How Many CZ Pullers Does a 1 GW Ingot Plant Need?

The short answer

For a plant making ingots for M10 or G12 TOPCon wafers, the count usually falls somewhere between about 25 and 55 pullers per gigawatt of annual output. The spread is wide because the number is a division: the silicon the plant must grow in a year, divided by what one puller can grow in a year. Both sides of that division change with technology choices.

That is why a puller count quoted without its assumptions is not very useful. Two plants with the same nameplate can need very different fleets, and a plant that buys too few pullers will not reach its gigawatt, while one that buys too many carries idle equipment and utilities.

How we calculate it

The method needs only four inputs, each of which a project can check for its own case.

• Silicon per watt (g/W) for the target wafer format and cell technology. ITRPV's 2026 roadmap puts TOPCon at about 1.84 g/W for G12, 1.68 g/W for M10 and 1.61 g/W for G12R
• Annual silicon demand: plant capacity in watts × g/W. For 1 GW at 1.68 g/W that is about 1,680 tonnes a year
• Output per puller per day in kilograms of ingot, averaged over the whole growth cycle, not the peak pulling speed
• Availability: the share of the year a puller is actually growing or recharging, after maintenance, crucible changes, hot-zone cleaning and stoppages
• Pullers needed = annual silicon demand ÷ (kg per day × 365 × availability)

Worked example: 1 GW of M10 TOPCon

Take 1 GW of M10 TOPCon at 1.68 g/W, so about 1,680 tonnes of silicon a year. The ITRPV figure counts all the silicon consumed per watt, including material lost after the puller in cropping and wire sawing, so treating it as the crystal the pullers must grow slightly overstates the requirement. That is a deliberately conservative simplification.

For puller output, one equipment maker told TaiyangNews in April 2025 that its 42-inch-crucible puller, with a charge of up to 1,400 kg, can deliver over 200 kg of ingot a day. That is a vendor's own claim for a large modern tool, so the example also runs two lower, illustrative cases for smaller or older hot zones. All three cases assume 90% availability, which is an assumption for illustration, not a published benchmark.

• 200 kg/day: about 66 tonnes a year per puller, so about 26 pullers for 1 GW
• 150 kg/day: about 49 tonnes a year per puller, so about 35 pullers
• 100 kg/day: about 33 tonnes a year per puller, so about 52 pullers
• The same 200 kg/day case for G12 at 1.84 g/W needs about 28 pullers

What changes the number

Every input moves with decisions a project makes early, which is why sizing belongs at the specification stage rather than after equipment is ordered.

• Hot zone and crucible size: a larger crucible and charge raise kilograms per run and per day
• Recharging: recharge Czochralski (RCz) refills the crucible between ingots without cooling the furnace, cutting idle time; ITRPV expects RCz to keep a dominant share of crystallisation
• Ingot diameter and pull speed, which together set how fast silicon leaves the melt
• Product quality windows: tight resistivity, oxygen or lifetime limits for n-type cells can slow growth or cost remelts
• Yield: crystals that lose their single-crystal structure are melted back, which spends puller time without adding output
• Ramp-up: first-year availability and yield are usually lower than steady state, so year-one output trails the nameplate
• Spares and buffer: plants often plan some margin above the calculated count to cover maintenance and ramp

A sanity check from a real plant

Published plant data rarely pairs a puller count with a capacity, but one example does. A China Daily report from June 2020 on Baotou Meike Silicon Energy described a 2.1 GW single-crystal pulling project, with four furnaces installed and 120 more forming the second part of the first phase. If all of those served the 2.1 GW line, which the report does not state, that is roughly 60 furnaces per gigawatt. That sits just above the top of this guide's range, which fits an earlier generation of pullers and wafer formats. Newer tools with larger charges are what pull the count down.

Why the puller count drives the rest of the plant

The puller count is the anchor for the rest of the ingot plant. Electrical load, argon supply, cooling water, vacuum pumping, floor area, crucible and hot-zone consumables, operators and maintenance crews all scale with the number of furnaces. Downstream, squaring, grinding and wire-saw capacity are sized to the ingot tonnage those pullers produce.

The timing matters in India. MNRE has brought ingots and wafers into the ALMM framework through List-III, effective 1 June 2028, so plants being specified now will be judged on output and quality soon after commissioning. Getting the fleet size right is one of the first decisions that determines whether a plant reaches its rated capacity.

Every plant is different

The range above is a starting point for planning conversations, not a procurement number. The right count depends on the specific puller model and hot zone, the wafer format and cell customers, the feedstock, the site's power and utilities, and the operating model. JRST Technology's Phase 1 blueprint sizes the exact line for a project, from puller count to utilities, and the project estimator at jrsttechnology.com/estimate is the place to start. For budget questions, see the ingot plant cost guide linked below.