Tearing apart the silver byproduct trap

Silver.

The sell-side note making the rounds this week says silver demand is softening. The solar sector is thrifting — total photovoltaic silver consumption is forecast to fall 19% in 2026 versus last year. Photovoltaic consumption is forecast to drop from 186.6 million ounces to around 151 million. The chyron writes itself: Silver Demand Weakens as Solar Sector Substitutes Away.

Here's the part nobody's reading past the headline: the deficit is getting wider. Not narrower. Wider. The Silver Institute's World Silver Survey 2026 projects a 46.3-million-ounce shortfall this year — up 15% from 2025. Sixth consecutive annual deficit. The solar thrifting story isn't fixing the imbalance. It's a bandage on an arterial bleed.

The reason the deficit keeps widening despite the solar pullback is straightforward once you look under the hood. The demand the panels left behind is being filled — and then some — by data centers, EVs, and grid infrastructure. AI data center buildouts alone are absorbing more than 42 million ounces annually. Every EV rolling off the line uses 67–79% more silver than the gas car it replaces. These are not speculative demand pools. They're purchase orders.

But the real story isn't demand. It never is with silver. The real story is why supply can't respond. And that story starts with a word the analyst reports keep glossing over: byproduct.

Story off. Numbers on.

Global silver mine production peaked in 2016 at 900.1 million ounces. It has declined at roughly 0.6% per year since. The World Silver Survey 2026 forecasts 844.1 million ounces for this year — a 0.3% decline from 2025. Recycling adds another 211.3 million ounces, a 7% bump and its highest level since 2012. Total supply: roughly 1.066 billion ounces. Total demand: 1.11 billion ounces. The math doesn't close.

This is the part the models get wrong every time. Silver is not like copper or zinc, where a price spike can incentivize new production. Nearly three-quarters of silver output is a side effect of mining something else. A copper mine in Peru doesn't produce more silver because silver is at $64. It produces more silver because copper economics justify running the mill. If copper falls, the silver goes with it — regardless of what the silver chart says.

I learned this the hard way in 2019 when I was long silver miners expecting the price to pull supply forward. It didn't. The lead and zinc mines that produce 29.4% of the world's silver don't care about the silver price. The copper mines that produce 28% of it don't either. You're a passenger in someone else's car.

Meanwhile, visible inventory is draining across every major exchange simultaneously. COMEX registered silver sits at 95.9 million ounces — it breached the psychologically important 90-million mark in February before partially recovering. Total COMEX holdings: 337.2 million ounces. LBMA vaults in London hold 28,431 tonnes. Total visible supply across COMEX, LBMA, SHFE, and SGE: 1.33 billion ounces as of September 9. That sounds like a lot until you remember the market pulls 46 million ounces per year more than it replaces.

This isn't cyclical softness. It's a structural mismatch between a supply base that answers to other metals and a demand base that keeps finding new reasons to consume silver. The deficit doesn't heal. It compounds.

China just flipped from seller to buyer. That's the part that should be keeping people up at night.

In 2025, China exported approximately 5,100 tonnes of silver and functioned as a meaningful contributor to global supply. By early 2026, that flow reversed. Beijing reclassified silver as a strategic material and restricted exports to a whitelist of 44 approved companies for 2026 and 2027. This is the same playbook they ran on gallium, germanium, and antimony. If you've been reading this letter, you've watched each of those supply valves close in sequence.

The Shanghai Gold Exchange is pricing silver $8 to $10 above London and New York benchmarks. That premium isn't noise. It's a physical market telling you that Chinese industrial buyers — the solar cell fabricators, the EV contact manufacturers, the electronics assemblers — are scrambling for metal that used to flow outward and now doesn't.

India is pulling from the other direction. The country spent a record $12 billion on silver imports in fiscal year 2025–2026 — up from $4.8 billion the year before. That's not jewelry demand. That's industrial consumption and physical investment running simultaneously hot.

So you have a supply base that can't expand because 74% of it is chained to someone else's mine economics. You have the world's largest exporter turning importer and gating outbound flow. You have the world's largest physical buyer setting import records. And the sell-side is writing notes about solar thrifting.

It's like watching someone celebrate a slow leak in the kitchen while the basement floods.

COMEX registered inventory hit 76 million ounces in April — a 13.4% coverage ratio against open interest. It's recovered slightly to 95.9 million, but the trajectory is clear. The paper market is writing claims on metal that is physically leaving the vaults and not returning. I've seen this movie before. The ending involves a lot of margin calls.

The trigger here isn't dramatic. It doesn't need to be. The structural math already fails — it just needs a catalyst to make the paper market acknowledge what the physical market has been saying for six years.

The choreography is mechanical, not mysterious. China's whitelist shrinks further, or Beijing applies the same escalation ladder it used on gallium — licensing, then delay, then effective embargo. Indian import demand stays hot. COMEX registered drops back toward 76 million ounces. At some point, a delivery month arrives where the number of contracts standing exceeds the metal available. That's when the paper price stops being the real price and the physical premium becomes the only quote that matters.

We've seen this sequence before. COMEX registered silver continued falling into March, with spot sliding from $67 in February to $58 by early April — a move that underscored just how vulnerable the physical market has become. The January all-time high of $118.45 wasn't driven by a headline — it was driven by a delivery squeeze that made the coverage ratio functionally irrelevant for a few days. Reuters warned the multi-year drawdown "raises squeeze risks." That was an understatement.

Where the edge sits: primary silver miners with direct spot exposure and reserves outside the Chinese supply chain. The 26% of output that actually comes from dedicated silver mines — those are the only producers who expand when silver prices justify expansion. The other 74% are bystanders. Broad precious metals ETFs average down the exposure with gold, which has different mechanics entirely. Silver royalty and streaming companies with primary-silver portfolios capture the upside without the mine-cost risk.

The consensus says silver is correcting because solar demand is falling. The vaults say 762 million ounces have left and no supply mechanism exists to put them back. When the narrative and the inventory diverge this far, I know which one to trust. It just takes longer than you'd like — and the timing, as always, is the part that costs you money.

Demand side says up. Supply side is bolted to someone else's floor. The gap doesn't close. It widens until something breaks.