents;
• other process constituents.

A normal municipal sewer assumption is not enough.

18. Industrial pretreatment may be essential

Semiconductor processes can generate wastewater requiring on-site segregation and treatment before municipal discharge.

The U.S. EPA’s electrical and electronic components effluent framework identifies wastewater from etching, cleaning, degreasing, cutting and grinding.

Local systems should require compatible pretreatment rather than pushing complex industrial chemistry downstream into ordinary sewage works.

19. Separate wastewater streams can improve recovery

High-strength or chemically distinct flows may be easier to treat if kept separate.

The site plan should reserve enough utility space for segregation, treatment and future upgrades.

A compact layout that leaves no room for improved wastewater management can create long-term operational constraints.

20. Wastewater treatment plants need land too

Large fabs may require substantial treatment facilities, tanks and chemical storage.

These should be shown as real land uses, not hidden under “ancillary utilities.”

The infrastructure footprint can materially affect setbacks, traffic and future expansion.

21. PFAS is a current manufacturing-transition issue

NIST’s June 2026 CHIPS R&D award for materials discovery explicitly includes PFAS-free alternatives for semiconductor manufacturing applications.

This is a strong signal that chemistry is evolving.

Planning should avoid freezing one process assumption forever. Permits need chemical-management and wastewater systems capable of adapting as materials change.

22. Planning should regulate impact, not choose process chemistry

A local authority normally should not specify which photoresist or etchant a fab must use.

Environmental and industrial regulators set technical controls.

Planning should require lawful chemical storage, treatment capacity, emergency response and environmental compliance.

Keep institutional responsibility clear.

23. Bulk chemical yards can create a major-hazard interface

Large quantities of acids, bases, solvents and gases may be stored on site.

The Major Accident Hazard Zone owner remains canonical where thresholds are triggered.

The fab capacity test should map hazardous inventories and whether nearby sensitive development remains compatible.

24. Bulk gas supply can change freight demand

Nitrogen, oxygen, argon and specialty gases may arrive by truck or be produced on site.

On-site production can reduce deliveries but increase equipment and power needs.

The application should describe the actual supply model.

25. Chemical deliveries require safe route planning

Hazardous materials should not be routed casually through residential streets or school zones when alternatives exist.

Map designated freight routes, gate access and staging.

The Logistics Layer and Warehouse Siting Map remain the broader freight owners.

26. Gate queuing can become a public-road problem

Large industrial campuses receive workers, chemicals, equipment, contractors and waste vehicles.

Provide internal queuing and security processing.

A strategic national industry should not create daily roadside congestion because the gate layout was treated as a minor detail.

27. Cleanroom buildings have unusual mechanical demands

Air handling, filtration, humidity control and temperature stability can dominate building services.

That means rooftop and yard mechanical equipment, noise and energy use can be substantial.

The site plan should show utility yards honestly.

28. Building height may come from process equipment

Cleanroom floor-to-floor heights, exhaust stacks and gas systems can make a fab taller than a conventional warehouse.

Use the Building Height Rule to distinguish occupied height from technical projections.

Industrial zoning written for ordinary sheds may not fit advanced manufacturing.

29. Vibration can influence site choice

Some semiconductor processes are sensitive to vibration.

Nearby heavy rail, blasting or industrial activity may affect manufacturing performance.

This is not normally a public health problem; it is a compatibility and investment problem.

Applicants should evaluate site conditions before requesting public infrastructure commitments.

30. Air emissions depend on abatement systems

Processes may use fluorinated gases, solvents and other chemicals.

Environmental permits should quantify emissions and controls.

The Airshed owner remains canonical. The fab capacity test checks whether the regional air-quality context can accommodate the proposed operations and backup systems.

31. Greenhouse-gas strategy should include process gases

Electricity is only one emissions source.

Some process gases have high global-warming potential.

A climate plan should distinguish:

• purchased electricity;
• on-site fuel;
• process emissions;
• refrigerants.

The Whole-Life Carbon Budget remains the broader climate owner.

32. Renewable electricity claims should be physically credible

A fab may contract for renewable energy.

That can reduce carbon intensity while the facility still requires a highly reliable grid connection.

Do not confuse contractual renewable sourcing with independence from transmission and local grid capacity.

33. Supplier ecosystems require additional industrial land

OECD’s 2026 semiconductor reports repeatedly emphasise local suppliers.

A successful fab may attract:

• equipment service firms;
• chemical suppliers;
• gas providers;
• precision engineering;
• logistics;
• packaging and testing.

Reserve or identify supplier land if the regional strategy expects clustering.

34. The supplier cluster should not be assumed to fit inside the fab site

Some suppliers need separate facilities, hazardous inventories or fast delivery access.

Map supplier typologies and likely locations.

A fab can succeed while the surrounding industrial land market becomes unexpectedly constrained.

35. Industrial land scarcity can shift lower-value firms elsewhere

A high-profile fab may bid up land prices.

Existing manufacturers can be displaced.

The Employment Land Needs owner remains canonical. The fab capacity test should consider whether strategic industrial diversity is being reduced by one large cluster.

36. Housing demand can arrive faster than the fab

Construction workers, engineers and suppliers may enter the region before full production.

The Housing Needs Assessment should receive workforce scenarios early.

Do not wait until rents rise sharply to connect industrial policy with housing planning.

37. Construction employment and permanent employment are different

A fab campus may create a large temporary construction peak and a smaller long-term operating workforce.

Housing and transport plans should model both phases.

Permanent housing should not be sized blindly to construction peak, but temporary shortage should not be ignored.

38. Specialist labour can have a regional catchment

Engineers and technicians may commute long distances if housing or schools near the site are unsuitable.

This can create new traffic patterns.

Map workforce residence scenarios rather than assuming employees live in the nearest municipality.

39. Skills policy should precede recruitment crisis

OECD’s current semiconductor work gives major attention to workforce development.

Technical colleges, universities and training centres need time to build programmes.

Land-use planners do not design curricula, but regional growth forecasts should inform education investment.

40. Training facilities can become part of the cluster

Cleanroom training, equipment labs and applied research may need specialised buildings.

These can fit university, innovation or industrial districts.

The fab strategy should reserve space for knowledge infrastructure rather than focusing only on production floor area.

41. Schools and childcare remain separate public-service owners

A growing workforce may increase enrolment and childcare demand.

The school and childcare owners should receive the population forecast.

The fab article does not take over their allocation decisions. It supplies the growth signal they need.

42. Hospitals and emergency services should understand chemical risk

Industrial incidents are rare but planning should identify:

• specialist response needs;
• ambulance access;
• decontamination capability;
• hazardous-material teams.

The local emergency system must understand the facility before operation begins.

43. Construction logistics can be enormous

Cleanroom modules, steel, mechanical equipment and utility infrastructure can require years of construction traffic.

The Construction Logistics Plan remains the owner for HGV routes, cranes, worker travel and staging.

The fab capacity test should not ignore construction because the economic debate focuses on finished operations.

44. Oversize loads need route verification

Large transformers, process tools and mechanical systems may require special transport.

Bridge capacity, turning radii and overhead clearance can become site constraints.

A site that cannot receive its equipment is not ready.

45. Airports and ports can matter to the ecosystem

High-value equipment and time-sensitive parts may rely on air freight. Large machinery may arrive by sea.

OECD’s Panama work highlights logistics as part of semiconductor opportunity.

Regional transport advantage should be measured through actual supply-chain needs rather than generic proximity claims.

46. Customs and secure logistics can affect site performance

Semiconductor inputs can be high-value and time-sensitive.

Border and customs processes are usually national functions.

Regional strategy should still recognise whether delays undermine the claimed logistics advantage.

47. Data connectivity is critical but rarely the binding land-use issue

Fabs depend on secure digital networks, suppliers and global operations.

The Broadband Map remains the digital-infrastructure owner.

Ensure redundancy, but do not let fibre availability distract from harder water and power constraints.

48. Cybersecurity requirements can shape physical infrastructure

Secure control rooms, redundant communications and access systems may need space.

Cyber rules sit outside ordinary planning.

Physical security and building requirements should be reflected in the approved site layout where they affect urban design or access.

49. Security should avoid unnecessary fortress urbanism

A fab needs controlled access and protection of intellectual property.

Industrial campuses can still have coherent public edges, safe footpaths and landscaped setbacks.

Security need not mean kilometres of neglected fence line.

50. Expansion space should be planned honestly

A company may describe a first fab while expecting several more.

The region should know the full plausible build-out.

Water, power and roads should be tested against both initial phase and maximum reserved capacity.

51. Future expansion should have decision gates

Reserve land without granting automatic approval for every later fab.

Require updated evidence for:

• water;
• electricity;
• air quality;
• wastewater;
• housing;
• transport.

A campus can grow while the region retains control over cumulative effects.

52. Do not count the same utility capacity twice

A city may court a fab, data centre and hydrogen project using the same “available” megawatts or water allocation.

The Plan Integration Scorecard exists to prevent this contradiction.

Maintain a shared infrastructure-capacity ledger across major projects.

53. The fab pipeline should be staged like the housing pipeline

Track:

• announced;
• incentive-agreed;
• permitted;
• under construction;
• commissioned.

Not every announcement becomes a functioning plant.

Public infrastructure should respond to realistic project probability and legal commitment.

54. National industrial policy can create local opportunity cost

A government may regard semiconductor capacity as strategically important.

Local land and water still have alternative uses.

Strategic importance should improve coordination and funding, not erase the need to compare local costs.

55. Economic multipliers should be treated carefully

Large investment headlines can include equipment, buildings and long-term capital rather than local wages.

Employment claims may count indirect jobs differently.

Planning should use transparent workforce assumptions for housing and services rather than promotional multipliers.

56. Tax revenue timing matters

Incentives can delay or reduce local revenue even as service demand rises.

Municipal finance planners should model:

• infrastructure cost;
• incentive period;
• property-tax treatment;
• service demand;
• long-term revenue.

The Financial Machine owner remains canonical for wider fiscal analysis.

57. Public infrastructure contributions should be explicit

If a fab requires a new water main, substation road or treatment expansion, identify:

• who pays;
• who owns it;
• who maintains it;
• whether capacity serves other users.

Hidden public obligations create weak decisions.

58. Shared infrastructure can improve the public case

A transmission upgrade may support housing and other industry. A reclaimed-water plant may benefit the whole district.

Design shared value where practical.

But avoid claiming a project is a public benefit merely because public infrastructure had to be built for it.

59. Environmental justice should examine both exposure and resource competition

Communities may experience:

• generator emissions;
• truck traffic;
• chemical risk;
• water competition;
• housing cost pressure.

The Environmental Justice Zoning Disparity Test remains the canonical equity tool.

The fab supplies a new high-intensity land use to that analysis.

60. A high-tech label does not make a use environmentally light

Cleanrooms look clean because contamination is controlled inside.

The industrial system still uses energy, water, chemicals and waste systems.

Planning should evaluate material flows rather than aesthetic impressions of advanced manufacturing.

61. Brownfield sites can be attractive—but contamination complicates construction

Existing industrial land may offer power and infrastructure.

Remediation can add time and risk.

The Brownfield owner remains canonical. The fab test should compare brownfield benefits with the need for exceptionally reliable construction and utilities.

62. Greenfield sites can simplify layout but consume land and infrastructure

Large campuses often seek open sites with expansion room.

This can require new roads, utilities and habitat loss.

A regional alternatives analysis should compare true system cost, not only land price.

63. Flood risk can threaten business continuity

A fab may contain equipment worth billions and highly sensitive production lines.

Even shallow flooding of utilities can cause large losses.

The Flood and Coastal Hazard owners remain canonical. Site selection should include future climate, access and utility resilience.

64. Water scarcity and flood risk can occur in the same region

A site may face drought most years and intense storms occasionally.

Water planning should include both supply reliability and stormwater management.

Climate resilience is not one-directional.

65. Seismic risk can affect both production and chemical systems

Earthquake-prone regions need structural and hazardous-material controls.

The Seismic Ground Map remains the hazard owner.

The fab capacity test asks whether the site’s strategic value and resilience requirements can be met economically.

66. Climate resilience should include supply chains

A fab can remain dry while its supplier road, port or substation fails.

Map critical off-site dependencies.

Business continuity is a network property.

67. A single-source utility connection is a hidden fragility

One water main, one substation or one bridge can become a point of failure.

Assess redundancy for critical systems.

The level of redundancy should match the economic and safety consequence of interruption.

68. Redundancy should not mean wasteful duplication

Two independent feeds may be justified.

Building two entire unused systems “just in case” may not be.

Use risk analysis and shared infrastructure where possible.

69. Waste management should include hazardous streams

Solid waste can include contaminated materials, filters, chemical containers and process residues.

Disposal routes and licensed handlers should be identified.

A region should not attract a fab without understanding where its regulated waste goes.

70. Waste transport can cross jurisdictional boundaries

A host municipality may approve the facility while another community hosts disposal or treatment.

Regional governance should make these externalities visible.

Do not define the project impact only by the factory fence.

71. Circularity opportunities should be realistic

Water reuse, solvent recovery and materials recycling can reduce impacts.

Not every waste stream can become a local circular product.

Require credible contracts, technologies and quality standards before counting circularity as guaranteed mitigation.

72. Public claims should separate targets from current performance

A company may have 2030 water or carbon goals.

Those are relevant.

The planning baseline should use current proven technology plus enforceable project commitments. Aspirational corporate targets should not substitute for infrastructure capacity.

73. Corporate sustainability reporting can support monitoring

Where metrics align, public authorities can use company water, energy and waste data to reduce duplicate reporting.

But regulatory monitoring should remain independently enforceable.

Voluntary disclosure and permit compliance are different systems.

74. Supplier concentration creates another resilience question

A fab cluster may rely on a small number of specialised chemicals or equipment firms.

Industrial strategy can encourage diversification and inventory resilience.

This is broader than municipal planning, but it affects land needed for strategic suppliers.

75. Critical minerals remain an upstream owner

Semiconductor manufacturing depends on specialty materials and minerals.

TPW-0231 already owns mining host-region planning.

This fab article begins when materials reach the manufacturing ecosystem. It should not duplicate extraction, Indigenous rights or mine-water questions.

76. Data centres remain a separate high-intensity industrial owner

Both fabs and data centres can demand power and water.

They serve different processes, labour markets, hazards and supply chains.

The Data Center Zone remains canonical. The capacity ledger should compare their competing utility claims without merging the articles.

77. Hydrogen remains a separate energy-infrastructure owner

A fab may use hydrogen or other gases in processes, but that does not turn the site into a hydrogen-production hub.

If dedicated hydrogen production or large storage is proposed, use the Hydrogen Planning Map for that component.

78. Schools should not become a labour-market afterthought

High-skill workers may choose locations based partly on family services.

Regional competitiveness can therefore depend on education quality.

The semiconductor capacity test should forecast population; school planning remains with its own owner.

79. Amenity planning also remains separate

Parks, retail, healthcare and culture help regions attract talent.

This does not mean a fab should receive an “amenity package” inside its land-use approval.

Regional talent strategy and neighbourhood planning should coordinate without cannibalising the amenity owners.

80. Geography matters through system proximity

A good fab region may combine:

• reliable power;
• abundant or reusable water;
• supplier access;
• airport or port connectivity;
• skilled labour;
• compatible industrial land.

This is site-and-situation geography, not a new geography owner.

The capacity test simply makes those dependencies operational.

81. The region should define a “no-go until upgraded” category

Some sites may be strategically attractive but currently lack water or grid capacity.

Mark them as future conditional sites rather than pretending they are ready.

This allows utilities to plan upgrades without creating premature development rights.

82. Conditional readiness can attract better investment

An investor may prefer a site where authorities clearly state:

• what capacity exists;
• what upgrade is funded;
• what date it arrives;
• what permits remain.

Transparent constraints can be more valuable than vague promises of unlimited capacity.

83. Pre-application review should involve utilities directly

The Pre-Application Conference owner provides the process.

For a fab, that conference should include:

• power utility;
• water utility;
• wastewater authority;
• environmental regulator;
• transport agency;
• emergency services.

A normal planning-only meeting is insufficient.

84. Confidentiality should not erase public-interest information

Semiconductor companies may protect process details and commercial secrets.

The public still needs understandable information on water, traffic, emissions, hazards and public cost.

Create a process that protects legitimate proprietary information while disclosing land-use impacts.

85. Security-sensitive information should be handled carefully

Detailed facility layouts or chemical inventories may have security implications.

Regulators need full information; public disclosure can be appropriately bounded by law.

Transparency does not require publishing every operational detail.

86. Monitoring should follow the expansion curve

A fab’s actual water and power use may differ from forecasts as tools and processes change.

Track:

• production phase;
• gross water intake;
• reuse;
• wastewater;
• peak power;
• generator testing;
• employment;
• traffic.

The Plan Monitoring Loop should detect when the next expansion gate is approaching regional limits.

87. Resource triggers should be defined before scarcity

If water use reaches an agreed threshold, require additional reuse or a new source before the next fab phase.

If grid reserve falls below a defined margin, delay expansion until reinforcement.

Triggers convert regional capacity into enforceable sequencing.

88. Trigger values should come from utilities and regulators

Planning should not invent arbitrary water or reliability numbers.

Use authoritative system thresholds and service standards.

The planner’s role is to connect those thresholds to land-use phasing.

89. The region should plan for project cancellation

A highly visible fab can be delayed or cancelled after public roads and utilities are built.

Stage public capital where possible. Design upgrades that remain useful to other industries and communities.

Reversibility is public-risk management.

90. The region should also plan for faster-than-expected success

If the first facility performs strongly, suppliers and expansions may arrive quickly.

Reserve corridors, industrial land and utility pathways before land fragmentation makes expansion expensive.

The Reserve Map provides the broader option-value method.

91. A worked example: water-rich site with weak grid

The municipality has abundant water and an industrial parcel but only one constrained transmission feed.

Rather than promise the site immediately, it classifies it as conditional and funds a grid study. The fab proceeds only after a redundant connection is committed.

The binding constraint was electricity, not land.

92. A worked example: dry-region fab proposal

The investor proposes aggressive water recycling and reclaimed-water supply.

The utility models gross intake, drought-year availability and wastewater reuse. Expansion beyond the first phase requires the reclaimed-water plant to reach a verified capacity threshold.

Water efficiency becomes a phasing condition rather than a marketing claim.

93. A worked example: established industrial cluster

The region already has chemicals, precision engineering, university labs and strong utilities.

The new fab needs less greenfield infrastructure but could crowd out existing industrial users.

The capacity test focuses on cumulative grid, water and land competition rather than building an ecosystem from zero.

94. A worked example: politically attractive but unready site

A rural government offers cheap land and incentives.

The site lacks reliable power, specialist wastewater treatment and workforce housing.

The capacity test shows that public upgrades exceed the land-price advantage.

The region avoids a stranded commitment before signing the incentive package.

95. Site certification should test utilities before incentive negotiations harden

Regions often market large industrial parcels as “shovel ready.”

For a semiconductor fab, that phrase is too vague.

A credible certified site should document:

• firm and interruptible electrical capacity;
• connection route and substation timing;
• raw-water source and drought condition;
• industrial wastewater receiving capacity;
• telecom redundancy;
• heavy-haul and construction access;
• hazardous-material response capability.

The point is not to guarantee every future process condition.

It is to prove that the region has tested the infrastructure questions that can kill a project after political commitments have already been made.

96. Utility reservation should have expiry and milestone rules

A proposed fab can reserve hundreds of megawatts or large water volumes long before construction begins.

If projects remain speculative, they can crowd out other users on paper.

Utilities should distinguish:

• inquiry;
• studied capacity;
• contracted capacity;
• energised capacity.

Reservation rules can require deposits, milestones or expiry dates where lawful.

The planning ledger should count only capacity that has a credible delivery path.

97. The region needs a queue for mega-projects, not separate promises

A semiconductor fab may compete with:

• data centres;
• hydrogen;
• battery plants;
• advanced manufacturing;
• new housing districts.

Each project team can be told independently that capacity is “available.”

Collectively, the promises may be impossible.

A shared regional queue should show committed and prospective loads against power, water, sewer, road and labour capacity.

This is where the Plan Integration Scorecard becomes operational rather than rhetorical.

98. Air permitting can become a critical-path land-use issue

Fabs use process chemicals and abatement systems that can create regulated air emissions.

The emissions profile depends on technology and process.

Planning should not invent semiconductor air limits.

It should identify early whether the proposed site can accommodate required stacks, scrubbers, backup generators and hazardous-air permitting without placing new sensitive uses in the wrong interface.

A late discovery that the air permit requires redesign can change building layout and schedule.

99. Reclaimed-water contracts should be tested against competing demand

Reclaimed water can reduce pressure on potable supply.

But a city may also plan that same resource for:

• irrigation;
• cooling;
• environmental flow;
• other industries.

The fab’s water strategy should therefore identify actual contracted volume, quality and seasonal reliability.

“Recycled water is available” is not a capacity finding.

A physically and institutionally deliverable supply is.

100. Water-quality variability can change treatment energy and waste

Ultra-pure-water systems are designed around source-water quality.

A source that becomes more saline, turbid or chemically variable can require additional treatment and create more residual waste.

Long-term capacity analysis should therefore consider source-water quality trends as well as quantity.

This is particularly important where drought, wildfire, algal events or upstream development can alter raw-water characteristics.

101. Workforce geography should be modelled by travel time, not county boundary

A fab may recruit from a wide region.

The practical labour market depends on:

• commute time;
• shift schedules;
• transit;
• housing cost.

A map that shows thousands of qualified workers within 80 kilometres may be misleading if congestion makes the trip two hours.

Model realistic peak and shift-change travel.

Workforce access is a transport and housing question as well as a skills question.

102. Supplier clustering can create second-order land pressure

Once a major fab commits, suppliers may seek nearby sites for gases, chemicals, equipment repair, clean components and logistics.

That is economically valuable.

It also means the original land allocation should not consume every serviced industrial parcel.

Reserve expansion and supplier geography before land prices spike.

A fab ecosystem needs room around the anchor, not only room under the anchor building.

103. Public infrastructure contributions should match the source of demand

A road or utility upgrade may serve the fab and many existing users.

The cost should not automatically fall entirely on either the public or the project.

Use the jurisdiction’s established tools for:

• utility connection charges;
• development agreements;
• impact fees where lawful;
• direct project works.

The Development Exaction owner remains the proportionality framework.

The fab capacity test identifies which upgrades are actually attributable to the project.

104. Expansion should be gated by measured performance, not only calendar date

A campus may announce several fabs over a decade.

Each phase should demonstrate that previous assumptions remain true.

Before the next phase, review:

• measured water use;
• wastewater quality;
• grid reliability;
• housing absorption;
• traffic;
• supplier land;
• emergency-service capacity.

If efficiency outperforms assumptions, later phases may fit more easily.

If the first phase consumes more than forecast, the region should adjust before doubling the problem.

Capacity planning is strongest when evidence from the operating fab becomes the input to the next land-use decision.

105. The Semiconductor Fab Capacity workflow

Step 1 — Define the exact semiconductor activity.

Step 2 — Map initial and full build-out.

Step 3 — Test reliable power and redundancy.

Step 4 — Build a complete water balance.

Step 5 — Test wastewater and industrial pretreatment.

Step 6 — Map hazardous chemicals, gases and emergency systems.

Step 7 — Model construction and operational logistics.

Step 8 — Reserve supplier and utility land.

Step 9 — Forecast construction and permanent workforce separately.

Step 10 — Feed growth into housing, schools, childcare and healthcare owners.

Step 11 — test cumulative effects with other major projects.

Step 12 — sequence expansion through capacity triggers.

Step 13 — monitor actual resource use and revise.

106. A Semiconductor Fab Capacity audit

Ask:

1. Is the semiconductor value-chain activity clearly defined?
2. Is full campus build-out disclosed?
3. Is peak electricity demand known by phase?
4. Is grid reliability adequate, not merely nominal capacity?
5. Is connection timing credible?
6. Is backup power included in air/noise analysis?
7. Is gross water intake quantified?
8. Is a physical water balance provided?
9. Is ultra-pure-water production understood?
10. Is recycling stated in physical as well as percentage terms?
11. Is drought-year supply secure?
12. Is reclaimed-water availability verified?
13. Is wastewater volume and composition understood?
14. Is industrial pretreatment adequate?
15. Is sufficient land reserved for water and wastewater equipment?
16. Are PFAS and other changing chemistries handled through adaptive controls?
17. Are hazardous inventories mapped?
18. Are chemical and gas delivery routes safe?
19. Is gate queuing internalised?
20. Are cleanroom mechanical systems included in noise and energy studies?
21. Are process and backup emissions assessed?
22. Is renewable sourcing distinguished from physical grid dependence?
23. Is supplier land identified?
24. Is industrial displacement risk considered?
25. Are construction and permanent workforce forecasts separated?
26. Is housing capacity tested?
27. Are skills and training institutions prepared?
28. Are public services given realistic growth scenarios?
29. Are construction logistics and oversize routes feasible?
30. Are airport, port and customs links relevant and credible?
31. Is climate and hazard resilience tested across off-site dependencies?
32. Is utility redundancy proportionate?
33. Are regulated waste routes known?
34. Are public infrastructure costs and owners explicit?
35. Is environmental justice assessed for pollution and resource competition?
36. Are full expansion phases subject to updated capacity gates?
37. Is one utility capacity ledger shared across competing mega-projects?
38. Are incentives evaluated against service and infrastructure costs?
39. Are monitoring data tied to expansion triggers?
40. Can public infrastructure remain useful if the fab is cancelled?

107. The deepest test is whether the region is hosting a factory or committing an ecosystem

A semiconductor fab is often discussed through investment value, jobs and national strategic importance.

Those matter.

But the planner’s harder task is to make the invisible regional commitments visible before they are locked in.

A fab consumes land, but it also consumes reliability. It needs a water system that can deliver under drought, a grid that can survive interruptions, a wastewater system that can handle industrial chemistry, logistics that can move sensitive equipment, a labour market that can supply skills, and a housing system that can absorb new workers without displacing everyone else.

None of those systems should be treated as someone else’s problem after the incentive agreement is signed.

The Semiconductor Fab Capacity Test succeeds when the region can answer, with evidence rather than enthusiasm, not only “Can we fit the building?” but “Can the whole regional system support the plant at full build-out, through droughts, outages, workforce growth, chemical transitions and future expansion, without promising the same water, power, land and public capacity to several incompatible futures at once?”

Sources and further reading

• OECD, Promoting the Development of the Semiconductor Ecosystem in Panama, 29 July 2026: https://www.oecd.org/en/publications/promoting-the-development-of-the-semiconductor-ecosystem-in-panama_50f1e10d-en.html
• OECD, Promoting the Development of the Semiconductor Ecosystem in Costa Rica, 22 April 2026: https://www.oecd.org/en/publications/promoting-the-development-of-the-semiconductor-ecosystem-in-costa-rica_35fe6797-en.html
• OECD, Promoting the Development of the Semiconductor Ecosystem in Mexico, 27 February 2026: https://www.oecd.org/en/publications/promoting-the-development-of-the-semiconductor-ecosystem-in-mexico_02c81dec-en.html
• U.S. National Institute of Standards and Technology, CHIPS Program Office, Final Programmatic Environmental Assessment for Modernization and Expansion of Existing Semiconductor Fabrication Facilities, 28 June 2024: https://www.nist.gov/system/files/documents/2024/06/28/Final%20PEA%20for%20Modernization%20and%20Expansion%20of%20Semiconductor%20Fabs%206-28-2024%20-%20OGC-508C.pdf
• NIST, Department of Commerce Announces Definitive Agreement with SandboxAQ for a $500 Million CHIPS R&D Award, 17 June 2026: https://www.nist.gov/news-events/news/2026/06/department-commerce-announces-definitive-agreement-sandboxaq-500-million
• NIST, Innovations in advanced processes and systems for semiconductor manufacturing, 30 June 2026: https://www.nist.gov/publications/innovations-advanced-processes-and-systems-semiconductor-manufacturing
• U.S. Environmental Protection Agency, Electrical and Electronic Components Effluent Guidelines: https://www.epa.gov/eg/electrical-and-electronic-components-effluent-guidelines
• U.S. EPA, Key EPA Actions to Address PFAS, current 2026 actions: https://www.epa.gov/pfas/key-epa-actions-address-pfas

Continue reading: Employment land · Drought capacity · Transmission corridors · Environmental justice · Critical minerals · Full Town Planning Series Index.