Tearing apart the tin solder supply crunch

Tin.

Financial media has discovered that AI datacenters use tin. The narrative is half right and entirely beside the point. Yes, demand is rising — AI server PCBs are tin-intensive, every solder joint on every board. But the story isn't on the demand side. It never is with this metal. The story is in a jungle in northern Myanmar that most commodity traders couldn't find on a map, and a licensing bureau in Jakarta that has been running three years behind schedule.

Tin is not a precious metal or a battery metal. It is the fastener. Every printed circuit board on earth — smartphones, EVs, medical devices, defense electronics, the servers running the AI models your fund manager is writing reports about — is held together by tin solder. Strip it out and none of it functions. There is no substitute at scale. The market for it is tiny by metals standards, roughly 400,000 tonnes per year globally, which means a disruption that would be a footnote in iron ore barely registers before it moves price 30 percent.

The Man Maw mine in Wa State, northern Myanmar, has been effectively closed since August 2023. That is three years of suspended output from a deposit that at peak supplied more than 70 percent of Myanmar's tin — which itself represented the world's second-largest producing nation. Three years. The mainstream press treats it as a recurring item, a headline to attach to price moves. It is not a headline. It is a structural hole in global supply that has been papering over itself with Chinese smelter drawdowns and thin LME inventory until there is nothing left to draw.

Meanwhile, Indonesia — the world's largest refined tin exporter — entered 2026 having voided its previous RKAB production quotas and required every licensed producer to resubmit from scratch. Jakarta has done this before. The market always assumes it resolves quickly. It has not resolved quickly. And this time two supply holes are open simultaneously.

Numbers only. The opinion is in the next section.

Myanmar holds roughly 700,000 tonnes of tin reserves — about 15 percent of global total, third after China and Indonesia. At Man Maw's operating peak, it fed Chinese smelters at a rate that kept the downstream supply chain functional. Tin concentrate flows from Myanmar to China fell 77 percent year-on-year through mid-2025. By July 2025, monthly flows had dropped to 933 tonnes. The International Tin Association granted the first restart permits for Man Maw in July 2025. As of early 2026, the mine is still operating at a fraction of capacity — de-watering, rehiring, infrastructure repair after the March 2025 earthquake. These are not fast processes.

Indonesia's RKAB cycle: the government declared all 2024 and 2025 production quotas obsolete at the start of 2026 and required full resubmission. State-owned PT Timah has permits. The private operators — who account for the majority of Bangka Island output — are still waiting. Jakarta seized 500 tonnes of illegally mined tin in a crackdown in early 2026, signaling that the compliance pressure is real and will not be waived for convenience. Indonesian exports that were already down 55 percent year-on-year in H1 2024 have not recovered to baseline.

LME tin warehouse stocks ended 2025 at roughly 8,000 tonnes — a two-year high that was built by a rapid 2,300-tonne December delivery. That cushion looked comfortable in January. LME and SHFE combined stocks cover approximately 6.3 days of global demand. Six days. For context: copper's equivalent buffer at the height of the 2021 squeeze was roughly 4 days. Tin is approaching that level without any of the fanfare that copper received.

The ITA's base-case projection for 2026 shows a 300-tonne surplus — but only if both Indonesia normalizes its RKAB process and Myanmar restarts meaningfully. Both conditions remain unconfirmed. The 300-tonne surplus is a rounding error in a 400,000-tonne market. One delayed boat out of Bangka Island and it flips.

The part nobody is tracking: the UWSA. The United Wa State Army controls Man Maw. It is not a government, not a publicly traded entity, and it does not issue press releases in English. Its permit approvals are, in the ITA's own careful phrasing, "notoriously opaque." The July 2025 restart licenses were granted by the UWSA, not by Naypyidaw, and the UWSA has not committed to a production schedule that any analyst can independently verify. Calling this a "mine restart" is doing a lot of work for what is actually a controlled trickle under a warlord bureaucracy that may or may not continue next quarter.

The AI datacenter narrative is the cover story. Every tech analyst writing about tin right now is anchored to the demand side — server PCBs, GPU substrates, the data infrastructure buildout. They are correct. Tin-halide perovskite solar cells are also emerging as a demand driver. Sodium-ion battery packs, which CATL is scaling, are tin-intensive. The demand picture for this metal through 2030 is genuinely strong. I am not arguing with it.

The weak link is that the fund positioning has been built entirely on a demand thesis, which means it is fragile to any supply-side disappointment for the wrong reason. If Man Maw delivers a surprise batch of concentrate to Chinese smelters in Q3, the longs flush. If Indonesia clears its RKAB backlog faster than expected, the longs flush. The structural deficit is real, but the trade is crowded in the same direction, which means the path to it is violent.

The secondary weak link is Indonesia's domestic political calculus. President Subianto's crackdown on illegal mining is genuine — 500 tonnes seized, 1,000 unlicensed mines ordered shut in Sumatra. This is not going to reverse. The RKAB re-submission requirement is structural, not administrative. Jakarta is redesigning its relationship with the tin industry, not processing a paperwork backlog. That takes years to normalize, not months. The market is pricing it as a temporary disruption. It is not temporary.

At 6.3 days of demand cover, tin has almost no shock absorber. There is no spare capacity to route around a disruption. The sequence, when it triggers, is fast.

Tin spiked above $50,000 per tonne in February 2026 — briefly, before pulling back to the $46,000 range. That move was driven primarily by speculation. The physical market had not yet forced the issue. When the physical market does force the issue, the move is not temporary and it does not pull back to $46,000.

The PCB manufacturers and electronics assembly plants that use tin solder are not set up to hedge this market the way an aluminum smelter hedges LME aluminium. Tin is too small, too illiquid, and for most of the last decade too stable to bother with a formal hedging program. They carry 30-to-60-day inventory buffers and rely on their solder suppliers to absorb price volatility. When the solder suppliers run out of cushion, the price passes through to the electronics supply chain directly. That is when the datacenter buildout companies start having conversations they did not expect to have about a metal nobody was watching.

The UWSA will not issue a timeline. Jakarta's regulators will not accelerate for the market's convenience. The electronics industry does not have a Plan B for tin. These three facts are sitting in the same machine together, and the machine has 6.3 days of buffer left.

I have watched small markets like this go from "well-supplied" to "where do I even get physical" in a matter of weeks. The nickel squeeze of 2022 took most of the industry by surprise and it is a bigger, more liquid market than tin. When tin goes, it goes without much warning. The loading docks in Bangka Island and the jungle roads out of Man Maw are the only early warning system that exists — and neither one is visible from a Bloomberg terminal. I would be watching them.