Snapshot

Specialty metals — cobalt, nickel, tungsten, and tin — are four distinct commodity markets with concentrated supply chains dominated by one or two countries, supplying inputs that advanced manufacturing cannot substitute away from. Global production in 2024: cobalt ~300 kt est., nickel ~3.5 Mt, tungsten ~81 kt, tin ~371 kt (refined). Combined mine-stage revenue roughly $70–80B/year. est. Each metal has a different supply-demand balance: nickel is in structural surplus (Indonesian capacity flood), cobalt swung from glut to deficit in 2025 after DRC export controls, tin is in narrow deficit (~2,200 t in 2024), and tungsten is tight due to Chinese export restrictions. No pure-play US-listed specialty metal miner exists; exposure comes through diversified miners Vale (nickel/cobalt), Glencore (cobalt/nickel), and BHP (nickel suspended).

Tin is the most direct AI link — every AI server's circuit board uses tin solder. Cobalt and nickel feed UPS battery banks and grid-scale storage co-located with data centers. Tungsten goes into cutting tools for precision-machining server components. Each metal's economics are distinct — nickel is drowning in Indonesian supply while cobalt had its tightest market ever in 2025. US-listed exposure only comes through diversified miners where specialty metals are a secondary business line.

The product & how money is made

• Cobalt — mined primarily as a byproduct of copper (DRC) or nickel (Indonesia, Canada). Refined into cobalt metal, cobalt hydroxide, or cobalt sulfate. 76% of 2024 demand went to batteries (EV cathodes use nickel-manganese-cobalt chemistry). The rest goes to superalloys (jet engines), catalysts, and magnets. China controls ~70% of global cobalt refining. est.
• Nickel — mined from laterite ores (Indonesia, Philippines) or sulfide deposits (Canada, Russia, Australia). Processed into Class 1 nickel (pure metal for batteries) or Class 2 (nickel pig iron / ferronickel for stainless steel). ~69% of demand is stainless steel; batteries are a growing but still minority share. Indonesian producers have the lowest costs (~$10,000–12,000/t). est.
• Tungsten — mined as wolframite or scheelite ore, processed into ammonium paratungstate (APT), then into tungsten carbide powder or pure tungsten metal. Used in cutting tools, drill bits, armor-piercing munitions, and high-temperature aerospace parts. Priced per metric tonne unit (MTU) of WO₃ content. Chinese state-linked producers control 82% of mine output.
• Tin — mined from cassiterite ore, smelted into refined tin ingots. 51% of demand is solder (every printed circuit board uses tin solder). Rest goes to tinplate (food cans), chemicals, and lead-acid batteries. The market is small (~371 kt refined) and thinly traded on the LME.

Cobalt Institute, "Cobalt Market Report 2024"; International Tin Association, "Global tin production falls in 2024"; USGS Mineral Commodity Summaries 2025 (tungsten); Stainless Steel World, "The world nickel market in 2025 and 2026."

Demand

Contracted / observable (recent actuals)

Cobalt Institute via mining.com (May 2025); Stainless Steel World (Nov 2024); USGS MCS 2025; ITA "Tin use in recovery cycle" (Oct 2024).

Forecast / projected

AI-specific demand: Tin is the most direct link — every AI server's circuit board uses tin solder. Cobalt and nickel feed UPS battery banks and grid-scale storage co-located with data centers. Tungsten goes into cutting tools for precision-machining server components. None of these AI demand channels is large enough to move the overall balance by more than a few percent. est.

Fastmarkets (Jan 2026) for cobalt 2026; Stainless Steel World for nickel; Core Consultants Group (May 2025) for tungsten CAGR; ITA for tin; BMI/Fitch Solutions for tin $35,000/t forecast.

Supply

Current capacity and production

Bottlenecks

• Cobalt: DRC export quota physically halted shipments — as of late Dec 2025, ~18,125 t of Q4 quota was unshipped. No DRC cobalt refining capacity exists; a sulfate refinery is planned for 2030. Indonesian cobalt-in-MHP (mixed hydroxide precipitate) is ramping from 46,300 t (2025) to 67,500 t (2026), partially offsetting DRC tightness.
• Nickel: No bottleneck. Indonesia is flooding the market. Combined LME + SHFE stocks reached 289,500 t by Oct 2025, the highest since May 2018. Chinese-origin material grew from 47% to 68% of LME warehouse stocks in 2025. Multiple Western producers (BHP, Glencore's Koniambo) have shut down or curtailed because Indonesian costs are too low to compete with.
• Tungsten: New non-Chinese mines take 5–7 years to develop. Almonty's Sangdong mine (South Korea) is the main near-term Western project, potentially adding ~7% of global supply. est. Scrap/recycling provides ~1/3 of supply when prices are high. est.
• Tin: Thin pipeline of mining projects globally. Myanmar's Man Maw mine (historically a major source of Chinese smelter feed) has been suspended; permit announcements have not translated into actual shipments. Chinese smelter output constrained by lack of concentrate.

Fastmarkets (Jan 2026); mining-technology.com (Aug 2024); USGS MCS 2025; ITA (Feb 2025); Core Consultants Group (May 2025); Stainless Steel World (Nov 2024, Apr 2025).

The gap

Three of four metals (cobalt, tungsten, tin) are in deficit or tight, with prices rising. Nickel is the outlier — Indonesia's massive capacity expansion created a structural surplus that took prices from $48,000/t (Mar 2022 peak) to ~$15,000–19,000/t. est.

Fastmarkets (Jan 2026) for cobalt deficit; Stainless Steel World for nickel surplus; Core Consultants for tungsten; mining.com for tin deficit; ycharts.com for cobalt price series; carboncredits.com for nickel price.

The players

Vale is 75%+ iron ore by EBITDA est.; its nickel is the largest among these three (177 kt) but it took a $4.6B impairment on Canadian nickel assets in 2025 and is reviewing its Thompson, Manitoba operations. BHP suspended its entire Nickel West operation in October 2024 due to Indonesian oversupply, spending ~$300M/year in care-and-maintenance with a review by February 2027. Glencore is the world's largest cobalt producer (36.1 kt in 2025) but cobalt is a small fraction of its $231B revenue, and it put its Koniambo nickel operation on care-and-maintenance.

stockanalysis.com for VALE statistics; BHP annual report 2024; Glencore preliminary results 2024; companiesmarketcap.com for Glencore; Vale 6-K filing (2025 results); BHP/Glencore production reports.

The price of exposure

Vale at $16.06 / $69B market cap. The energy transition metals segment (nickel + copper + cobalt + PGMs) generated $8.2B revenue and $3.4B EBITDA in FY2025 — about 22% of the company's $15.5B total adjusted EBITDA. At $69B market cap, $69B × 22% = ~$15B implied value for the specialty metals exposure, or ~4.4× segment EBITDA. est. Nickel all-in cost is $12,158/t against LME ~$15,000–19,000/t est.. Forward P/E of 7.9 reflects analyst expectations that iron ore holds up. Annual dividends of $4.3B (5.6% yield) are funded primarily by iron ore cash flows.

BHP at $90.85 / $236B market cap. Nickel operations are fully suspended. BHP is spending ~$300M/year to keep Nickel West in care-and-maintenance. At today's price, an owner gets zero current nickel production and a call option on restart if nickel prices recover above ~$20,000–25,000/t. est. BHP's value is copper (growing via the OZ Minerals acquisition) and iron ore. Specialty metals exposure: effectively zero until the February 2027 review.

Glencore at $8.07 / $95B market cap. Produced 36.1 kt of cobalt (world's largest) and 71.9 kt of nickel in 2025. At cobalt ~$56,000/t, that cobalt production is worth ~$2B in revenue; nickel at ~$15,000/t adds ~$1.1B. Combined ~$3.1B of specialty metals revenue against $231B total — about 1.3%. est. Glencore's marketing/trading arm ($2.3–3.5B EBIT guidance) earns from physical commodity trading spreads, providing indirect specialty metals leverage through volatility. The FY2024 net loss ($1.6B) reflected coal price declines and impairments. Forward P/E of 13.9 implies analyst consensus expects recovery. est.

Dilution check: Vale has been relatively stable at ~4.26B shares. BHP and Glencore have both conducted buybacks — Glencore authorized $1B in buybacks for 2025. No significant dilution risk at any of the three.

What to deep-dive next

• DRC cobalt quota mechanics. The DRC can adjust export quotas quarterly. The 87,000 t/yr quota against ~244 kt/yr production means ~64% of DRC output is being withheld or redirected. est. Understanding the government's revenue needs vs. market-control ambitions determines whether cobalt stays tight or reopens.
• Indonesian nickel cost curve. At what nickel price do Indonesian HPAL plants become unprofitable? That price (~$12,000–14,000/t est.) is the floor. If enough Indonesian capacity shuts, the surplus narrows. BHP's restart decision hinges on this.
• Cobalt-free battery chemistry adoption rate. LFP batteries (no cobalt, no nickel) took ~40% of the global EV battery market in 2024, up from ~15% in 2020. est. If LFP continues taking share, cobalt's 76% battery demand exposure shrinks structurally.
• Tungsten: Almonty's Sangdong ramp. If successful, this South Korean mine adds ~7% of global supply outside China est. — the first meaningful non-Chinese new supply in years. Almonty Industries (ALY on TSX) is the closest to a pure-play Western tungsten company.
• Tin: Myanmar Man Maw restart. Myanmar holds ~700,000 t of tin reserves (15% of global). est. Whether mining actually resumes determines whether the tin deficit widens or closes.
• Vale's Thompson strategic review outcome. If Vale exits Canadian nickel (high-cost sulfide), its nickel production shrinks but margins improve. The $4.6B impairment signals management may be done investing in nickel at current prices.

Sources & confidence