Technology

Who Controls the Critical Minerals Powering AI?

Published

October 1, 2026 10:04 am

Srijaa Chatterjee

Design

Amy Kuo

Three Critical Minerals Behind the AI Supply Chain

Key Takeaways:

China accounts for 99% of estimated global processed gallium production and 69.2% of rare earth mine production.
The Democratic Republic of the Congo produces 52% of the world’s tantalum, followed by Rwanda and Nigeria.
These lesser-known materials are used in high-performance semiconductors, magnets, and power-management components found across AI hardware.

AI infrastructure depends on far more than chips and electricity. It also relies on a range of critical materials, including gallium, rare earths, and tantalum, that are essential to semiconductors, magnets, and power-management systems.

These three materials are highlighted in Figure 2.3 of the World Bank’s World Development Report 2026 because their production is concentrated in a relatively small number of countries.

The figures here are 2025 estimates from the U.S. Geological Survey’s Mineral Commodity Summaries 2026, using processed production for gallium and mine production for rare earths and tantalum.

Gallium: China’s 99% Share

Here is the World Bank’s breakdown of estimated processed gallium production:

Country/Economy	Share of World Production (%)
🇨🇳 China	99.0
🌐 Other countries	1.0

China accounts for 99% of the total. However, that does not mean China holds 99% of the world’s gallium underground. Gallium is generally recovered as a byproduct of processing bauxite for aluminum, and China’s enormous aluminum industry has helped it develop gallium extraction at scale. Earlier oversupply and low prices also contributed to producers elsewhere exiting the market.

Gallium matters because compounds such as gallium nitride (GaN) can operate efficiently at high voltages and temperatures. These properties make gallium useful in high-performance semiconductors and power electronics, an increasingly important consideration as AI data centers require more computing power and electricity.

Rare Earths: More Than One Element

Country/Economy	Share of World Production (%)
🇨🇳 China	69.2
🇺🇸 United States	13.1
🇦🇺 Australia	7.4
🇲🇲 Myanmar	5.6
🇹🇭 Thailand	1.2
🇮🇳 India	0.7
🌐 Other	2.6

China produces 69.2% of the total, followed by the United States at 13.1% and Australia at 7.4%.

Despite the name, “rare earths” are actually a group of 17 elements. Different members serve different roles across computing hardware. Neodymium and praseodymium, for example, are used in powerful permanent magnets found in hard drives and other equipment, while other rare earths have applications in semiconductors, displays, optics, and data storage.

Tantalum: Power Management for AI Servers

Tantalum production has a very different geographic footprint:

Country/Economy	Share of World Production (%)
🇨🇩 Congo, Dem. Rep.	52.0
🇷🇼 Rwanda	16.0
🇳🇬 Nigeria	15.6
🇧🇷 Brazil	7.6
🇨🇳 China	3.2
🇦🇺 Australia	2.0
🇪🇹 Ethiopia	1.6
🇷🇺 Russian Federation	1.2
🌐 Other	0.8

The Democratic Republic of the Congo supplies 52% of estimated mine production, while Rwanda and Nigeria contribute 16% and 15.6%, respectively. Together, those three countries account for 83.6%.

Tantalum’s AI connection is largely about capacitors. Polymer tantalum capacitors can provide high capacitance in a compact package along with stable electrical performance, making them useful in demanding power systems. Industry reporting in 2026 points to AI servers as a growing source of tantalum capacitor demand.

When Mineral Supply Becomes a Strategic Bottleneck

Production concentration does not automatically translate into an AI advantage. Advanced chips, fabrication capacity, energy infrastructure, and capital all matter. However, when one or a handful of countries dominate production of a critical material, export restrictions or other supply disruptions can ripple through global technology supply chains.

That risk is already visible. China has introduced export controls on gallium-related items and selected medium and heavy rare-earth items in recent years. Those measures have encouraged governments and companies elsewhere to pursue alternative suppliers and new production capacity. In Central Africa, concentrated tantalum production presents a different set of challenges, including potential supply disruptions and responsible sourcing concerns.

As AI infrastructure expands, securing the minerals behind chips, servers, and power systems could become increasingly important alongside securing the chips themselves.

Learn More on the Voronoi App

To see where the U.S. is most dependent on foreign mineral supplies, see Ranked: U.S. Import Reliance for 37 Critical Minerals on the Voronoi app.

Don't Miss

Mapped: Where U.S. Jobs Are Most Exposed to AI

Technology

Mapped: Where U.S. Jobs Are Most Exposed to AI

See which U.S. states have the highest share of AI-exposed jobs, led by Washington, Virginia, and Washington, D.C.

Published

October 1, 2026 7:17 am

Bruno Venditti

This visualization ranks all 50 states and Washington, D.C. by the share of workers employed in occupations considered highly exposed to AI disruption.

Where AI Exposure Is Highest Across the U.S.

Key Takeaways

Washington has the highest share of AI-exposed jobs, at 5.7% of its workforce.
California has the largest number of AI-exposed positions overall, at roughly 724,000 jobs.
Mississippi has the lowest share of AI-exposed jobs in the study, at 1.9%.

Artificial intelligence is becoming capable of performing tasks across a growing range of white-collar and technical occupations.

This visualization maps all 50 states and Washington, D.C., by the share of workers employed in occupations considered highly exposed to AI-related disruption. It also highlights the states with the largest and smallest numbers of workers in these roles.

The data for this visualization comes from SmartAsset, using U.S. Bureau of Labor Statistics data and research from the Virginia Economic Information and Analytics Division. The analysis covers 26 occupations identified as having particularly high exposure to potential AI-related disruption.

AI exposure does not necessarily mean these jobs will disappear. Instead, it reflects how susceptible their tasks may be to changes such as automation, weaker hiring demand, wage pressure, or restructuring as AI tools become more capable.

Why Washington Tops the Map

Washington ranks first, with 5.7% of its workforce employed in highly AI-exposed occupations.

The state’s large technology sector helps explain its position, with companies such as Microsoft and Amazon supporting a significant concentration of software developers, programmers, database specialists, and other digital roles.

Rank	State or District	AI-Exposed Jobs (%)
1	Washington	5.7%
2	Virginia	4.6%
3	District of Columbia	4.5%
4	California	4.0%
5	Utah	4.0%
6	Maryland	3.9%
7	Colorado	3.7%
8	New Hampshire	3.7%
9	Texas	3.6%
10	North Carolina	3.5%
11	South Dakota	3.5%
12	Oregon	3.4%
13	New Jersey	3.4%
14	Massachusetts	3.2%
15	Minnesota	3.2%
16	Georgia	3.2%
17	Arizona	3.1%
18	New York	3.1%
19	Tennessee	3.1%
20	Nebraska	3.1%
21	Florida	3.0%
22	Connecticut	2.9%
23	Michigan	2.9%
24	Missouri	2.8%
25	Illinois	2.8%
26	Wisconsin	2.8%
27	Rhode Island	2.8%
28	West Virginia	2.7%
29	Kansas	2.7%
30	Pennsylvania	2.7%
31	Iowa	2.7%
32	Delaware	2.7%
33	Vermont	2.7%
34	Ohio	2.6%
35	Alabama	2.6%
36	Maine	2.5%
37	Idaho	2.5%
38	Montana	2.5%
39	Nevada	2.5%
40	Alaska	2.4%
41	South Carolina	2.4%
42	North Dakota	2.3%
43	Kentucky	2.3%
44	Oklahoma	2.3%
45	Louisiana	2.3%
46	Indiana	2.2%
47	Arkansas	2.2%
48	Wyoming	2.2%
49	New Mexico	2.1%
50	Hawaii	2.1%
51	Mississippi	1.9%

Virginia follows at 4.6%, reflecting its mix of technology firms, federal contractors, and knowledge-based employment. Washington, D.C., ranks third at 4.5%, while California and Utah round out the top five at 4.0% each.

At the other end of the ranking, Mississippi has the lowest share in the study, at 1.9%.

California Has the Most AI-Exposed Jobs

California has the largest absolute number of workers in highly exposed occupations, with approximately 724,000 positions.

Texas follows with about 500,000, while New York and Florida each have roughly 301,000.

Rank	State	States With the Most AI-Exposed Jobs
1	California	724K
2	Texas	500K
3	New York	301K
4	Florida	301K
5	Washington	202K

At the other end of the scale, Wyoming has just 6,000 AI-exposed positions, followed by Alaska and Vermont at about 8,000 each.

Rank	State	States With the Fewest AI-Exposed Jobs
1	Wyoming	6K
2	Alaska	8K
3	Vermont	8K
4	North Dakota	10K
5	Montana	13K

What Types of Jobs Are Most Exposed to AI?

The 26 occupations span technology, communications, administration, finance, and other knowledge-based fields.

Mathematicians rank as the most exposed occupation, followed by proofreaders, correspondence clerks, court reporters, and media and communication workers. Computer programmers, database administrators, web developers, software developers, writers, translators, payroll clerks, and bookkeeping workers also appear on the list.

Many of these roles involve processing information, generating or reviewing text, working with structured data, or performing routine digital tasks, all areas where generative AI and other automation tools have advanced quickly.

Still, exposure should be interpreted as the potential for jobs to change rather than a direct estimate of how many positions will ultimately be eliminated.

Learn More on the Voronoi App

To learn more about this topic, check out this graphic on the smartest AI models in 2026.

Technology

Ranked: The 10 Youngest U.S. Billionaires on the Forbes 400

America’s youngest billionaires are increasingly tied to AI, with seven of the top 10 building fortunes in the industry.

Published

September 30, 2026 8:08 am

Sofie Gilbert

Ranked: The 10 Youngest U.S. Billionaires on the Forbes 400

Key Takeaways

Seven of America’s 10 youngest Forbes 400 billionaires built their fortunes in AI.
Anthropic accounts for four of the 10, with their combined fortunes totaling an estimated $62 billion.
Nine of the 10 are self-made, but the lone heir, Lukas Walton, is the richest at $44.4 billion.

Artificial intelligence isn’t just creating some of America’s most valuable startups. It’s also creating a new generation of billionaires.

The 2026 Forbes 400 includes founders and executives behind Anthropic, OpenAI, Surge AI, and Cognition among its youngest members, highlighting how quickly AI wealth has risen into the upper ranks of America’s richest people.

This visualization ranks the 10 youngest members of the Forbes 400 by age and shows the companies behind their fortunes.

Net worth figures are as of Sept. 4, 2026, and a fortune of at least $4.4 billion was required to make this year’s Forbes 400.

How AI Is Reshaping America’s Youngest Billionaires

The youngest person in the group is 30-year-old Steven Hao, cofounder of AI coding company Cognition, with an estimated fortune of $5.8 billion.

At the other end of the ranking, several billionaires are 39 years old, including Walmart heir Lukas Walton and Anthropic cofounders Daniela Amodei and Tom Brown.

Name	Age	Net Worth (Billions)	Source of Wealth:
Steven Hao	30	$5.8B	Cognition
Palmer Luckey	33	$5.9B	Anduril
Sam McCandlish	36	$15.5B	Anthropic
Jack Clark	37	$15.5B	Anthropic
Greg Brockman	38	$25.5B	OpenAI
Edwin Chen	38	$18.0B	Surge AI
Daniela Amodei	39	$15.5B	Anthropic
Tom Brown	39	$15.5B	Anthropic
Vlad Tenev	39	$7.1B	Robinhood
Lukas Walton	39	$44.4B	Walmart

Seven names on the list made their fortunes in AI. The remaining fortunes come from Walmart, fintech company Robinhood, and defense technology company Anduril.

Four Anthropic Cofounders Are Worth $15.5 Billion Each

Anthropic accounts for four of the 10 youngest billionaires: Sam McCandlish, Jack Clark, Daniela Amodei, and Tom Brown.

Forbes estimates that each holds a stake worth $15.5 billion, putting their combined fortunes at roughly $62 billion. The four range in age from 36 to 39.

Other AI fortunes include OpenAI cofounder Greg Brockman, whose estimated net worth stands at $25.5 billion, Surge AI founder Edwin Chen at $18 billion, and Cognition cofounder Steven Hao at $5.8 billion.

Most of These Fortunes Are Tied to Private Companies

Another defining feature of the ranking is how much of the wealth comes from private companies. Eight of the 10 fortunes are based on private-company stakes valued using recent funding rounds, according to the source note accompanying the Forbes data.

Unlike public stocks, private companies don’t have continuously traded market prices. New funding rounds can reset company valuations and significantly change the estimated paper wealth of major shareholders.

That dynamic is especially important for privately held AI companies, where rising valuations have created multibillion-dollar fortunes well before many of these businesses have reached the public markets.

The Only Heir Is Also the Richest

For all the new wealth being created in technology, the richest person in the ranking represents a much older source of billionaire wealth.

Name	Source of Wealth:	Self-made or inherited
Steven Hao	Cognition	Self-made
Palmer Luckey	Anduril	Self-made
Sam McCandlish	Anthropic	Self-made
Jack Clark	Anthropic	Self-made
Greg Brockman	OpenAI	Self-made
Edwin Chen	Surge AI	Self-made
Daniela Amodei	Anthropic	Self-made
Tom Brown	Anthropic	Self-made
Vlad Tenev	Robinhood	Self-made
Lukas Walton	Walmart	Inherited

At 39, Lukas Walton has an estimated net worth of $44.4 billion stemming from retail giant Walmart. He is the only person among the 10 whose fortune is identified as inherited.

His fortune also stands above every individual AI billionaire shown.

The ranking captures two very different paths to extreme wealth within the same generation: a decades-old retail fortune at the top and a wave of newer fortunes built around some of the world’s fastest-growing technology companies.