Goldman Sachs says iron ore's marginal production cost has jumped over 20% in two years to about $95 a tonne, setting a $90-95 near-term price floor even as it still expects a Q4 slide to $88.
At a glance
- Goldman Sachs' updated cost curve puts iron ore's marginal cost of production -- the 90th percentile, grade-adjusted, all-in cost -- at about $95 a tonne, up more than 20% from roughly $75 a tonne in 2024.
- The bank now treats $90-95 a tonne as a near-term price floor, estimating more than 200 million tonnes of global supply is already unprofitable at $95 and over 350 million tonnes is cash-negative below $90.
- Freight is the single biggest driver behind the higher floor: Brazil-to-China shipping rates have doubled to about $40 a tonne, compounding higher diesel and labor costs and declining ore grades.
- The most exposed supply is low-grade Indian exports and high-cost mines in Australia, Brazil, Canada, Ukraine and West Africa; some have already announced output cuts.
What happened
Goldman Sachs told clients this week that the floor under iron ore prices has moved higher, and it isn't likely to come back down easily. A team led by analyst Paul Young updated the bank's proprietary global iron ore cost curve -- a ranking of every major mine by what it actually costs to produce a tonne of ore -- and found the marginal cost of production, measured at the 90th percentile on a grade-adjusted, all-in basis, has climbed more than 20% in two years, from roughly $75 a tonne in 2024 to about $95 a tonne now. Freight is doing much of the damage: shipping rates from Brazil to China have doubled to around $40 a tonne, layered on top of rising diesel and labor costs, mine depletion in China and India, and ore grades that keep declining as the easiest deposits run out. Goldman now treats $90-95 a tonne as a solid near-term support level, estimating that more than 200 million tonnes of global supply -- mostly low-grade Indian exports and high-cost output from Australia, Brazil, Canada, Ukraine and West Africa -- is already unprofitable at $95 a tonne, with over 350 million tonnes turning cash-negative below $90. Some of those higher-cost mines have already announced production cuts. Iron ore is trading comfortably above that floor for now: the SGX-linked 62%-grade benchmark was quoted near $97-98 a tonne on September 21, while China's Dalian Commodity Exchange futures rose 0.70% to 715.5 yuan (about $106.83) a tonne the same day as mills restocked ahead of October's National Day holiday.
The details
A cost curve is simply every producer in a market ranked from cheapest to most expensive. The 90th percentile point on that curve is not an average -- it is close to the single most expensive tonne the world still needs to dig up, ship and sell to keep supply and demand in balance. That tonne matters more than any other price point because of what happens when the market price falls below it: the producer sitting at that spot starts losing cash on every tonne shipped, and eventually stops shipping. Less supply then tightens the market and pushes price back toward that cost level. That self-correcting mechanism is why Goldman is comfortable calling $90-95 a tonne a floor rather than just an estimate -- it is the price at which the market's own economics start forcing supply out.
What makes this call notable is how much that floor has moved, and why. A 20%-plus rise in two years is a big jump for a cost structure that usually changes slowly. Freight explains a large share of it on its own: Brazil-to-China rates doubling to roughly $40 a tonne falls straight onto the delivered cost of every seaborne cargo from the world's second-largest exporter. Layered on top of that is a harder problem to reverse -- mine depletion. As easily accessible, higher-grade deposits in China and India get worked out, producers are left mining lower-grade ore, which means moving more rock to get the same amount of iron, raising the cost per tonne shipped even before freight and labor are added in.
The tension in Goldman's own note is worth sitting with rather than smoothing over. A higher cost floor is a genuinely bullish structural signal, but the bank's near-term price path points the other way -- a slide to $88 a tonne by the fourth quarter of 2026, below the low end of its own stated floor, as China's steel mills return to oversupply. That is not a contradiction so much as two different clocks running at different speeds: the cost floor is a multi-year structural story, while a quarter-to-quarter price move can still dip through it if Chinese demand weakens enough, even if high-cost producers eventually respond by cutting output once losses become unsustainable.
Goldman's longer-dated forecasts add a further wrinkle: $96 a tonne for 2027, easing back to roughly $85 a tonne in real terms by 2030. That trajectory suggests the bank does not see today's $90-95 floor as permanent either -- freight rates could ease, or new lower-cost supply could eventually enter the market, shifting the whole curve down again over time. The floor is real for now, but it is a snapshot of today's cost structure, not a guarantee about where costs sit in five years.
Why it matters
India shows up on both sides of this story. Goldman specifically names low-grade Indian iron ore exports among the supply most exposed to unprofitability at $90-95 a tonne, meaning marginal Indian exporters could face real pressure to cut volumes if prices test the lower end of that range. At the same time, Indian steelmakers -- who mine and consume most of their own ore but still reference global benchmark pricing when weighing import contracts for higher-grade material -- now have a clearer sense of where the floor sits and why, rather than just watching daily price swings without knowing what is structurally supporting them. A cost-driven floor is also a different kind of signal than a demand-driven one: it holds even if Chinese steel demand stays weak, which matters for anyone in India's metals trade trying to judge how much further prices could realistically fall.
Our read
Outlook: neutral. Goldman's updated cost curve is a genuine structural bullish signal -- a marginal cost of production up more than 20% in two years to about $95 a tonne, with over 200 million tonnes of global supply already unprofitable at that level. But the bank's own near-term price path points the other way, forecasting a slide to $88 a tonne by the fourth quarter of 2026 as China's steel sector returns to oversupply, and its 2030 long-run estimate of about $85 a tonne in real terms suggests the floor itself is not permanent. The structural support and the near-term bearish forecast pull in opposite directions from the same report.
What to watch
- Whether additional high-cost producers announce output cuts as prices approach the low end of Goldman's $90-95 floor
- China's steel-sector oversupply trend into the fourth quarter of 2026, given Goldman's own forecast of a slide to $88 a tonne
- Brazil-to-China freight rates, the single largest driver of the higher cost curve, for any reversal or further increase
- Whether iron ore prices actually trade down into the $90-95 range, which would be the first real test of whether the floor holds in practice
For information only, not investment advice.
Iron price in India
metalscost.com India reference price as of 2026-10-03.
Detailed analysis
Timeline
- 2026-09-21: Goldman Sachs, in a client note led by analyst Paul Young, sets a $90-95 a tonne cost-curve floor for iron ore and raises its 2026 full-year price forecast to $93 a tonne, while flagging a slide to $88 a tonne in the fourth quarter as China's steel sector returns to oversupply.
- 2026-10-01: China's week-long National Day holiday begins, running through October 7 -- the pre-holiday restocking pull is part of why current benchmark prices are trading above Goldman's stated floor.
Supply Drivers
Goldman estimates more than 200 million tonnes of global iron ore supply -- mostly low-grade Indian exports and high-cost output from Australia, Brazil, Canada, Ukraine and West Africa -- is already unprofitable at $95 a tonne, rising past 350 million tonnes cash-negative below $90, with some higher-cost producers already announcing production cuts.
Mining Production
Iron ore's 90th-percentile, grade-adjusted, all-in marginal cost of production has risen more than 20% in two years to about $95 a tonne, driven by Brazil-to-China freight rates doubling to roughly $40 a tonne, higher diesel and labor costs, and declining ore grades from mine depletion in China and India that require moving more rock for the same iron content.
What could lift prices
- Iron ore's marginal cost of production has risen structurally to about $95 a tonne, up from roughly $75 in 2024, giving prices a firmer near-term floor than the market has had in recent years.
- More than 200 million tonnes of global supply is already unprofitable at $95 a tonne, rising past 350 million tonnes cash-negative below $90 -- with some high-cost producers already cutting output, which would tighten supply if it continues.
- The largest driver, doubled Brazil-to-China freight rates, is a structural shipping-cost shift rather than a temporary spike, making it a less reversible source of price support.
What could weigh on prices
- Goldman's own near-term forecast still points down, not up -- a slide to $88 a tonne by the fourth quarter of 2026, below the low end of its stated $90-95 floor, as China's steel sector returns to oversupply.
- Current spot prices, near $97-98 a tonne on the SGX-linked benchmark and 715.5 yuan on the Dalian Commodity Exchange, are already trading above the floor, meaning the support level has not yet been tested.
- Goldman's own 2030 long-run benchmark of about $85 a tonne in real terms implies the cost curve itself is not fixed and could drift lower again over time.
Country impact
| Country | Impact | Reason |
|---|---|---|
| Australia | High | Australia is named among the countries whose higher-cost mines are most exposed to unprofitability under Goldman's updated cost curve, even as the country remains one of the world's dominant seaborne iron ore exporters. |
| Brazil | High | Brazil is both a major source of the high-cost supply Goldman flags and the origin point of the freight route -- Brazil to China -- whose doubled shipping rate is the single largest driver of the higher cost floor. |
| China | High | China sits on both sides of this story -- mine depletion inside the country is cited as a driver of the higher global cost curve, while Chinese steel demand is the swing factor Goldman says could push prices down through the stated floor in the fourth quarter. |
| India | Medium | Goldman specifically names low-grade Indian iron ore exports as among the supply most exposed to unprofitability at current cost levels, while Indian steelmakers separately reference the same global benchmark when pricing import contracts for higher-grade ore. |
Industry impact
| Industry | Effect | Reason |
|---|---|---|
| Mining | Neutral | A higher cost floor squeezes the industry's highest-cost producers toward unprofitability and potential shutdowns, while simultaneously supporting realized prices for lower-cost producers who sit well below the new $90-95 marginal cost level. |
| Steel Manufacturing | Negative | A structurally higher iron ore cost floor raises the underlying raw-material cost for steelmakers globally, adding to pressure on mills that are already running on thin or negative margins in markets like China. |
Who gains, who loses
- Low-cost, tier-one iron ore producers: A price floor set by the industry's most expensive producers protects the wider margin of lower-cost miners whose production costs sit well below the new $90-95 marginal cost level, even as prices stay range-bound.
- High-cost iron ore producers and low-grade Indian exporters: Goldman estimates more than 200 million tonnes of supply from Australia, Brazil, Canada, Ukraine, West Africa and low-grade Indian exports is already unprofitable at $95 a tonne, with some producers already cutting output.
Other ways this could play out
- If Chinese steel demand weakens further into the oversupply Goldman expects, prices could fall through the $90-95 floor directly, forcing more high-cost producers into cuts before the floor holds in practice rather than just on paper.
- If Brazil-to-China freight rates ease from their current doubled level, or if new lower-cost supply enters the market, the cost curve itself could shift back down, pulling the floor lower again even without a change in steel demand.
Price risks
- Prices are currently trading above the $90-95 floor, so downside risk remains if China's steel oversupply deepens the way Goldman expects heading into the fourth quarter.
- The floor itself is not fixed -- Goldman's own long-run 2030 estimate of about $85 a tonne in real terms implies the cost curve could shift lower again if freight rates ease or high-cost supply exits the market for good.
Historical comparison
- 2024 to 2026: Goldman Sachs' iron ore marginal cost of production, on a 90th-percentile grade-adjusted all-in basis, has risen more than 20% in two years, from about $75 a tonne in 2024 to roughly $95 a tonne now.
- Goldman's 2027 and 2030 forecasts: Goldman forecasts 61%-grade iron ore averaging $96 a tonne in 2027, before its long-run 2030 benchmark eases to about $85 a tonne in real terms (roughly $95-100 nominal), suggesting today's cost floor is not viewed as permanent.
Technical view
Price is mixed relative to its 20-period and 50-period moving averages, showing no clear trend alignment.
Computed from metalscost.com's own stored price history.
Australia
Brazil
China
India