Tearing apart the platinum recycling trap

Platinum.

The financial press has a story for you. The World Platinum Investment Council's Q1 2026 report showed a 268,000-ounce surplus — the first in four years. Headline writers called it a turning point. "The deficit cycle is over." "Recycling fills the gap." If you've been paying attention to what actually moves platinum, you already know what that story is missing.

The surplus is real. The story behind it is not what they think it is. South African mine supply didn't recover — it was flat. The Q1 surplus came almost entirely from a one-time surge in Chinese jewellery scrap, fabricators liquidating inventory they had stockpiled during the 2025 price trough. That's not a structural fix. That's an inventory flush. It's a one-quarter phenomenon masquerading as a market regime change.

The machine everyone is betting on to fix platinum supply is autocatalyst recycling. Spent catalytic converters, collected from scrapyards, refined at a handful of specialist facilities, metal returned to market. The recycling loop. Clean. Circular. Logical.

The problem is that the loop has a kink in it that nobody is measuring correctly, and I want to walk you through exactly where the pipe is bent.

Numbers only. Here is what the recycling machine actually looks like when you strip out the narrative.

The recycling pipeline has three mechanical stages, and each has its own delay clock.

Stage one: vehicle scrappage. A car has to be retired before its catalytic converter enters the recycling stream. Global vehicle scrappage rates are relatively stable, but the population of vehicles worth scrapping — those with high-PGM-loading converters from the 2015–2022 era — is only now reaching typical end-of-life age. That's the good news. The bad news is in stages two and three.

Stage two: collection and consolidation. Spent converters have to be collected by scrapyards, sorted, and sold to specialist processors. When PGM prices were depressed from 2022 to 2024, scrapyards ran a rational play: hoard. Sit on the inventory. Wait for prices to recover. This is legal, widespread, and not tracked in any official data series. The WPIC calls it "collector hoarding" and identifies it as a key reason why recycled supply was running 17% below the decade average even as the ELV scrappage pool was theoretically large enough to support much higher volumes.

Stage three is the one that doesn't make the WPIC charts: working capital financing. A specialist refiner that acquires a large batch of spent converters doesn't get paid when the metal ships. They get paid when the refining cycle completes — which takes six to ten weeks. At current platinum prices and refining volumes, that's a significant capital requirement. Higher platinum prices, paradoxically, make this problem worse: the metal is worth more, the working capital required to process it goes up, and smaller operators get squeezed out of the pipeline just when you need them most.

Mine supply isn't riding to the rescue either. Total mine production peaked just over 6 million ounces in 2021 and the 2026 forecast sits around 5.5 million ounces — a 9% structural decline — with the WPIC projecting output to remain "broadly flat" this year. The Mogalakwena underground expansion, Valterra's big answer to the grade decline problem, is in trial-mining phase. First real ore won't surface at scale until well past 2030.

The market thinks recycling is a dial you can turn. It is not. It is a pipeline with multiple constriction points, and only the first one — scrappage volumes — responds to price in a clean, measurable way.

The scrapyard hoarding problem is structural and invisible. Spent converter inventories sitting in yards in the US Midwest, central Europe, and South Korea are not reported anywhere. They show up in the data only when they move — as a sudden supply improvement that analysts then attribute to price-responsiveness, writing headlines about how recycling is "working." What's actually happening is the hoarding cycle unwinding. And that cycle doesn't unwind linearly. It unwinds in batches, when a yard operator decides prices are high enough to justify the paperwork, the logistics, and the working capital cost of converting stockpiled converters into cash.

Here's the thing nobody is writing about: China just reclassified platinum as a strategic critical mineral. That changes the incentive structure for Chinese holders of above-ground stocks in a very specific direction. It does not point toward selling. It points toward strategic accumulation. The West is counting on a Chinese inventory buffer that China has just signalled it intends to hold rather than release.

Meanwhile, on the Western mine supply side: South African electricity tariffs rose roughly 60% between 2021 and 2026. Diesel import costs for deep-level operations run through Hormuz. Over 20,000 mining jobs have been shed since 2024. The Western Limb — the historic heart of Bushveld platinum production — is in what industry executives are now openly calling a "managed retreat." I've been watching this happen in slow motion for three years. The first time a Sibanye shaft manager used the phrase "managed retreat" in a public context, I kept reading the sentence three times thinking I'd misread it. I hadn't.

The refining infrastructure for spent PGM autocatalysts runs through approximately five major facilities globally — in the UK, Belgium, Germany, South Africa, and Japan. None of them was built to handle a sudden step-change increase in throughput. Adding refining capacity isn't a six-month project. It's a capital investment with a multi-year lead time, at a price level that processors aren't yet convinced is durable. So they wait. The scrapyard operators wait. Everyone waits, and the WPIC writes a report saying recycling supply will grow 9% this year, which is technically accurate and structurally insufficient.

The WPIC is forecasting a 240,000-ounce deficit for 2026 as a full year — meaning the Q1 surplus gets more than wiped out over the following three quarters. That forecast assumes recycling performs as modelled. Based on what we've just walked through, that assumption deserves scrutiny.

The hydrogen angle is real but slow. WPIC projects 875,000 to 900,000 ounces of annual platinum demand from hydrogen applications by 2030. Right now it's a rounding error — under 3% of total automotive demand. But the direction is one-way, and it's entering a supply structure that's already running at a deficit without it. The demand stack is building from the bottom. By the early 2030s, hydrogen adds a layer that the recycling loop — even a fully optimised version of it — wasn't designed to absorb.

Where does capital go if this thesis holds? Not into broad precious metals ETFs — those are diluted with gold, silver, and ETF flows that trade on Fed rate expectations and have nothing to do with PGM mechanics. Not into the big producers with deep hedging programmes, who've effectively sold forward the upside to manage their cost base. The edge, if there is one, is narrow: mid-tier, low-debt PGM miners with direct Bushveld exposure and minimal forward hedging, plus the handful of specialist recycling-adjacent operators whose throughput economics improve non-linearly when both platinum prices and scrap volumes rise simultaneously.

The consensus sees a balanced 2026 and says the deficit cycle is done. The mechanics say the recycling fix is slower than modelled, the mine decline is structural and inelastic, and the Chinese buffer is being reclassified out of Western reach. When the model and the machine disagree, I've learned to look at the machine. The machine is still running short.