Pax Silica’s water gamble in drought-prone Central Luzon

Pax Silica, envisioned as a semiconductor and artificial intelligence hub in New Clark City, Tarlac, is one of the most ambitious industrial projects in the Philippines. Its viability, however, depends heavily on water security, a critical concern for both industry and surrounding communities.

The project is expected to require 130 million liters of water per day. For its rainfall and reservoir calculations, this analysis uses an annual requirement of 39 billion liters, an amount equivalent to the irrigation needs of thousands of hectares of rice land.

The Bases Conversion and Development Authority (BCDA) has said groundwater will not be tapped and that rainwater harvesting and reservoir construction will instead be used.

Article continues after this advertisement

But an assessment of rainfall, reservoir capacity, construction costs and wastewater recycling suggests that rainwater alone would not be enough to meet Pax Silica’s projected needs, particularly during El Niño years.

Tarlac receives an average of 1,761 millimeters of rainfall annually, but precipitation is highly seasonal. Peak months from June through September receive 237 to 324 mm, while the dry months from February through April receive less than 30 mm.

During El Niño years, rainfall can decline by 40% to 60%, increasing the risk of drought.

Under normal conditions, the 1,600-hectare Pax Silica site could theoretically capture about 28 billion liters of rainwater, still 11 billion liters below the annual requirement used in this analysis.

In El Niño years, rainfall capture could decline to about 17 billion liters, leaving a deficit of 22 billion liters.

Article continues after this advertisement

Even assuming full capture efficiency, rainfall alone would not meet the project’s water needs.

Meeting demand would therefore require extensive reservoir infrastructure and high levels of wastewater recycling. Both would involve substantial financial costs and land-use trade-offs.

Article continues after this advertisement

What the rainfall can provide

Water security is a significant challenge in Philippine development.

Central Luzon, the country’s rice granary, faces increasing pressure from climate variability, aquifer depletion and industrial expansion.

Pax Silica’s projected demand of 130 million liters per day would represent a major new water requirement in the region. With BCDA saying groundwater will not be tapped, rainwater harvesting and reservoir storage become central to the proposed alternative.

Rainfall in Tarlac averages 1,761 mm annually but is highly seasonal. Peak months from June through September receive 237 to 324 mm, while the dry months from February through April receive less than 30 mm.

During El Niño years, rainfall can decline by 40% to 60%, exacerbating drought risk, according to PAGASA and the World Bank.

In a normal year, the 1,600-hectare Pax Silica site could theoretically capture about 28 billion liters of rainwater.

Based on the annual requirement of 39 billion liters used in this analysis, this would leave a shortfall of 11 billion liters.

In El Niño years, rainfall capture could fall to about 17 billion liters, widening the deficit to 22 billion liters.

The figures indicate that even at full capture efficiency, rainfall alone would not meet the annual requirement used in the analysis.

How much reservoir space would be needed

Storing 39 billion liters of water annually would require reservoirs on a scale that would occupy a significant part of the Pax Silica site.

At an average depth of 5 meters, 1 hectare can store about 50 million liters.

At that depth, about 780 hectares of reservoirs would be required to store 39 billion liters.

Increasing the average depth to 10 meters would reduce the land requirement to about 390 hectares, but engineering and safety risks would increase.

Between 390 and 780 hectares of Pax Silica’s 1,600-hectare site would therefore have to be devoted to reservoirs.

That would represent about 25% to 50% of the site, creating significant land-use trade-offs as reservoir construction competes with industrial development, agriculture and community space.

The cost of storing water

Reservoir construction costs vary according to size and design.

A small reservoir covering 50 hectares is estimated to cost P825 million to P2.75 billion.

A medium reservoir covering 200 hectares could cost P5.5 billion to P16.5 billion, while a large 780-hectare reservoir could cost P55 billion to P110 billion.

Lining reservoirs to prevent seepage would add an estimated P2.1 billion to P4.3 billion.

Total reservoir construction costs, including lining, could therefore reach P60 billion to P115 billion, or about $1.09 billion to $2.09 billion.

Annual maintenance, including desilting, pumping and treatment, could add P1 billion to P2 billion.

These figures do not include the opportunity cost of land that otherwise could be used for industrial or agricultural purposes.

When rainfall falls short

Scenario modeling highlights the vulnerability of relying primarily on rainwater.

In normal years, rainfall capture would fall short of the annual requirement used in this analysis by about 11 billion liters.

In El Niño years, the deficit could rise to 22 billion liters.

Even if reservoirs occupied as much as half of the site, the amount of water available to fill them would still depend on rainfall.

This means reservoir capacity alone would not eliminate the risk of shortages.

Wastewater recycling would therefore have to play a major role in reducing Pax Silica’s demand for new freshwater supplies.

Why water recycling matters

Experience from semiconductor manufacturing centers in Taiwan, South Korea and the United States shows that significant volumes of water can be recovered and reused.

Taiwan Semiconductor Manufacturing Co. (TSMC) has achieved water recycling rates above 88% to 90%, while Samsung and SK Hynix in South Korea recycle about 40% to 47%.

For Pax Silica’s projected demand of 130 million liters per day, advanced recycling systems capable of recovering 85% to 95% of wastewater could allow about 110 million to 124 million liters to be reused each day.

This would reduce net freshwater intake to about 6 million to 20 million liters per day.

Such reuse, however, would require advanced treatment systems to remove heat, heavy metals, fluorides and toxic organic compounds from semiconductor wastewater.

These systems include membrane bioreactors (MBR), reverse osmosis (RO), advanced oxidation processes (AOP) and, ultimately, zero liquid discharge (ZLD).

How other semiconductor hubs recycle water

Taiwan provides one of the leading examples of semiconductor water recycling.

TSMC achieved a 90.3% recycling rate in 2023, supported by large-scale water reclamation plants in Tainan and the Southern Taiwan Science Park.

Chronic droughts pushed Taiwan to prioritize semiconductor water reuse, making recycling an important element of industrial water security and national competitiveness.

In South Korea, Samsung and SK Hynix recycle about 40% to 47% of their water. Both companies are expanding in-house treatment facilities to reduce their dependence on municipal water supplies.

In the United States, Intel works with municipalities to reclaim water. Its Arizona facilities recycle about 65% of their water use and aim to reach 90% by 2028.

These approaches offer different models for Pax Silica, from highly integrated in-house recycling to partnerships with local water providers.

How much water Pax Silica could recover

If Pax Silica adopted advanced recycling systems with recovery rates of 85% to 95%, about 110 million to 124 million liters of its projected 130 million-liter daily requirement could be reused.

This would reduce the need for new freshwater to about 6 million to 20 million liters per day.

Such a reduction would significantly ease pressure on freshwater supplies, although supplemental sources would still be required, particularly during droughts.

The effectiveness of the system would depend not only on how much wastewater is recovered but also on how thoroughly it is treated before reuse.

How semiconductor wastewater is treated

Semiconductor wastewater can contain heat, fluorides, heavy metals, solvents and toxic organic compounds such as tetramethylammonium hydroxide, or TMAH.

Treatment is therefore necessary before the water can be reused.

The treatment process considered in this analysis begins with cooling and equalization tanks to reduce temperature and stabilize wastewater flow.

A membrane bioreactor removes suspended solids and organic matter, with chemical oxygen demand reductions of 92% to 97%.

Reverse osmosis then removes 90% to 95% of dissolved salts, fluorides and metals.

Advanced oxidation processes are used to break down refractory organic compounds such as TMAH.

At the highest level of treatment, zero liquid discharge uses evaporation and crystallization to recover 95% to 99% of water, leaving solid waste for disposal.

The cost of advanced treatment

Advanced recycling requires substantial capital investment and continuing operating costs.

MBR-RO hybrid systems are estimated to cost $3 million to $10 million per semiconductor fabrication plant.

Full ZLD systems are estimated to require capital expenditures of $15 million to $50 million.

Operating expenditures, driven partly by energy use and membrane replacement, are estimated at $0.50 to $2 per cubic meter treated.

For Pax Silica’s projected water demand, annual ZLD operating costs are estimated at P800 million to P2 billion.

Over 20 years, ZLD operating expenditures could total P16 billion to P40 billion, or about $291 million to $727 million.

Over 40 years, they could reach P32 billion to P80 billion, or about $582 million to $1.45 billion.

Over 50 years, the total could rise to P40 billion to P100 billion, or about $727 million to $1.82 billion.

What international experience shows

Taiwan demonstrates that water recycling rates above 90% are achievable, but only through multistage treatment, reclamation infrastructure and substantial investment.

The approach has helped increase the semiconductor industry’s resilience during periods of drought.

South Korea’s lower recycling rates of about 40% to 47% leave a larger share of semiconductor water requirements dependent on outside supplies.

Its systems may involve lower costs but provide less protection against supply disruptions.

Intel’s partnership model in Arizona offers another approach. Working with municipalities can reduce the amount of infrastructure an individual company must build, although the model also depends on external water systems.

A similar partnership could be considered for Pax Silica if local water districts have sufficient infrastructure and treatment capacity.

For Pax Silica, adopting systems capable of recycling 85% to 95% of wastewater could reduce net freshwater intake to 6 million to 20 million liters per day.

The analysis estimates capital expenditures of P825 million to P2.75 billion for the recycling infrastructure considered, with annual operating expenditures of about P800 million to P2 billion.

Without sufficient treatment, semiconductor wastewater containing fluorides, heavy metals and toxic organic compounds would not be suitable for potable use.

Where reservoirs could be built

Capas, Tarlac, where Pax Silica is located, is hydrologically connected to the wider Central Luzon water system.

Capas sits within the drainage area of the Tarlac River, a major tributary of the Agno River.

The O’Donnell River, which flows through Santa Lucia in Capas, joins the Tarlac River.

The area lies within the Agno River Basin.

The Agno system receives runoff from Mount Pinatubo and surrounding mountains, with the Tarlac River and its tributaries, including the O’Donnell River in Capas, feeding into it.

Capas also has low-lying areas influenced by lahar deposits from Mount Pinatubo.

These areas could function as catch basins but would also face risks from flooding and siltation.

The engineering challenges

Reservoir construction in the area would have to account for heavy runoff and lahar-related siltation in the Agno River Basin.

Reservoirs would require reinforced embankments as well as systems for sediment management.

They would also need to be lined with geomembrane or concrete to limit seepage.

The analysis estimates seepage losses of 20% to 30% without adequate lining.

For a 780-hectare reservoir, lining alone could cost P2.1 billion to P4.3 billion, or about $39 million to $78 million.

Deeper reservoirs could reduce the amount of land required but would introduce additional engineering and safety considerations.

Combining reservoirs and recycling

Reservoirs alone would not solve Pax Silica’s water problem because their usefulness would depend on how much rainfall is available to fill them.

Recycling alone would also leave the project needing some new freshwater.

The analysis therefore considers a system combining reservoir storage with high levels of wastewater recovery.

Even with reservoirs, Pax Silica would need recycling rates of 85% to 95% to substantially reduce net freshwater demand.

ZLD represents the highest level of water recovery considered in the analysis.

Reservoir construction, including lining, is estimated to cost P60 billion to P115 billion upfront, or about $1.09 billion to $2.09 billion.

ZLD operating expenditures are estimated at P800 million to P2 billion annually.

The long-term cost

Over 20 years, ZLD operating costs could total P16 billion to P40 billion, or about $291 million to $727 million.

Over 40 years, the total could reach P32 billion to P80 billion, or about $582 million to $1.45 billion.

Over 50 years, operating expenditures could reach P40 billion to P100 billion, or about $727 million to $1.82 billion.

When these costs are combined with reservoir construction, the total integrated cost is estimated at P76 billion, or about $1.38 billion, over 20 years under the lower-cost scenario.

Under the higher-cost scenario, the total could reach P215 billion, or about $3.91 billion, over 50 years.

Reservoir construction would account for the largest upfront expense, while ZLD would create significant recurring costs over the life of the project.

The combined cost of reservoir construction and recycling would therefore represent a substantial part of Pax Silica’s infrastructure requirements.

The policy and governance questions

The scale of Pax Silica’s water requirements raises broader questions about land, agriculture, household water supplies and regulation.

Reservoir construction could displace land that might otherwise be used for industrial development or agriculture, creating potential consequences for food production.

The estimated P55 billion to P110 billion cost of the reservoir itself would also add to the project’s other major infrastructure requirements.

The analysis identifies risks to surrounding communities if aquifers were contaminated or overdrawn, potentially affecting potable water supplies for millions of residents.

Climate variability presents another concern.

El Niño droughts could leave reservoirs underfilled precisely when stored water is most needed.

Without clear water allocation rules, industrial demand could also compete with household and agricultural requirements.

Water recycling could reduce Pax Silica’s net freshwater demand by as much as 95%, but achieving that level of reuse would require large investments in advanced treatment.

The annual cost of ZLD — estimated at P800 million to P2 billion — would have to be weighed against the potential social and economic costs of aquifer depletion and reduced potable water availability.

What Pax Silica would need to do

Under the scenarios examined in this analysis, Pax Silica would need recycling rates of at least 85% to 95% to substantially reduce its dependence on new freshwater supplies.

ZLD is the highest treatment standard considered and could provide near-total water recovery.

Such recycling would add billions of pesos to project costs but would also reduce pressure on aquifers and other potable water sources.

Transparent reporting of recycling rates and treatment protocols would be necessary, together with independent monitoring.

Partnerships with local water districts could also help reduce some costs if those systems have sufficient capacity.

Any such arrangement would still require advanced on-site treatment, including RO, AOP and ZLD where needed.

The bottom line

Rainwater harvesting at the scale required by Pax Silica is technically possible but would require substantial amounts of land and capital.

Reservoirs covering 390 to 780 hectares and costing P55 billion to P110 billion would be needed to store the annual volume used in this analysis.

Even with that capacity, El Niño droughts could leave water supplies below projected requirements.

The analysis therefore indicates that a rainwater-only approach would face significant constraints without additional water sources or substantial public investment.

Water recycling could reduce Pax Silica’s net freshwater demand by as much as 95%, but only through billion-peso investments in advanced treatment systems.

Recycling rates of 85% to 95% would significantly reduce the project’s exposure to water shortages.

ZLD operating expenditures are estimated at P800 million to P2 billion annually.

Reservoir construction with lining is estimated at P60 billion to P115 billion upfront, while ZLD operating expenditures could total P16 billion to P100 billion over 20 to 50 years.

The resulting integrated cost ranges from an estimated P76 billion, or about $1.38 billion, over 20 years to P215 billion, or about $3.91 billion, over 50 years.

Pax Silica’s proposed water security system, combining reservoir construction and ZLD recycling, would therefore involve estimated costs of $1.38 billion to $3.91 billion over the periods examined.

Taiwan’s semiconductor industry has absorbed investments on a similar scale, supported by strong state subsidies and national prioritization.

For the Philippines, the projected costs raise questions about economic viability and the trade-offs among industrial development, agriculture and potable water supplies. /dm

[Teodoro C. Mendoza, Ph.D., is a retired professor and scientist at the Institute of Crop Sciences, College of Agriculture and Food Science, University of the Philippines Los Baños.]