Zinc Aluminum Smelter Capacity US: The Invisible Infrastructure Holding Up Everything Else

US zinc and aluminum smelter capacity decline eliminated domestic gallium supply and cut sulfuric acid production. The invisible infrastructure nobody talks about controls everything downstream.

Zinc and aluminum smelter capacity in the United States has been declining for decades — and the consequences of that decline extend far beyond the metals themselves into gallium supply, sulfuric acid production, silver output, and industrial chemical availability.

Zinc smelting produces gallium as a byproduct. Aluminum smelting produces gallium through a different process route. Close the zinc and aluminum smelters, and you close the domestic gallium supply — the metal essential to directed energy weapons and advanced semiconductor devices. The connection is not obvious to anyone who doesn’t map the full industrial metabolism, which is exactly the kind of systems thinking Craig Tindale argues we have lost.

The same logic applies to sulfuric acid. Zinc and copper smelting produce sulfur dioxide as a byproduct, which is captured and converted to sulfuric acid through the contact process. Sulfuric acid is the essential reagent in copper mining and refining. Close the smelters, lose the sulfuric acid, create a dependency on imported reagents for the copper mining operations you are trying to expand domestically. The circular dependency is complete and largely invisible to policymakers.

The US aluminum smelting industry has been particularly hard hit. Primary aluminum production requires enormous quantities of electricity at prices that domestic utilities cannot consistently provide at competitive cost. The result has been a steady contraction of domestic smelting capacity, with production shifting to regions with cheaper hydroelectric power — and to China, which built aluminum smelting capacity at the scale the global market required and priced it below what Western competitors could match.

Rebuilding zinc and aluminum smelter capacity in the US is not glamorous. It is also not optional if the downstream dependencies on gallium, sulfuric acid, and silver are to be addressed. The infrastructure that nobody talks about is frequently the infrastructure that everything else depends on.

Siemens, €143 Billion Backlogged, and the Electrification Fantasy

Siemens has a €143 billion transformer backlog and a five-year wait time. The AI buildout can’t happen without electricity. The electricity can’t happen without transformers.

Siemens’ current order backlog for electrical transformers: €143 billion. Current wait time if you order a transformer today: five years.

Five years. For a transformer. The kind you need to connect a data center, a factory, a charging network, or a renewable energy installation to the grid.

This single data point should end the conversation about whether America can build the AI infrastructure it has announced on the timeline it has announced. It can’t. Not because the financing isn’t there. Not because the land isn’t available. Not because the technology doesn’t work. Because the physical hardware required to connect these facilities to electrical power is backlogged for half a decade at the world’s leading manufacturer.

Craig Tindale cited this in his Financial Sense interview as one of the clearest illustrations of the gap between the financial narrative around AI and the material reality. We have Nvidia chips sitting in inventory, undeployed — not because there’s no demand, but because the data centers that would house them can’t get power connections. The transformer is the bottleneck, and the transformer backlog is the direct result of two decades of underinvestment in electrical infrastructure manufacturing capacity.

The rural electrification analogy is apt. In the 1930s, bringing electricity to rural America required an enormous coordinated buildup of generation capacity, transmission infrastructure, and distribution hardware. It took years and required deliberate government intervention to overcome market failures in low-density areas. We are attempting something of comparable complexity — multiplying the electrical capacity of major industrial corridors to support AI, EV charging, and re-shored manufacturing — without having built the manufacturing capacity to produce the equipment that would make it possible.

Tindale’s prediction: by late 2027, the electricity constraint on the AI buildout becomes undeniable and public. The stories about transformers, substations, and grid interconnection queues — already visible to those paying attention — become the dominant narrative. The AI hype cycle collides with the infrastructure reality cycle. Position accordingly.

The Statistical Surge: Why America’s Industrial Fires Aren’t Random

Systematic analysis of 27 industrial incidents reveals a pattern of infrastructure decay, not random accident.

Between 2024 and 2026, something changed in the data on industrial incidents across North America. Fires at aluminum smelters. Explosions at chemical processing plants. Equipment failures at facilities that had been running, more or less quietly, for decades. Individually, each event has an explanation — a valve left open, a maintenance cycle deferred, an aging compressor that finally gave out. Collectively, they form a pattern that demands a different explanation.

Craig Tindale, a systems analyst with four decades of infrastructure planning experience, began cataloguing these incidents systematically after noticing that a single New York aluminum smelter suffered three separate fires in rapid succession — each one interrupting a recovery from the last. The cumulative cost ran into billions. That sequence, he argued, wasn’t bad luck. It was a symptom.

Tindale reviewed 27 documented incidents and cross-referenced official investigative reports. His finding was straightforward: the common thread wasn’t sabotage, wasn’t regulatory failure, wasn’t a single point of negligence. It was systemic deterioration. America’s industrial midstream — the smelters, refineries, chemical networks, and processing plants that sit between raw material extraction and finished manufacturing — had been allowed to decay for two decades while capital flowed elsewhere.

When the Biden administration’s green energy push arrived with its enormous demand on industrial capacity, it hit infrastructure that was no longer fit for purpose. The bill of materials required to rebuild wasn’t available. The workforce trained to operate these systems had dispersed. The safety protocols had atrophied. And so things broke — not because of any single decision, but because of a thousand decisions made over twenty years to defer, divest, and offshore.

Key findings from Tindale’s analysis:

Industrial complexity — a published metric tracking the diversity and depth of a nation’s production capacity — has been declining in the U.S. for years. Each closure of a processing facility doesn’t just remove capacity; it removes the knowledge base, the supplier relationships, and the safety culture that surrounded it. These don’t reconstitute automatically when demand returns.

The FOMC’s monitoring frameworks, built on neoclassical price theory, assume closed facilities reopen when demand justifies it. That assumption requires that the human capital, physical plant, and supply chains remain available. They don’t. Once dispersed, they take a decade or more to rebuild — if they rebuild at all.

Bottom line: Track industrial incident frequency as a leading indicator. A rising thermal event rate isn’t a maintenance story. It’s a sovereignty story.