Tearing apart the tin solder supply chain

Tin.

Four weeks ago it was trading at a record high — $59,040 a tonne on June 3. Today it's sitting near $50,550, a seven-week low, down roughly 14% from the peak. Bloomberg's explanation fits in a chyron: "AI spending fears hit metals." CNBC ran the Broadcom earnings miss and drew a straight line from Hock Tan's guidance to the tin chart. If you only watched the screens, you'd think the solder market just lost its biggest customer.

It didn't.

Solder isn't a growth story you can downgrade on a single earnings call. It's a physical input. You cannot manufacture a semiconductor without it. You cannot build an AI server without it. You cannot build the data center that houses the server without it. The Broadcom miss was about networking revenue guidance — one company's forward estimate for one product line. The tin market's response was to reprice the entire physical supply chain of the metal that holds the digital economy together.

That's not analysis. That's an algorithm reading a headline.

Narrative off. Supply diagram on.

Global tin consumption ran approximately 378,900 tonnes in 2025. Refined production grew roughly 3% year-on-year. Demand grew 3.5%. That half-point gap doesn't sound like much until you remember that the tin market was already in deficit — 2,200 tonnes short in 2024, according to the International Tin Association. The buffer was gone before this year started.

Read that callout again. Silicon wafer shipments — the single best proxy for physical solder demand — hit 3,275 million square inches in Q1, the strongest year-on-year growth this cycle. That data was published April 29. Thirty-five days before tin hit its record. Then Broadcom's earnings print on June 3, and the same market that just priced in record semiconductor throughput decided the AI trade was over.

The demand side didn't change. What changed was sentiment. And sentiment doesn't unsolder a chip.

Now look at supply. Two countries — Indonesia and Myanmar — together account for roughly 40% of global tin exports. Both are impaired.

Indonesia's refined tin exports fell to 46,000 tonnes in 2024 — a multi-year low. In May 2026, the country shipped just 3,246 tonnes, down over 40% year-on-year, as licensing bottlenecks strangled what was left of the legal supply chain. PT Timah, the state miner, controls 81% of Indonesia's national tin-mining permits. The government is simultaneously cracking down on illegal operations and failing to process legal export licenses on time. The result is a supply valve that's jammed in both directions.

Myanmar's Wa State — which accounted for 70% of the country's tin exports — banned all mining in August 2023. Output fell 50% in 2024. The Man Maw mine operators secured mining permits in 2025 and began slowly resuming production in early 2026, but progress has fallen short of expectations — the ITA doesn't expect full resumption until 2027. The mine still needs dewatering. The loading docks aren't loading much.

Here's the part the AI-selloff narrative can't reach.

The tin market is small. Roughly 378,900 tonnes a year. For context, the copper market moves 25 million tonnes. Zinc does 14 million. Tin is a rounding error on most commodity desks — which is exactly why it moves so violently when physical flows shift. A few thousand tonnes of disrupted supply, in a market this size, isn't noise. It's the whole signal.

The Broadcom earnings miss triggered algorithmic selling across the entire semiconductor-adjacent complex. The SOX index dropped 10% in a single session. Tin, because it carries the "AI metal" tag now, got dragged along. I've seen this before in smaller commodity markets — the narrative tail wags the physical dog until it can't. I lost money on a nickel position in 2019 trading the story instead of the stockpile. You learn, eventually, to check the warehouse before you check the headline.

The Indonesian crackdown is the weak link nobody on the screens is watching. President Prabowo's government has shut down an estimated 1,000 illegal tin mines across Bangka-Belitung. Those operations may have accounted for up to 80% of the islands' output. The Indonesian Tin Exporters Association (AETI), an industry group, estimates 12,000 tonnes per year were leaving the country illegally. That supply didn't get replaced — it got deleted.

Meanwhile, the legal licensing pipeline is clogged. The shift from annual to triennial mining licenses — meant to streamline the process — created a regulatory bottleneck that's been running since early 2024. In a previous licensing crackdown, PT Timah estimated 3,000 tonnes of shipments were held up by licensing failures in March 2018 alone. This isn't a temporary disruption. It's an institutional restructuring of the world's largest refined tin exporter, happening in real time, with no clear timeline for resolution.

Shanghai warehouse stocks have been drawn down to about 9,286 tonnes. China flipped from net exporter to net importer of refined tin in the second half of 2024 and hasn't flipped back. The world's largest tin consumer is now competing with everyone else for shrinking physical supply. That's the quiet part. That's the part the Broadcom chyron can't see.

The sequence from here is mechanical.

Tin did this exact move in 2021–2022. Price dropped 20% on a macro scare, the physical market stayed tight underneath, and within three months it ripped back to new highs. The difference now is the supply picture is worse. Indonesia's export capacity has been structurally reduced by the crackdown, not temporarily paused by a licensing delay. Myanmar's Man Maw restart is real but slow — full resumption isn't expected until 2027. The pipeline of new mines globally is, to put it politely, thin. ITA projects a 13,000-tonne deficit by 2030 without major new investment.

The AI demand narrative will come back. It always does. Silicon wafer shipments are growing at 13% year-on-year. Data centers are still being built. The solder doesn't care about Broadcom's quarterly guidance — it cares about whether 190 GW of announced hyperscale capacity gets built or doesn't. And if it gets built, every rack needs tin.

Where capital doesn't go: diversified mining majors where tin is a rounding error on the revenue line. Where it does go, if the physical squeeze reasserts: pure-play producers with unhedged production tied to spot. Companies with reserves outside the two choke-point jurisdictions. Alphamin's Bisie mine in the DRC, guiding 20,000 tonnes for 2026 and contributing roughly 6% of global supply, is the obvious name. But the broader point isn't one stock. It's the structure.

The paper market just repriced tin on a narrative about whether AI companies will spend slightly less next quarter. The physical market is repricing tin on whether there's enough metal to solder the chips that already exist. In my experience, when those two disagree, the physical market wins. It just takes longer than you'd like — and the drawdown in between is where most people lose their conviction.