In 2008, a tornado bore down on Jackson, Mississippi.
Five warning sirens were supposed to sound across the city.
None of them did.
Thieves had ripped out the copper wiring that powered them.
Today, AT&T is offering $10,000 bounties to anyone who helps catch copper thieves in Texas.
The FBI classifies copper theft as a threat to critical U.S. infrastructure.
Telecom companies reported over 15,500 incidents of copper theft and cable vandalism in the past twelve months alone — nearly double the prior year.
In Denver, police logged 500 cases of copper theft in a single year.
In Las Vegas, $1 million worth of HVAC systems were stripped from buildings — not for the equipment, but for the copper coils inside.
In California, thieves shut down state government websites by ripping copper wire from a mile-long stretch of highway.
In Oakland, a DMV office had to close after someone stole the copper from its walls.
When people risk electrocution in the middle of the night to steal a metal from the ground, something deep is happening in the economy.
That something has a name: a structural shortage.
And it’s about to collide with the largest infrastructure buildout in human history.
The Number That Should Keep Every AI Investor Awake
On January 29th, 2026, copper prices exploded 11% in a single session on the London Metal Exchange — the biggest one-day move since the 2008 financial crisis.
The price hit $14,527 per tonne, an all-time record.
That same day, Meta announced it would nearly double its AI spending to $135 billion for 2026.
Coincidence? Not even close.
Here’s the math that nobody in the AI trade is doing:
A conventional data center uses between 5,000 and 15,000 tonnes of copper. A hyperscale AI data center — the kind being built to run Nvidia’s latest systems — can require up to 50,000 tonnes. Per facility.
That’s not a typo. A single AI data center can consume as much copper as a small city’s entire electrical grid.
Now multiply that by the dozens of gigawatt-scale campuses that Microsoft, Amazon, Meta, and Google are racing to build. Then add the grid upgrades, the substations, the transmission lines, the transformers — every single one of them packed with copper.
S&P Global published a study in January 2026 that lays it out in black and white: the world faces a 10 million metric tonne copper deficit by 2040. Demand is projected to surge 50% from current levels. And supply? It’s not even close.
Daniel Yergin, vice chairman of S&P Global, summarized it in one devastating sentence: “Copper is the great enabler of electrification, but the accelerating pace of electrification is an increasing challenge for copper.”
Translation: $650 billion in Big Tech capex doesn’t matter if you can’t wire the servers.
The 29-Year Problem
Here’s where the story turns from alarming to absurd.
The United States sits on copper reserves comparable to Canada and Australia combined. There is no shortage of copper in the ground. The shortage is in our ability to get it out.
According to S&P Global, it takes an average of 29 years to bring a copper mine from discovery to production in the United States. Twenty-nine years. That’s the second-worst timeline in the world — behind only Zambia.
In Australia, it takes 20 years. In Ghana, roughly 10. In the U.S., seven to ten of those years are consumed by the permitting process alone.
Since 2002, only three new mines have come online in the United States. None of them were on federal land.
The poster child for this dysfunction is Resolution Copper, a project run by Rio Tinto and BHP in Arizona. It sits on top of the second-largest untapped copper deposit on Earth — an estimated 40 billion pounds. Enough to supply 25% of America’s copper needs for decades.
The deposit was identified in the 1990s. Rio Tinto and BHP have spent over $2 billion developing it. And as of February 2026, not a single tonne of copper has been extracted.
Why? Because Oak Flat — the mesa above the deposit — is sacred land for the San Carlos Apache, who have held ceremonies there since time immemorial. The Supreme Court rejected the tribe’s appeal in May 2025. Lawsuits from environmental groups are still active. The earliest possible production date is sometime in the 2030s.
I’m not here to take sides on the ethics of that fight. Both the economic and cultural stakes are real and enormous.
But the investment implication is undeniable: even when we know exactly where the copper is — even when it’s the second-largest deposit on Earth — it can take 30+ years to bring it to market.
Now ask yourself: if AI data center demand is tripling by 2035, and the copper to build them takes decades to mine, where exactly is the supply going to come from?
The Perfect Storm
Let me stack the demand drivers, because the AI story is only one piece:
AI & Data Centers: S&P Global expects total data center capacity to reach 550 gigawatts by 2040 — more than 5x today’s levels. At 27-33 tonnes of copper per megawatt of installed capacity, this alone represents millions of tonnes of incremental demand.
Electric Vehicles: A single EV uses 3-4x more copper than a gasoline car — roughly 80-100 kg per vehicle. Global EV adoption is accelerating.
Grid Expansion & Replacement: The U.S. grid needs trillions of dollars in upgrades. Every transformer, every cable, every substation runs on copper.
Defense: Global defense spending could double to $6 trillion by 2040. Military hardware is copper-intensive.
Renewables: Every wind turbine, every solar installation, every battery storage system requires copper in the wiring, the connections, the inverters.
S&P Global estimates that AI/data center demand and defense demand alone will add 4 million metric tonnes of combined annual copper consumption by 2040 — each roughly tripling from today’s levels.
And on the supply side? Mine output grew less than 2% in 2024. Chile and Peru — which together produce nearly 40% of the world’s mined copper — are struggling with lower ore grades, water shortages, and permitting delays. Anglo American just cut its 2026 production forecast. Freeport’s Grasberg mine in Indonesia shut down through Q2 2026.
Wood Mackenzie projects a 6 million-tonne supply shortfall by 2035.
This isn’t a cyclical blip. This is a structural crisis unfolding in slow motion.
Why I’m Writing This Now
Copper hit its all-time high less than a month ago. JPMorgan is projecting $12,500/tonne for Q2 2026. UBS sees a structural bull case. Deutsche Bank expects mine supply to remain under pressure through at least mid-2026.
Yet copper mining stocks — the companies that actually pull this metal out of the ground — are trading at valuations that suggest the market hasn’t connected the dots.
The mega-cap miners are priced for moderate growth.
The mid-tier producers are trading at single-digit earnings multiples. And the junior explorers — the companies sitting on the deposits that will supply the next generation of demand — are, in many cases, trading near their lows.
This is the exact same setup I identified in the transformer supply chain six months ago: a critical physical bottleneck in the AI buildout, where committed capital expenditure from the largest companies on Earth must flow through a handful of companies that can’t be bypassed, substituted, or fast-tracked.
Except the copper thesis is bigger. Transformers are a $20-33 billion market. Copper is a $200+ billion market. And the deficit is structural, not cyclical.
I’ve spent the last two weeks building a complete copper supply chain model — from mining through smelting through fabrication — identifying the companies positioned at each chokepoint where the math of supply and demand creates inevitable pricing power.
What’s Behind the Paywall
For premium subscribers, I’m sharing my complete copper investment thesis:
✅ My 6 copper picks across the value chain — from major producers to junior explorers, with exact tickers, entry zones, position sizes, and 24-month price targets
✅ The “29-Year Arbitrage” — why the market’s inability to price long-duration supply constraints creates a specific, tradeable opportunity in 3 mid-cap miners sitting on permitted, near-production deposits
✅ The Smelter Bottleneck — why China controls 40-50% of global copper smelting capacity, why that’s about to become a crisis, and the 2 companies building Western alternatives
✅ My copper demand model through 2035 — breaking down consumption by sector (AI/data centers, EVs, grid, defense, renewables) with pricing scenarios at $12,000, $14,000, and $16,000/tonne
✅ The “Copper Recycling” sleeper play — with 30% of global copper demand met by recycling and ore grades falling 40% since 1991, urban mining is becoming economically compelling. I’ve found one company that’s positioned to dominate this shift.
✅ The tariff trade — Trump imposed 50% tariffs on copper imports in August 2025. U.S. copper trades at a persistent premium to the LME. I explain how to position for both scenarios: tariffs tighten further, or they expire.
The first company I’m profiling is a mid-tier producer with a fully permitted mine in a Tier 1 jurisdiction, first production expected within 18 months, and a resource base that could supply 3% of U.S. demand for 25 years. It’s trading at 0.3x NAV.
I believe this stock triples before a single tonne of copper ships.