How Many Wafers Make 1 GW, and How Many Wafers Come From One Ingot?

The short answer

Two counts drive every ingot-wafer plant plan: how many wafers the market needs for each gigawatt of cells, and how many wafers the plant gets from each ingot it grows. Both are simple divisions, and both move with the wafer format, the cell efficiency and the wafer thickness.

For M10 TOPCon at today's typical parameters, a gigawatt needs roughly 128 million cells, and a single multi-metre ingot gives somewhere in the region of 15,000 wafers. The sections below show the working, so a project can replace each assumption with its own.

Step 1: watts per wafer

A cell's rated power is roughly its efficiency times its area times standard illumination of 1,000 W per square metre. In February 2026 pv magazine reported an industrial-size TOPCon record cell built on an M10 wafer with an effective area of 313.3 cm². Production cells run below record efficiency, so this guide assumes 25% for illustration, which gives about 7.8 W per cell.

Divide one gigawatt by that and the answer is about 128 million cells. Some wafers break or are rejected at sorting and in the cell line, so the worked example allows 5% extra, an illustrative assumption, for about 134 million wafers a year.

Wafers per gigawatt by format

Wafer format changes the count directly. M10 is a 182 mm square wafer, agreed as a common size by seven manufacturers in June 2020. G12 is a 210 mm square wafer, and G12R is a rectangle of about 182 by 210 mm. At the same efficiency, a bigger wafer carries more watts, so a gigawatt needs fewer of them. Scaling by nominal area gives the illustrative counts below, before the loss allowance.

• M10 (182 × 182 mm): about 128 million wafers per GW
• G12R (about 182 × 210 mm): roughly 15% more area than M10, so about 110 million wafers per GW
• G12 (210 × 210 mm): roughly a third more area than M10, so about 96 million wafers per GW
• Each 1 percentage point of cell efficiency changes the count by about 4%, because watts per cell scale with efficiency

Step 2: wafers per millimetre of silicon

A diamond-wire slicer cuts a brick of squared ingot into wafers in one pass. The spacing of the wire, called the pitch, is the as-cut wafer thickness plus the kerf, the slot of silicon the wire removes. TaiyangNews, reporting the ITRPV roadmap in September 2026, puts TOPCon wafers at around 130 µm, and production diamond wire was reported at 36 to 40 µm in 2022 by InfoLink, with the kerf a little wider than the wire.

This guide therefore assumes a pitch of about 0.18 mm, which is roughly 5.5 wafers per millimetre, or about 5,500 wafers per metre of brick. In April 2025 one equipment maker told TaiyangNews its newest slicer takes bricks up to 950 mm long, a vendor claim, which at this pitch is about 5,280 wafers in a single cut.

Step 3: wafers per ingot

A Czochralski ingot is grown several metres long and then cropped and cut into bricks. The cone-shaped shoulder at the top and the tail at the bottom are cut off as crop ends, together with any sections that fail resistivity or lifetime tests. Ingot length and the share lost to cropping vary with the puller, hot zone and process, and are rarely published, so the figures below are labelled assumptions.

Worked example: a 3 m ingot that loses about 10% of its length to crop ends and test samples leaves about 2.7 m of usable brick. At 0.18 mm pitch that is roughly 15,000 wafers per ingot, before slicing breakage. A shorter ingot, heavier cropping or a thicker wafer lowers the number; a longer ingot from a recharge or continuous Czochralski process raises it.

Worked example: ingots per gigawatt

Put the two steps together and the plant's daily rhythm appears. At about 134 million M10 wafers a year and about 15,000 wafers per ingot, a 1 GW plant must grow and slice roughly 8,900 ingots a year, or about 24 to 25 a day. Those are the ingots the puller fleet has to deliver and the bricks the slicers and line must keep up with, which is why the same counts sit underneath the crystal-puller and wire-saw sizing guides linked below.

How much of the silicon ends up in the wafer

Silicon has a density of about 2.33 g per cubic centimetre, so a 182 mm M10 wafer about 130 µm thick holds roughly 10 g of silicon. At about 7.8 W per cell that is around 1.3 g of silicon per watt in the finished wafer.

The ITRPV 2026 roadmap, as reported by TaiyangNews and PV Tech, puts polysilicon use for M10 TOPCon at about 1.68 g per watt. On these illustrative figures, roughly three-quarters of the polysilicon ends up in the wafer. The rest leaves as kerf, crop ends, squaring off-cuts and broken wafers, some of which can be recycled. That gap is why thinner wafers, finer wire and longer usable ingots matter so much to a wafer plant.

What changes the count

Most of these are fixed when the wafer specification, cell technology and equipment are chosen, so the wafer count is a specification-stage decision, not an operating detail.

• Wafer format: larger wafers mean fewer wafers per gigawatt but more silicon per wafer
• Cell efficiency: higher efficiency means more watts per wafer and fewer wafers per gigawatt
• Wafer thickness and wire diameter: both set the pitch, and so the wafers per millimetre of brick
• Half-wafer slicing: cutting half-size wafers directly doubles the piece count without changing the silicon
• Ingot length and cropping: longer ingots and tighter specifications at the crown and tail lift wafers per ingot
• Yield: breakage and rejects at slicing, cleaning, sorting and in the cell line all add to the wafers a plant must make

Why it matters in India now

MNRE has brought wafers into the ALMM framework through List-III from 1 June 2028. A plant specified now has to be sized in wafers, not only in gigawatts, because its customers buy wafers by format and thickness, and the format mix is shifting: the 2026 ITRPV roadmap expects G12 and G12R to cover most of the market within ten years while the square M10 phases out.

Every plant is different

The figures above are a starting point for planning conversations, not a design basis. The right counts depend on the format, efficiency, thickness, wire, ingot length and yields a specific project agrees with its customers and equipment suppliers. JRST Technology's Phase 1 blueprint turns these into a project-specific wafer and ingot balance alongside the machine list, utilities and site, and the project estimator at jrsttechnology.com/estimate is the place to start.