Every other sector in this document — AI, robotics, defense, EVs, energy infrastructure — depends on the same dozen-or-so raw materials . Rare earths for the magnets in motors. Lithium for the batteries. Copper for the wires. Silicon carbide for the next generation of power chips. Industrial gases for every fab.
Here's the wrinkle: China spent twenty years locking up most of the world's supply of all of them. Now the West is racing to rebuild a parallel non-Chinese supply chain — and that costs trillions, takes a decade, and creates a generational opportunity for the few Western companies that already have mines, refineries, and processing capacity.
You don't have to pick which downstream theme wins. If AI wins, materials win. If robots win, materials win. If EVs win, materials win. If reshoring wins, materials win. The same dozen rocks sit underneath every story.
You cannot build a motor without a permanent magnet. You cannot build a permanent magnet without rare earths. China makes 90% of both.
Every electric motor — in an EV, a wind turbine, a robot, a missile guidance system, a smartphone vibrator — uses a neodymium-iron-boron (NdFeB) permanent magnet . For high-temperature applications (motors, especially EV drive motors) you additionally need dysprosium and terbium (the "heavy" rare earths). China refines ~90% of rare earths globally and produces ~95% of permanent magnets . The MP Materials and USA Rare Earth names appear in AI Choke 01 (rare earths flow to AI lasers and magnets); they belong here too because the magnet thesis spans every other sector. We add net-new exposure via Lynas — the only at-scale ex-China rare-earth refiner.
China imposed export controls on heavy rare earths in 2025. If the Nov 27, 2026 deadline isn't extended, prices spike and any non-Chinese supply becomes strategically priceless.
Only at-scale ex-China rare-earth miner + refiner
Australian miner that operates the only commercial-scale rare-earth refining facility outside China — in Malaysia. Building a second refinery in Texas (Lynas USA, ~2026 first production) with US Department of Defense partnership. Owns the Mount Weld mine in Western Australia, one of the highest-grade rare-earth deposits on Earth.
Covered in detail in AI Choke 01 — included here for completeness
Both names sit primarily in the AI rare-earths chokepoint above. The thesis here is identical with an added flavor: EV motors and humanoid robots use the same magnets as AI laser systems . Any pure-play rare-earth name is implicitly a multi-sector chokepoint. See AI Sector for full cards.
Lithium is in oversupply — until it isn't. The 2025-26 trough is a setup, not a verdict.
Two years ago lithium carbonate was $70,000/ton. Today it's around $14,000/ton. A massive new supply wave (Chinese lepidolite, African hard-rock) flooded the market just as Chinese EV demand normalized. Result: every major lithium producer is unprofitable or barely break-even . That's exactly the kind of moment that sets up the next cycle — capacity expansions are being cancelled, marginal mines are closing. The companies that survive the trough with the lowest cost structures will own the next upswing. Albemarle (US, integrated), SQM (Chile, brine), and Arcadium (merger of Allkem + Livent — global presence) are the three pure-plays for Western investors.
EV demand is structurally rising. Grid storage demand is exploding. Robot batteries (small but high-quality cells) are starting to matter. Even at depressed prices, low-cost producers print cash; high-cost producers die. The reset is the opportunity.
Largest US lithium producer + global specialty chemicals
World's largest lithium producer by revenue. Operates mines + brine + processing facilities in Chile, Australia, North Carolina, Nevada. Also has a specialty chemicals business (catalysts, bromine) that provides cushion. Cyclical pure-play on lithium prices recovering — and a contrarian bet given current pessimism.
Chilean Atacama brine — lowest-cost lithium production globally
Chilean miner that extracts lithium from Atacama Desert brines — by far the lowest-cost lithium production on Earth . The Codelco partnership (signed 2024) extends Atacama operations into 2060 and gives Chile partial ownership stake. Cyclically depressed; structurally favored. Also #1 in iodine globally, #2 in specialty fertilizers — non-lithium businesses provide cushion.
Allkem + Livent merger — global integrated lithium
Smaller, more leveraged, more volatile. Argentina (brine) + Australia (hard rock) + North Carolina (downstream lithium hydroxide) — integrated supply chain. Rio Tinto announced an acquisition offer in late 2024 that closed at a controversial sub-fair-value price for some shareholders. If you missed the Rio deal you can still own Rio for lithium exposure now. Tracking thesis only.
The world will need 50% more copper by 2035. We're not opening enough mines.
Copper goes into every wire, every motor, every transformer, every cable, every battery interconnect, every data-center busbar . The International Energy Agency projects global copper demand growing ~50% by 2035 driven by EVs, the grid, AI, and electrification. Supply: roughly flat . New mines take 15+ years from discovery to first production due to permitting, environmental review, and indigenous land rights. The math doesn't work — which means prices must rise to ration demand, or capacity must come online faster. Either way, the producers with existing mines win.
You can substitute aluminum for copper in some applications. You cannot substitute copper for itself. The deficit is structural and multi-year.
Largest publicly traded copper producer — Grasberg, Cerro Verde, Morenci
The largest publicly traded copper producer in the world . Operates Grasberg (Indonesia — also the world's largest gold mine as byproduct), Cerro Verde (Peru), and Morenci (Arizona). Highest leverage to copper price among the majors — earnings move ~$1B per $1/lb copper price change. Indonesia operating environment is the perennial wild card.
Lowest-cost copper producer — Peru + Mexico operations
Less reported, less covered than FCX but with lower cash cost per pound and longer mine life . Toquepala and Cuajone in Peru, Buenavista in Mexico. Pays meaningful dividend. The "compounder" of copper miners — slower beta but durable.
Boring, monopolistic, and a chokepoint that touches every industrial process on Earth.
Every semiconductor fab, every steel mill, every hospital, every fertilizer plant, every space-launch facility runs on industrial gases : oxygen, nitrogen, hydrogen, argon, helium, plus exotic specialty gases (fluorine compounds, dopants). The industry is a tight oligopoly : Linde (US/Germany), Air Products (US), Air Liquide (France) plus Japan's Nippon Sanso. Linde and Air Products together hold ~50% of global industrial gas revenue. Multi-decade take-or-pay contracts with utility-like cash flows. Boring, secular, and largely insulated from any single end-market cycle.
The two leaders are the literal pipes. Every fab built — for AI chips or anything else — requires gas. Every blue/green hydrogen project requires their equipment. Capex by Linde and APD is a leading indicator for global industrial activity.
Largest industrial gas company globally — utility-like compounder
Largest industrial gas company in the world after the 2018 Praxair-Linde merger. ~30% global market share. Supplies oxygen, nitrogen, hydrogen, argon, plus electronics specialty gases (the latter to TSMC, Samsung, Intel — directly chokepoint-relevant to AI). 80%+ of revenue is take-or-pay. Best-in-class capital allocation: routinely 15%+ ROIC. Operating margins 25%+.
#3 globally + bet-the-company hydrogen pivot
More mid-cap than Linde, more exposed to large hydrogen projects, and recently shaken up by activist investor Mantle Ridge — which forced board changes and a strategic refresh in 2024-25. The NEOM hydrogen plant (Saudi Arabia, $8B+ capex) is the bet-the-company project. If it works, APD is the leader of the hydrogen transition. If not, APD becomes a value trap. Higher beta than LIN.
Two niche material plays that punch above their weight in AI, EVs, and defense.
As power densities rise — in EV inverters, in data-center power supplies, in solar inverters — traditional silicon power semiconductors hit physical limits. Silicon carbide (SiC) and gallium nitride (GaN) are "wide bandgap" semiconductor materials that handle higher voltages, higher temperatures, and higher frequencies. Wolfspeed is the pure-play SiC substrate maker — bet-the-company turnaround story in 2026. Meanwhile, aluminum and steel have their own pockets of strategic relevance — particularly for defense (every missile, every aircraft body) and infrastructure. Century Aluminum and Nucor are the cleanest US-listed plays.
SiC substrate is a true single-process bottleneck — growing SiC crystals is harder than silicon by orders of magnitude. Only a handful of facilities globally can do it. Aluminum smelting capacity in the US has shrunk 70% in 20 years; the few survivors are strategic.
Largest pure-play SiC substrate & device maker — restructuring play
Once the darling of SiC, now a restructuring penny stock . Operates the largest 200mm SiC substrate fab on Earth (Mohawk Valley, NY). Pivoted out of LED business to focus on power devices for EVs + industrial. Filed Chapter 11 in mid-2025 and emerged restructured. The asset is real and strategic; the equity is highly speculative .
Largest US steel producer — electric-arc furnace + low cost
Mini-mill model = lower cost than integrated mills. Diversified end-markets : construction, auto, energy, durable goods. New West Virginia sheet mill ($3B+ capex) targets data-center construction supply. Aggressive buybacks. Less exciting than tech but a quality cyclical compounder.
Only US primary aluminum producer with green-field expansion plans
US aluminum smelting has collapsed from ~5 million tons/year in 1980 to ~700,000 tons today. Century is one of the last men standing — Hawesville KY, Sebree KY, Grundartangi (Iceland). Got a $500M DOE grant to build the first new US smelter in nearly half a century — strategic for defense and infrastructure. Highly leveraged to aluminum + power costs.
Largest US flat-rolled steel + iron ore — auto + appliance + defense
Lost the Nippon Steel / US Steel deal in 2024 but remains the only fully integrated US flat-rolled producer . Volatile, cyclical, but uniquely positioned in any defense/auto/appliance reshoring scenario. Trades like a deep value name today. Optionality on Section 232 tariffs and a potential US Steel re-do under different ownership.