GPC Market Outlook 2026: The Oversupply Case, Stated Fairly, and Why We Still Don't Buy It

Published 8 September 2026 · Ferro-Vanadium.com Editorial Desk · 9 min read · Analysis, not a price forecast

How a tonne of GPC actually gets made

Most market commentary on carbon additives treats the product as if it arrives from a factory that buys a raw material. The chain is longer and each link has its own constraint, which is why supply responds slowly.

Refinerydelayed coker Greenpetroleum coke Calciner~1,200–1,350 C CPC~USD 500/t Graphitizationabove 2,800 C2,500–3,600 kWh/t Screening byparticle size Packing25 kg / 1,000 kg Container25–28 jumbo bags The graphitization step is the one that is slow to build and expensive to idle. Everything upstream is a refinery decision; everything downstream is a sizing decision.

Two consequences follow from that diagram. First, the upstream constraint is a refinery stream - green petroleum coke of suitable sulfur - and refineries do not build cokers because the carbon market wants them to. Second, the capital sits in the middle, at the furnace. A calciner is a commodity asset; a graphitization furnace running above 2,800 °C under inert atmosphere is not something you order and commission in a quarter.

The bear case, stated as well as we can state it

If we were arguing the other side, here is the case we would make, and it is not a weak one.

• GPC is a process upgrade on a commodity, not a distinct product. The input - calcined petroleum coke - was assessed around USD 500 per tonne in China for Q2 2026. The output sells for two to three times that. A multiple that wide invites entrants, and entrants are exactly what compresses it.
• The demand side is cyclical and not growing fast. GPC goes into steel ladles and foundry melts, both of which track industrial production. Published value growth of around 5 percent a year is respectable but not the profile of a market that can absorb a wave of new furnaces.
• Published price ranges are wide enough to hide discounting. When a grade table shows USD 850 to 1,350 for the same nominal specification, some of that width is real product difference and some of it is a seller meeting a buyer. Wide published bands are often a symptom of a market weakening underneath.
• There is no mechanism to enforce discipline. No exchange, no settlement, no reference contract. Producers facing a quiet quarter can cut price quietly, and the market finds out months later.

That is a coherent argument. We have seen it work in other ferroalloy and carbon markets, and the absence of a settlement price is a genuine structural weakness, not a technicality.

Why we still don't accept it

Because the cash floor is higher than the bear case needs it to be

We built the cost stack on our price page, so we will only take the conclusion here. Adding published CPC input to published electricity consumption - 2,500 to 3,600 kWh per tonne at the NDRC 36-city industrial tariff of RMB 0.620 per kWh for July 2026, converted at USD 1 = CNY 6.71098 - gives roughly USD 771 to USD 834 per tonne on the commodity route and USD 1,113 to USD 1,206 per tonne on the low-sulfur route, before any capital recovery or margin.

Now compare that to where the quoted low-sulfur band sits: USD 1,050 to 1,350. The bottom of the quoted band is below our estimated cash cost. That is the single most useful fact in this market, because it tells you the bottom of the band is not a sustainable resting place - it is a level producers pass through on their way to cutting output, not a level they settle at.

Because the two published growth estimates describe different baskets

One research house puts graphitized petroleum coke at roughly USD 2.50 billion for 2025. Another puts graphite petroleum coke at USD 4.18 billion for 2025 rising to USD 5.99 billion by 2032 - which works out to a compound annual growth rate of about 5.3 percent over seven years.

Both cannot be right about the same basket. At published price levels of roughly USD 850 to 1,050 per tonne, a USD 2.50 billion market implies about 2.4 to 2.9 million tonnes, and a USD 4.18 billion market implies about 4.0 to 4.9 million tonnes. That is a gap of well over a million tonnes between two reports nominally about the same product, which tells you the category boundary is doing more work than the underlying data.

And it points at something we think is true and rarely said: part of the reported value growth is grade mix, not volume. When buyers trade up from 98.5 percent to 99.2 percent fixed carbon, the market's value rises without a single additional tonne being consumed. Value growth in this market is partly a substitution story dressed as a demand story.

Because the constraint is upstream, in a refinery

This is the argument we find most persuasive. The published supply-side view holds that no structural oversupply is expected, because graphitization furnaces are capital-heavy and low-sulfur feedstock is constrained. We agree, and would add a corollary: the feedstock constraint is not one the GPC industry can fix. A refinery builds a coker to process residue, not to serve carbon additive buyers. If the pool of suitable green coke does not grow, no amount of furnace investment downstream changes the effective ceiling.

Three things that would change our read

We would rather be falsifiable than confident. These are the specific, observable events that would move us, in rough order of how closely we are watching them.

• Low-sulfur CPC falling below roughly USD 650 per tonne and staying there for a full quarter. Published 2026 guides put medium-sulfur CPC at USD 750 to 900, so a sustained break below USD 650 would move the cash floor itself rather than producer margin. That is the one development that would genuinely open room for a lower GPC band. Watch the spread between generic and low-sulfur CPC, not the level of either.
• Evidence of furnaces idling rather than discounting. Producers cutting output is the normal response above cash cost being breached. If instead you see quoted bands widening at the bottom while shipments hold up, the discipline argument is weaker than we think. Ring your suppliers and ask about utilisation, not about price - the first question gets a more honest answer.
• A new graphitization corridor built on very cheap power. At 2,500 to 3,600 kWh per tonne, power is USD 231 to 333 per tonne in China at the July 2026 tariff. A location with materially cheaper electricity changes that line and, with it, the calculus of building. This is the slowest of the three to show up and the hardest to reverse once it does.

Where the real risk sits

Not in price. The risk we would actually budget for in 2026 is specification drift: a lower fixed carbon or higher sulfur feedstock gradually substituted into a grade that keeps its label. It does not show up on an invoice. It shows up as falling carbon recovery at the furnace, a few tenths of a percent at a time, and by the time it is visible in your melt chemistry you have already paid for several containers.

The defence is boring and effective: batch-level certificates of analysis, retained samples, and a periodic independent check on the two parameters that carry the price - fixed carbon and sulfur. If your supplier cannot produce a per-batch COA with a nitrogen figure on it, that is information.

Published specifications for good material cap nitrogen at or below 0.03 percent, hydrogen at or below 0.01 percent and oxygen at or below 0.30 percent, alongside fixed carbon at or above 99.0 percent and sulfur at or below 0.05 percent. Buyers who only ever compare fixed carbon and sulfur are ignoring the element that most often decides whether a substitution works - a point we work through in numbers on the applications page.