In the third quarter of 2026, Excelsior Nickel Cobalt (ENC), a new nickel plant in the Indonesia Morowali Industrial Park in Central Sulawesi, reached half its design output within weeks. Then it ran short of water. The mine that feeds it had 7 mm of rain between the start of August and 21 October, against a ten-year average of 474 mm for those months, and the plant cut back to 30% of capacity until the wet season returns, usually in December [4] [2]. The analysts at CRU expect the other plants of its kind to make similar cuts [2], and Indonesian news coverage has warned that battery-grade nickel supply for Chinese electric-vehicle makers is at risk [1].
A plant that is losing output for lack of water looks like an obvious customer for a company that sells water. We followed that idea through the plants, their suppliers and the arithmetic. The plants can probably afford the water. What a new company cannot find is a way of supplying it that the plant owners and the water firms already working in Indonesia could not match.
Why a nickel plant drinks a river
Morowali’s newer plants extract nickel by high-pressure acid leaching: crushed ore is mixed with sulphuric acid and water and heated under pressure until the metals dissolve, and a nickel-cobalt product called mixed hydroxide precipitate is then recovered from the solution and sold to battery-material makers. CRU estimates that the process uses about 200 cubic metres of water for every tonne of nickel, at least twice as much as the older furnace smelters, which mainly need water for cooling and dust control. CRU stresses that this is an estimate across plants, not a measured figure for any one of them [2].
At least five such plants sit inside the Morowali park: four that were operating in 2025, with a combined design capacity of about 165,000 tonnes of nickel a year [12], plus ENC, designed for more than 72,000 tonnes [5]. At CRU’s average intensity, plants of that size would need roughly 130,000 cubic metres a day running flat out (our arithmetic, in the notes). They share the park’s infrastructure, which is why CRU expects one plant’s shortage to be its neighbours’ too [2]. Nickel Industries, which developed ENC and holds 46% of it [5], reports that the dry weather is now also cutting output at its furnace smelters in the park, to under half of capacity in October [4] [3].
The money at stake is real. ENC’s cut of 20 percentage points on a 72,000-tonne design is a production rate about 14,400 tonnes of nickel a year lower, which at CRU’s intensity corresponds to roughly 7,900 cubic metres of water a day. That figure shows the scale, not a measured shortfall. Huayue Nickel Cobalt (HNC), an older acid-leach plant in the park in which Nickel Industries holds 10%, earned adjusted EBITDA (earnings before interest, tax, depreciation and amortisation) of US$4,416 per tonne of nickel in the same quarter [4]. Divided by 200 cubic metres, that puts each lost cubic metre of water at roughly US$22, or about Rp400,000 at the exchange rate Liputan6 used that week [1]. That is our illustration: ENC is not HNC, and the plant’s real water balance is not public. Even so, it suggests that a plant facing a cut would pay many times what water usually costs to make.
The suppliers a plant can already call
A plant that wants more water has three options. It can wait for rain, which costs output but nothing else. It can have its own engineers and the contractors who built it add treatment or desalination capacity, owned by the plant. Or it can buy water as a service from a specialist. Several specialists already sell each form of that service in Indonesia:
| Supplier | What it sells | What we could not find |
|---|---|---|
| PT Padi Hijau Buana (PHB) | Rental of modular treatment units for fresh, waste and sea water; water sold under build-own-operate contracts [6] [7] | Any published tariff or Morowali stock |
| Veolia Water Technologies Asia | Mobile treatment units for emergencies through to multi-year service, including reuse [8] | Its fleet in Indonesia; local prices |
| PT Mizui Osmosa Teknovisa | Treatment plants built for the Morowali park, with local spare stock and operator training [9] | Water-sale terms; verified customer list |
| PT Krakatau Tirta Industri, Adika Tirta Daya | Industrial water built, owned and operated for the client, including a smelter contract on Sumbawa (KTI) and portable desalination (Adika) [10] [11] | A Morowali contract |
None of them publishes a price, so we cannot say whether they are cheap or expensive. What the list does show is that the obvious services (short-term rental, mobile desalination, outsourced water plants and local spares) are already on offer to these buyers. A new company would need something those firms lack.
Every plant runs dry in the same weeks
In most rental businesses, that something is utilisation. A company that owns equipment and moves it from customer to customer keeps it working more of the year than any single customer could, so it can charge less than the cost of owning. The idea the commissioned research tested was a shared or relocatable water supply for the Morowali plants, earning its edge exactly that way.
The drought takes that edge away. The plants in the park draw on the same sources and the same weather, so they all need extra water in the same months, and none of them needs it once the rains come back [2] [4]. Moving the equipment to the other nickel centres does not help much either. CRU reported in early October that Weda Bay and Obi Island in North Maluku had also had virtually no rain since August [2]. Nickel Industries reported later that month that its own furnace smelter on Halmahera was unaffected [4]. Either way, a second site gives the equipment work only while it is short of water too, and that is in the same season. Once the rains return there is nowhere for it to go.
The shortage is also not an annual event. Nickel Industries calls the 2026 conditions unusually severe, and the ten-year average for August to October, 474 mm, is far from a drought [4]. Equipment bought for this season may wait years for the next one. The one field survey of idle seawater plants in Indonesia that we found is not encouraging: a government-built plant on Belakang Padang island was out of service when researchers visited in 2023, mainly because repairs moved slowly through the bureaucracy [15].
A desalination unit that works four months in a bad year
The economics come down to one relationship: most of what a cubic metre of desalinated water costs is the machine, and the machine costs the same whether it runs all year or for one dry season.
The best public Indonesian costing we found is a 2016 engineering study of a 1,000-cubic-metre-a-day seawater desalination plant in Bali [13]. At 90% availability through the year, it put the full cost at Rp20,829 per cubic metre in the first year, including capital recovery, and modelled a selling price of Rp24,300. A 2026 equipment-only budget for a plant of the same size, excluding intake, tanks, civil works and freight, came to US$644,000 [14]. If the Bali plant ran only about 120 days, through one dry season, its fixed costs (capital, maintenance and staff) would fall on less than half as much water. On the study’s own figures the full cost would rise to about Rp33,000 per cubic metre, above its selling price.
What a cubic metre of desalinated water costs, by how long the plant runs
A water-short plant may still pay that price, and more. If each cubic metre restores output worth something like Rp400,000, even Rp50,000 a cubic metre for a dry season is a bargain. So the problem is not that no one would pay. The problem is that the same arithmetic tells the plant owner to buy the unit itself, and nothing gives a new company a cheaper way of providing it. A new supplier pays the same for the machine and the same for idle months as the owner does. It cannot spread that cost across customers who all need water at once. And PHB and Veolia already rent mobile units to plants that would rather not own one [6] [8]. If the need runs all year instead, the comparison gets harder still. On the commissioned research’s assumptions (installed cost Rp20 billion, a water price of Rp25,000 and an operating cost of Rp10,000 per cubic metre), a plant that owns the unit pays it off after about 1,333 operating days, roughly four years of steady use. That kind of long contract is exactly what PHB, KTI and Adika already sell.
The research had sized the first market at Rp6.6 billion to Rp16.5 billion a year for a new supplier in the Morowali park. That assumed 330 days of useful work. On the same volumes and prices, a dry-season-only business would earn about Rp2.4 billion to Rp6 billion, and only in drought years. Revenue on that scale, earned only in drought years, would struggle to carry a specialist company’s engineering and sales team as well as its idle equipment.
The best argument for trying anyway
The strongest case for a new company is a different job: recovering water from the plants’ own waste streams under a long contract, which would cut what they draw from rivers in every season, not only in droughts. That would give a supplier year-round work. But reuse is already a core offer of Veolia, PHB and Adika [8] [6] [11]. A permanent duty is also exactly the kind of capacity a plant prefers to own. And we found no public figure for how much water could be recovered at Morowali, or any sign that an outsider could get access to those streams. It is a question worth asking, not yet a business.
What would change the answer
What is documented: the shortage, the output cuts, the high water intensity of the process and a market of established water suppliers. What we infer: that every plant in the park is short at the same time and that, for a few dry months, water is worth far more to these plants than it costs to make. What we do not know, and what would decide it, is whether any plant has a year-round extra water need it would rather buy than build, and at what price.
The cheapest test is a set of conversations, not equipment. Ask two or three utility engineers at the park’s acid-leach plants for their dry-season and wet-season water deficits and what owning reuse or desalination capacity would cost them. Then compare that with one build-own-operate quotation for the same duty. The idea reopens only if a year-round deficit exists and an outside supplier could offer it for less than the plant would pay to own the capacity. A run of dry seasons as severe as 2026’s would also change the arithmetic for standby capacity. If you run water at a Morowali plant, or have priced a contract there, that is the evidence we are missing.
Research Notes
The sources, calculations and evidence records behind this investigation. Numbered citations in the article link here; ↩ returns to the text.
Sources and provenance
[1] Liputan6: “Pembuat Robot Humanoid Ini Jadi Miliarder Termuda…”, 9 November 2026 ↩
Full text read. The headline concerns a Chinese robot maker; an appended section reports, citing CRU via the South China Morning Post, the drought-driven HPAL cuts and the risk to Chinese battery supply. Used for that report and for the exchange rate the article assumes (Rp18,069 per US dollar). Its “370,000 tonnes in the second half” misstates CRU, which forecasts about 370,000 tonnes for the whole year [2]; we do not use it.
[2] CRU: “Water shortages limit Indonesian HPAL nickel production”, 2 October 2026 ↩
Full text read. Supports: Morowali August rainfall falling from 222 mm to below 4 mm (citing Nickel Industries); ENC cutting from 50% to 30% until the December wet season; Weda Bay and Obi Island also nearly rainless since August; water intensity of about 200 m³/t for HPAL and about 70 m³/t for furnace smelting, as range estimates; the expectation that other HPAL plants, often sharing infrastructure, will cut similarly; and a forecast fall in Indonesian MHP output from about 208 kt to 180 kt in the second half of 2026 if they do.
[3] Nickel Industries: operations ↩
Company page, sections on HNC, ENC and furnace smelters (read by the commissioned research). Supports the company’s ownership of plants in the park.
[4] Nickel Industries: September 2026 Quarterly Activities Report, 22 October 2026 ↩
Read directly for this article (pages 1–2 and 6–7). Supports: ENC reaching 50% of nameplate before being constrained by water; 7 mm of rain at the Hengjaya Mine from the start of August to 21 October against a ten-year August–October average of 474 mm; full ramp-up delayed until the wet season, “usually in December”; RKEF output in the park cut to 44% (HNI and RNI) and 48% (ONI) of nameplate in October, while ANI on Halmahera was unaffected (128%); HNC adjusted EBITDA of US$4,416/t Ni, 45% lower than the previous quarter; ENC cash cost US$13,943/t Ni and no ENC sales in the quarter.
[5] Nickel Industries: “ENC HPAL Project Receives IUI”, 29 October 2026 ↩
Read directly. Supports ENC’s expected output of more than 72,000 t of nickel a year and the industrial licence permitting commercial sales.
[6] PHB: industrial water treatment and rental ↩
Service page and price FAQ, read by the commissioned research. Supports modular rental for fresh, waste and sea water and reuse; no tariff is disclosed.
[7] PHB: sales of water, BOT and BOO ↩
Read by the commissioned research. Supports water sold under build-operate-transfer and build-own-operate contracts; no tariff.
[8] Veolia Water Technologies Asia: mobile water services ↩
Service sections read by the commissioned research. Supports emergency-to-long-term mobile treatment and reuse offers; regional marketing, not a Morowali service-level promise.
[9] Mizui Osmosa: Morowali Industrial Park WTP contractor ↩
Read by the commissioned research. Supplier claims of Morowali-focused treatment engineering, consignment spares and operator training; not independently verified project performance.
[10] Krakatau Tirta Industri: industrial water ↩
Offering and project portfolio read by the commissioned research. Supports outsourced, operated industrial water including a named smelter deployment on Sumbawa.
[11] Adika Tirta Daya: Build Own Operate ↩
Read by the commissioned research. Supports build-own-operate industrial water, recycling and portable desalination with guaranteed quantity and quality; no tariff.
[12] Earthworks: Filtered Tailings in Indonesia, 2026 ↩
Read directly (pages 36–43, Table 1a). Lists seven HPAL plants operating as of June 2025, four of them in the Morowali park: HYNC 60,000 t Ni, QMB 50,000 t, ESG 30,000 t and Meiming 25,000 t (capacities vary between cited sources). It adds that two more HPAL facilities in the park came online after June 2025 without naming them, so five sites (these four plus ENC) is a conservative count. A plant inventory, not evidence of water deficits.
[13] Yoshi and Widiasa: small-scale seawater reverse-osmosis techno-economics, Reaktor 16(4), December 2016 ↩
Read directly. Bali case study for 150–1,000 m³/day plants: capital Rp12,015,000 per m³/day of capacity; 90% plant availability; 13% interest; 10-year life; first-year cost at 1,000 m³/day Rp20,829/m³ (capital 6,066; maintenance 658; groundwater tax 4,950; electricity 6,676; chemicals 1,000; cartridges 500; membranes 658; labour 321); modelled first-year selling price Rp24,300/m³. Ten years old and Bali-specific (including a local water tax), so used only to show the cost structure.
[14] GlobalUF: “How Much Does a Seawater Desalination Plant Cost?”, 20 October 2026 ↩
Read directly. A supplier’s selected 2026 quotations; an October 2026 proposal for 1,008 m³/day was US$644,000 FOB China, equipment only. The supplier says these are not a price list or the cost of an operating plant.
[15] Savitri et al.: field survey of SWRO on very small Indonesian islands, PLOS Water, November 2024 ↩
Abstract and findings read. The Belakang Padang seawater plant was not operating in March 2023, mainly because repairs were slow to approve. An analogy for the upkeep risk of rarely used plants, not evidence about industrial operators.
Competitor table
| Alternative | Type | Offering | Pricing | Model and channel | Source |
|---|---|---|---|---|---|
| Curtail until rain | Do nothing | Run at lower output | Lost contribution; no fee | Management decision | [2] [4] |
| Owner and its EPC contractors | Internal | Owned treatment, desalination or reuse, integrated with park utilities | Internal cost not public | Capital project | [3] [4] [9] |
| Equipment purchase (e.g. GlobalUF) | Indirect | Treatment package; site works by buyer | US$644,000 FOB for ~1,000 m³/day, equipment only | Direct quotation | [14] |
| PHB | Direct | Rental, EPC, O&M, water under BOT/BOO | Quotation only | Engineering sales | [6] [7] |
| Veolia Water Technologies Asia | Direct | Mobile treatment, reuse, long-term service | Quotation only | Regional direct sales | [8] |
| Mizui Osmosa | Direct | Morowali treatment plants, spares, training | Quotation only | Project and procurement sales | [9] |
| Krakatau Tirta Industri | Direct | Industrial water, desalination, O&M under BOO | Project-specific | Named industrial contracts | [10] |
| Adika Tirta Daya | Direct | BOO/BOT feedwater, recycling, portable desalination | Quotation only | Direct project sales | [11] |
A missing price is not evidence that a supplier is expensive. A missing Morowali contract is not evidence that a supplier is absent there.
Evidence ledger
Observed: the water constraint at ENC and the furnace smelters in the park; ENC’s cut and expected December recovery; the existence of rental, mobile, reuse and BOO offers in Indonesia [2] [4] [6]–[11]. Inferred: that all HPAL plants in the park are short at the same time (CRU’s expectation and shared infrastructure [2]); that water during a cut is worth far more than it costs to make (HNC EBITDA divided by an industry-average intensity). Contradictory: CRU’s report of North Maluku drought against Nickel Industries’ report that its Halmahera smelter was unaffected; furnace smelters use far less water, so both may be true. Missing: any plant’s water balance, any Morowali water tariff, the recurrence of droughts this severe, why plants curtailed instead of renting (source, permits, power or lead time).
Economics workings
Park water need (illustration). Four 2025 plants 165 kt [12] + ENC 72 kt [5] = 237 kt Ni/yr × 200 m³/t [2] = 47.4 million m³/yr ≈ 130,000 m³/day at full design output.
ENC’s cut. 20 points × 72,000 t = 14,400 t/yr rate × 200 m³/t ÷ 365 ≈ 7,890 m³/day. A withdrawal proxy, not an engineered deficit.
Value of water (illustration). HNC adjusted EBITDA US$4,416/t Ni [4] ÷ 200 m³/t = US$22/m³ × Rp18,069/US$ [1] ≈ Rp399,000/m³. EBITDA per tonne understates the margin on extra output (fixed costs are already paid) but is HNC’s, not ENC’s, and fell 45% in the quarter. Weakest input: the plant-specific intensity.
Dry-season cost (estimate on [13]’s figures). Fixed costs per m³ at 90% availability (328.5 days): capital 6,066 + maintenance 658 + labour 321 = Rp7,045. Spread over 120 days: 7,045 × 328.5 ÷ 120 = Rp19,286. Variable costs: 4,950 + 6,676 + 1,000 + 500 + 658 = Rp13,784 (membranes treated as variable). Total ≈ Rp33,070/m³ against Rp20,829 year-round and a modelled price of Rp24,300. In 2016 rupiah and with Bali’s groundwater tax, which may not apply in Morowali.
Dry-season revenue (illustration, research assumptions). 1,000 m³/day × 120 days × Rp25,000 = Rp3.0bn; annual capital recovery at 15% on an assumed Rp20bn installed = Rp3.0bn; operating cost at an assumed Rp10,000/m³ = Rp1.2bn; result about −Rp1.2bn. At 330 days: Rp8.25bn − 3.0bn − 3.3bn = Rp1.95bn before overhead. The price needed to break even in 120 days is about Rp35,000/m³, which a water-short plant could afford; the obstacle is that the plant can buy the same unit.
Owning breakeven. Rp20bn ÷ (Rp25,000 − Rp10,000) ÷ 1,000 m³/day ≈ 1,333 operating days: about four years at 330 days, or about 11 dry seasons of 120 days.
Market size. Base: five HPAL sites in the park [4] [5] [12]. The research’s narrowing: 40% contestable = two site-equivalents × 1,000–2,000 m³/day × 330 days × Rp20,000–25,000 = Rp13.2–33bn/yr; an entrant winning half = Rp6.6–16.5bn/yr. Dry season only (120 of 330 days): Rp2.4–6.0bn/yr, in drought years. The research’s wider sizing (eight-plant national TAM Rp566–1,510bn/yr on an assumed Rp10,000–20,000/m³; SAM 15%; year-5 SOM 10% of SAM = Rp8.5–22.7bn/yr) uses year-round water, most of which plants already supply themselves. Weakest input: the contestable share of year-round water.
Routes considered
- Shared Cluster Desalination Hub: one plant serving several sites to raise utilisation; fails because the sites are short at the same time.
- Dry-Season Bridge Capital and Deficit-Sized Dynamic Asset: relocatable or right-sized units for dry windows; already rented by PHB and Veolia [6] [8], and there is nowhere to move them after the rains.
- Brine Recovery Blending Unit: year-round water recovery; kept as the lead to reopen (above), pending a plant water balance.
- Seasonal Water Credit Hedging: needs surplus capacity elsewhere in the same season, which the drought removes.
- Pre-qualified Water Pool, Phased Water Exchange, Cluster Qualification Engine and Utility Orchestrated Bidding: procurement and vetting services for a handful of buyers who buy water projects rarely; too few transactions for a company.
- Forecast-Triggered Procurement: earlier contracting does not stop the equipment idling after the season.
- Modular Intake Conditioning Skid, Distributed Quality Gateways, Park-Networked Modular Treatment and Demand-Responsive Supply Skid: variants of modular rental already sold; they treat water but cannot create it when the intake runs short.
- Compliance Polishing Interface: addresses quality, while the documented constraint is quantity.
- Standardized Park Interface for Water Duty, Duty-Verified Escrow for Incremental Water and Pooled Discharge and Source Rights Pool: contract or access terms incumbents can offer, or rights no source shows an outsider could obtain.
Open questions
- Do any of the park’s plants have a year-round water deficit, and how large? Plant utility engineers could answer.
- What does a build-own-operate water contract in Morowali cost against owner-built capacity? PHB, KTI, Adika or Mizui could quote one.
- Why did plants cut output instead of renting water when it seems worth so much: no usable source, permits, power, or lead time?
- How often does a dry season this severe recur in Morowali? Climate and hydrology analysts could say.
Leave a Reply