Tearing apart the Simandou supply overhang

Simandou.

Bloomberg has a piece today about iron ore's resilience near $105 a tonne. The narrative is "Chinese stimulus is holding the floor." CNBC is doing the same thing it always does: explaining a price by pointing at a policy. Neither of them has mentioned the 650-kilometer railway cutting through the Guinean rainforest that just started running ore to the coast at scale.

On January 17, the Winning Youth docked at Majishan port in Zhejiang province after a 46-day run from West Africa. Cargo: 200,000 tonnes of 65% iron content ore from Simandou, the largest untapped iron ore deposit on Earth. That was the first shipment. By February, vessels were departing Guinea on roughly weekly schedules. The ramp-up is running ahead of the construction timeline that was supposed to constrain it.

The financial press framing is that Simandou is a "long-term story" — something to monitor in 2028 or 2029 when full capacity arrives. That framing is wrong in the way that matters. The supply shock doesn't begin when Simandou hits 60 million tonnes. It begins when the market reprices the path to 60 million tonnes. That repricing is a 2026 event. It may already be starting.

This isn't a Guinea story. It's a Pilbara story. The mine that just opened in West Africa was financed and structured specifically to reduce China's dependence on the two companies that built the Australian iron ore market — BHP and Rio Tinto. One of those companies, Rio Tinto, is also a joint venture partner in Simandou. They helped fund the machine that is about to reprice their own product. I've seen management teams do a lot of things. I've never seen one greenlight their own margin compression before.

Numbers. Strip the narrative.

China's port stockpiles are sitting at 177.5 million tonnes across 47 facilities — a number that nearly touched the all-time record set in March. The steelmaking demand that was supposed to absorb those tonnes is not there. China's domestic steel output hit a seven-year low in 2025 and is projected to slide another 1% in 2026. The property sector has been the explanation for two straight years, which is usually the moment to start looking for a different explanation.

The structural shift is in Chinese steel's end-use composition. Construction was 50% of China's steel consumption a decade ago. That number is falling, and it isn't falling toward zero — it's falling toward a new, lower equilibrium that manufacturing, EVs, and grid infrastructure partially replace but don't fully offset. Electric arc furnaces, which run on scrap rather than iron ore, are 12% of Chinese steel production today and headed for 18% by the early 2030s. That's not a forecast. That's already in the capital deployment data.

Meanwhile, on the supply side, the trajectory looks like this:

The math on the BHP embargo is worth dwelling on. Beijing ordered Chinese mills to stop buying two specific BHP ore grades — jimblebar fines and jingbao fines. The official framing is a negotiating tactic. That may be true. But those embargoes were issued precisely at the moment Simandou ore started clearing Chinese customs for the first time. Whether the timing is coincidence or coordination, the effect is the same: Chinese mills are being structurally conditioned to reduce Pilbara dependence while the alternative supply chain is being proven out. I've watched Beijing run that playbook with soybeans, with LNG, with semiconductor equipment. It does not tend to reverse.

The market narrative says $95 is the 2026 floor because Chinese stimulus and infrastructure spending will keep absorption strong enough to balance new supply additions. Here's the assumption buried inside that number: Simandou's ramp is slow and constrained by the permanent crushing facility, which isn't complete until the second half of 2026. The narrative is counting on an infrastructure bottleneck to give the market time to adjust.

The weak link is that the ramp isn't behaving like a constrained ramp. Shipping frequency went from one vessel every three weeks in December to weekly departures by February — a cadence that implies 8 to 10 million tonnes annualised before the permanent crushing facility is even finished. UBS satellite imagery of both the mine gate and port stockpiles at Morebaya confirms accumulation is tracking ahead of the original timeline. The infrastructure bottleneck that was supposed to buy the market 18 months of adjustment time is eroding faster than the $95 consensus assumes.

Now stack the second problem on top of the first. The ore Simandou produces grades at 65% iron. The Pilbara benchmark — the price the whole seaborne market anchors to — is 62% Fe. When 65% ore arrives at Chinese ports in volume, it doesn't just add to total supply. It displaces the 62% product in the mix, because it's chemically superior and produces better steel at lower sintering energy cost. Chinese mills running on razor-thin margins don't keep buying the 62% ore at a premium over the alternative. They don't have the margin to be sentimental about their supply chain.

The grade premium spread — the price difference between 65% Fe Brazilian ore and the 62% Fe benchmark — was $18 per tonne in August 2025 when mill margins were briefly positive. By the time rebar margins went negative in September, that spread had compressed to $12. The spread is a real-time read on whether mills can afford to pay for quality. They are currently running on compressed margins with record port inventory. They cannot afford to pay much for quality. When Simandou's 65% product arrives at scale, it doesn't command the full grade premium. It displaces the 62% price instead.

The CFTC and DCE positioning data shows managed money has been net long iron ore futures on the stimulus thesis. They are positioned for the demand side of the equation. Nobody in the paper market is short on a supply thesis that requires reading satellite imagery of a port in Guinea. The machines don't check Morebaya. They check the PMI prints and the property starts. When those numbers are good enough to hold the narrative together, the longs stay on. The physical reality at the port — inventory near record, new supply arriving weekly — keeps building underneath them.

The sequence runs like this. It doesn't require a catalyst. It just requires the current physical reality to become legible to the paper market — which it eventually always does.

The capital flow consequence is not evenly distributed. Rio Tinto is in a genuinely strange position: they are a Pilbara producer and a Simandou joint venture partner simultaneously. If iron ore reprices lower because Simandou succeeds, Rio Tinto's Pilbara unit economics compress while their Guinea investment appreciates. They have built a hedge against themselves. It's either the best corporate strategy in the mining sector or a slow-motion demonstration of what happens when a mining company grows large enough that it can no longer keep track of which direction it's betting.

Fortescue has no such hedge. 100% Pilbara, no grade premium above the 62% benchmark, no Guinea exposure. They are the pure-play downside on this trade. BHP is similar but has the scale to survive a $10 price move. Fortescue's economics get uncomfortable below $90. Their cost of production hasn't fallen nearly as fast as the headline guidance suggests — the all-in sustaining cost includes a lot of green energy capex that isn't discretionary anymore.

The honest hedge on this thesis: China's stimulus arithmetic could be bigger than the supply math. If Beijing decides to deficit-spend on infrastructure at a scale that absorbs both the current port inventory and the incoming Simandou tonnes, the price floor holds and the whole argument above is a theoretical exercise. It wouldn't be the first time Beijing chose to paper over a supply surplus with a demand injection. But they've done that twice in the last decade with iron ore already. The bins fill up faster each time. At 177.5 million tonnes with weekly deliveries arriving from West Africa, there's a limit to how many times you can print your way out of a warehouse problem.