From mining to finished product — quartz mine · metallurgical silicon · polysilicon · wafer · closed-loop waste recovery (ZLD + STC) · ~1 GW firm power · workforce and education system. Analysed through eight expert lenses.
The project is not a single plant but a fully vertically integrated cluster: quartz is mined, purified stage by stage, and processed all the way to wafer. Every waste stream is recovered in closed loops. Uzbekistan's advantage is precisely here: Oman imports its raw material — we take ours from the ground.
The geologist's view: domestic quartz is no longer an assumption. The Tashkent Institute of Chemical Technology (Nomazov & Aripova, 2025) measured the composition of quartz from Uzbek deposits:
| Source | SiO₂ | Al₂O₃ | Fe₂O₃ | Assessment |
|---|---|---|---|---|
| Nurota vein quartz | 98.74% | 0.45% | 0.02% | High purity |
| Jerdanak quartzite | 97.72% | 0.77% | 0.12% | Illite admixture |
The SiO₂ grade is sufficient for MG-Si (metallurgical silicon). Uzbekistan has 20+ quartz deposits and beneficiation experience (JSC "Kvarts"). This is a real advantage against Oman's "import dependency" weakness.
Official strategic recognition: Uzbekistan's Critical Elements list includes silicon (Si) and high-purity silica/quartz — with vast, largely untapped potential (Mirkamalov et al., 2025). That is, the state recognises this material as strategic.
Today we have chemical composition (TKTI), but no JORC-compliant resource statement. Uzbekistan uses the GKZ (Soviet) classification, but international investors (IFC, banks) require a resource statement to the JORC or NI 43-101 standard: tonnage + grade + confidence category (measured/indicated/inferred). Only a drilling programme delivers this.
| Item | Requirement | Status |
|---|---|---|
| Candidate deposits | Jizzakh · Navoi · Sirdarya + Nurota/Jerdanak | ⚠ To select |
| SiO₂ range | ≥98% (confirmed for MG-Si: 97.7–98.7%) | ✓ Have |
| Boron (B) + phosphorus (P) ppm | Most critical for polysilicon — not yet measured | ✗ Needed |
| Post-beneficiation purity | Does it reach 6N (pilot test) | ✗ Needed |
| 30-year reserve | ~9 Mt quartz (JORC-compliant) | ✗ Drilling |
| Mining cost | $/tonne → confirm the $1.3/kg feedstock estimate | ✗ Needed |
Signal to UzSIF / Geology committee: "We hold scientifically confirmed initial evidence; we are ready to begin Gate-1 (drilling + B/P + JORC)."
For polysilicon, boron (B) and phosphorus (P) are the most important, because they corrupt the semiconductor property and are extremely hard to remove from quartz.
| Element | Target (indicative) | Method |
|---|---|---|
| Boron (B) | < 0.1 ppm | ICP-MS |
| Phosphorus (P) | < 0.3 ppm | ICP-MS |
No B/P measurement currently exists for any Uzbek deposit — this is normal. Only an ICP-MS lab + international assay + JORC expert can establish it. Final polysilicon spec is even tighter (ppba level).
After beneficiation, Al₂O₃, Fe₂O₃, TiO₂, CaO, MgO must drop to ppm levels. Pilot test stages:
| Parameter | Minimum (indicative) |
|---|---|
| Drill holes | ~12 (more possible depending on deposit size) |
| Depth | 150–200 m |
| Output | 3D geological model · tonnage + grade + confidence category |
| Duration | ~6 months · certified by a JORC expert |
"According to preliminary scientific data, the Jizzakh and Navoi regions hold promising quartz resources for a polysilicon industry. Their suitability must be confirmed through independent laboratory assay, drilling and a JORC audit."
Only confirmed high-purity quartz is presented as the feedstock base: Nurota vein quartz (98.7%) and Jerdanak quartzite (TKTI). Uzbekistan has a broad mineral base, but silica sand ≠ polysilicon quartz, and each deposit (e.g. Jeroy — a phosphorite deposit, not polysilicon) must be verified separately. Maintaining this distinction preserves credibility before the geology committee.
From quartz to polysilicon there are 5 chemical-physical stages. Each requires distinct equipment, energy and a safety regime.
| Stage | Process | Main equipment | Energy (kWh/kg) |
|---|---|---|---|
| 1 · Carbothermic | SiO₂ + 2C → Si + 2CO (1800°C) | Submerged arc furnace (SAF) | ~11–13 |
| 2 · TCS synthesis | Si + 3HCl → SiHCl₃ + H₂ | Fluidized-bed reactor | ~5 |
| 3 · Distillation | Fractional purification (remove B, P, metals) | Distillation columns | ~8 |
| 4 · Siemens CVD | SiHCl₃ + H₂ → Si (1100°C, grows on rods) | Siemens reactors (CVD) | ~45–50 |
| 5 · Crushing/packing | Crushing, quality control, packaging | Clean room, automation | ~2 |
| Total (to polysilicon) | ~70–75 | ||
The base target is solar-grade 6N (99.9999%) — for solar panels. Semiconductor-grade 11N (microchips) is an entirely different, far more complex level; only 5 companies manage it (Wacker, Hemlock, OCI, Tokuyama, Mitsubishi). It is left as a later, separate phase.
Who is the licensor/EPC? The Siemens process and STC closed loop require know-how (Western options: centrotherm and others). Power quality: a Siemens reactor cannot tolerate even a one-second interruption — this makes firm baseload power mandatory in the energy system.
This is the heart of the project's ESG / "low-carbon" certification and what the EU buyer demands. The two main waste streams of the polysilicon process — silicon tetrachloride (STC, SiCl₄) and wastewater — are fully recovered.
ZLD + closed-loop STC raises capex by ~10–15%, but it: (1) solves water scarcity, (2) earns the "green" certificate (EU premium condition), (3) lowers operating cost by reducing chlorine/feedstock purchases. Oman built exactly this system.
The wafer stage is built in Uzbekistan by a Turkish partner. Polysilicon → ingot → wafer. This is your product's adjacent, immovable buyer.
| Stage | Process | Waste recovery |
|---|---|---|
| Ingot growth | Czochralski (CZ) — monocrystal pulling | Pot remnant → re-melt |
| Slicing | Thin wafers via diamond wire saw | Kerf loss (cutting powder) → recovery |
| Wash/sort | Cleaning, quality control, packaging | Connects to the ZLD water system |
The 100kt of polysilicon splits in two and reaches the EU via wafer. This is the project's clear answer to "who buys it?"
| Buyer | Volume | Why them | Margin* (all-in $7.8/kg) |
|---|---|---|---|
| 🇹🇷 Turkey wafer (in UZ) | ~50 kt | Adjacent · non-China · Turkey–EU customs union | premium |
| 🇮🇳 India | ~50 kt | Building 20+ GW wafer, ~zero domestic polysilicon, ALMM policy | +$2/kg (thin) |
| 🇪🇺 Europe (via wafer) | — | ~63 GW/yr demand; €2.2 bn EU programme | +$12/kg |
| 🇺🇸 USA (optional) | — | UFLPA/FEOC mandate non-China | +$16/kg |
*Margin from the Energy-Cost model (separate Excel file). India is price-sensitive; US/EU premiums are wider.
The non-China premium is real but the window is narrow — Oman (100kt), Qatar and the US are filling it. The adjacent Turkey wafer plant + India offtake keep you in this race. An early LOI/MOU turns "demand" into "a buyer".
The full chain (MG-Si + polysilicon + wafer) requires ~1 GW of firm (continuous) power — ~8,800 GWh per year. ACWA Power / Masdar IPP model: they build it, you buy via a 20–25 year PPA.
| Metric | Value | Note |
|---|---|---|
| Firm power (full chain) | ~1 GW | MG-Si + poly + wafer |
| Electricity cost | ~$2.6–2.8/kg | Better than grid $4.4; above China $1.65 |
| Energy cluster capex | ~$2.5–3 bn | Owned by ACWA/Masdar (PPA) |
| CO₂ footprint | ~14 kg/kg | Within "low-carbon" range (China ~36) |
More gas = cheaper/firm, but higher CO₂ (weakens the premium). More renewables = clean, but higher capex. This balance is computed in a separate Energy-Cost Excel model — varying the gas/renewable share shows $/kg cost, firm-coverage % and CO₂ footprint in real time.
Total area ~2,800 hectares — mostly the solar farm. Industrial core ~600 ha.
Location factors: Navoi — Masdar solar experience, industrial zone, but water-scarce (ZLD mandatory). Jizzakh/Sirdarya — water/logistics may be better. Final choice assessed by proximity to the quartz mine + energy + water + rail connection.
The full cluster generates ~6,000–8,000 direct jobs (2–3× more indirect). The largest need is at the wafer plant.
| Level | Role | Field |
|---|---|---|
| Engineer / PhD | Process, R&D, quality | Chemical engineering · metallurgy · materials science · power |
| Technician | Equipment, automation, maintenance | Instrumentation · mechanical · electrical |
| Operator | Furnace, reactor, CZ, slicing | Chemical operators · clean room |
| HSE / environmental | Safety, ZLD, chlorine chemistry | Industrial safety · environment |
The most critical timing factor: training must start in 2028–29, not in 2032. If it does not begin 3–4 years ahead, there will be no qualified workforce when the plant commissions.
| Institution | Field | Role |
|---|---|---|
| Tashkent Inst. of Chemical Technology (TKTI) | Chemical engineering, quartz beneficiation | Core partner (the quartz study came from here) |
| Navoi State Mining & Tech University | Mining, metallurgy | MG-Si and mining workforce |
| Tashkent State Technical University | Power, automation, mechanics | Technicians and engineers |
| Turin Polytechnic (Tashkent) | Materials science, mechatronics | International-standard engineers |
| Foreign partner (Germany/Korea) | Siemens process, CVD know-how | Mentorship, internship, licensing |
1) Send the first wave of engineers in 2028–29 to plants in China/Korea/Germany for hands-on training (the fastest route). 2) Build an on-site training-simulator centre. 3) Training contracts with equipment vendors. 4) Redesign university programmes to fit real needs.
The first table UzSIF asks for. Each stage separately, with owner and model. Note: wafer and energy are carried by partners (Turkey, ACWA/Masdar), so Uz Silicon Tech's own capex is ~$4.5–5.5 bn.
| Component | Capex (preliminary) | Owner / model |
|---|---|---|
| 1 · Quartz mine | ~$0.1–0.2 bn | Uz Silicon Tech / state |
| 2 · MG-Si (arc furnace) | ~$0.4–0.5 bn | Uz Silicon Tech |
| 3 · TCS synthesis | ~$0.5–0.7 bn | Uz Silicon Tech |
| 4 · Distillation (purification) | ~$0.5–0.7 bn | Uz Silicon Tech |
| 5 · CVD (Siemens reactors) | ~$1.8–2.2 bn | Uz Silicon Tech (largest) |
| 6 · ZLD + closed-loop STC | ~$0.4–0.6 bn | Uz Silicon Tech |
| 7 · Land + infrastructure | ~$0.3–0.5 bn | Uz Silicon Tech / state |
| Your portion (1–7) | ~$4.5–5.5 bn | Uz Silicon Tech |
| 8 · Wafer plant (20 GW) | ~$2.0–4.0 bn | 🇹🇷 Turkey partner (JV) |
| 9 · Energy cluster (~1 GW) | ~$2.5–3.0 bn | ACWA/Masdar IPP (PPA) |
| TOTAL CLUSTER | ~$9–12.5 bn | Multi-party |
"$1.5–2 bn" (v1.0) is outdated. The fully integrated cluster is ~$9–12.5 bn. But due to the asset-light structure your balance sheet is ~$4.5–5.5 bn — the rest is off-balance via partner/PPA.
For the polysilicon core only (TCS→CVD), ~$3.6–4.2 bn. MG-Si, wafer and the energy cluster are outside this.
The world's largest non-China polysilicon plant — the proven template for our project:
Signal to investors: "Oman built 100kt for $1.6 bn, financed by the IFC + a sovereign fund. We replicate the same model, with a domestic-feedstock advantage."
Note: Oman figures are from confirmed public sources. The CAPEX percentage structure is an industry-typical estimate, not any company's internal number.
An investor's first question: "who owns it?" This is the indicative cap table of the polysilicon company (SPV). Note: energy (ACWA/Masdar) and wafer (Turkey) are separate vehicles; they are primarily contractual partners (PPA / offtake), with optional minority stakes.
Percentages indicative. Core principle: the state/sovereign fund as anchor (Oman/OIA model), a DFI (IFC) with debt+equity, a strategic industrial partner bringing technology.
Revenue model (100kt): price × volume. Cost from our Energy model — cash ~$5.8/kg, all-in ~$7.8/kg.
| Polysilicon price | Revenue/yr | EBITDA | EBIT | Verdict |
|---|---|---|---|---|
| $12/kg (blended premium) | $1,200M | $620M | $420M | Strong |
| $10/kg | $1,000M | $420M | $220M | Healthy |
| $8/kg | $800M | $220M | $20M | Break-even |
| $6/kg (China price) | $600M | $20M | −$180M | LOSS |
The all-in break-even point is ~$7.8/kg. At the Chinese commodity price ($6) the project makes a loss. So the project works only in premium markets ($10+/kg, FEOC/EU) — not in price competition with China. This is the financial proof of the entire "non-China" strategy.
Payback (your ~$5 bn portion): ~8 years at $12/kg · ~12 years at $10/kg. Acceptable for an infrastructure project.
All figures are linked to the Energy-Cost model. Final IRR/NPV is computed in the Feasibility Study with the full financing structure.
Quartz+MG-Si in-house (not in Oman) · cheap renewable potential · neutral geopolitics · state support · non-China/FEOC position.
Landlocked · gas-heavy grid (ESG tension) · no polysilicon experience/workforce · all-in cost above China · subscale risk below 100kt.
EU/US diversification · India wafer boom (needs polysilicon) · 2028–30 rebalancing · semiconductor ecosystem.
Oman/Qatar/US/Morocco · premium compression · China price war · gas deregulation · 4-party coordination.
For the Presidential Administration and the Cabinet of Ministers — national-level outcomes:
The most important view for investors: which stage when, and which "gate" stops the project if not passed. Without passing Gates 1–3, there is no FID.
FID ~end of 2028 → construction ~3–4 years → production in 2032, aligned with the rising demand/shortage cycle.
The master plan is intended for four groups — each reads a different section.
Expects exactly this kind of document. Signal: "we are ready, you start Gate-1 and be the orchestrator." Capex + Gate timeline + offtake are for them.
The geology section is in the right tone: "we have an assumption, you confirm with JORC." B/P, reserves, drilling are for them.
The energy section and ~1 GW firm requirement matter most to them. PPA model + 2 anchor buyers give them bankability.
20 GW wafer → 50 kt adjacent non-China polysilicon + Turkey–EU customs route = a golden opportunity for them.
Vertical integration (domestic quartz) is a real moat. But "who puts in equity?" is still unanswered. A sovereign fund must anchor it.
Non-China demand exists, but Oman/Qatar/US are filling it. The adjacent Turkey wafer + India offtake are a strong position.
~1 GW new power + 6–8k jobs. State orchestration and an energy guarantee are decisive.
Energy PPA + polysilicon offtake — without both it is not financed. The IFC model (Oman) exists.
"$1.5–2 bn" is outdated. All-in cost is above China — the business relies on the premium market, not price competition.
Who is the licensor/EPC? The reactor cannot tolerate interruptions → firm power mandatory. ZLD/STC raise capex.
SiO₂ confirmed for MG-Si. Boron/phosphorus measurement and reserve estimate are the rest of Gate-1.
The process is known and proven. 6N is the target; 11N a later phase. STC closed-loop + ZLD is the right approach.
Four independent mega-decisions (poly · wafer · energy · offtake) must be signed at the same time. Only a state/sovereign fund as orchestrator can break this.