The Metal Quietly Driving Up the Cost of AI

Tungsten prices have risen by more than 550% in parts of the global market since China tightened export controls, reaching record levels and sending shockwaves through the semiconductor supply chain. The surge is exposing a vulnerability few policymakers or technology companies have seriously discussed: artificial intelligence ultimately depends on raw materials that remain strategically concentrated.

The immediate consequence is rising production costs for advanced semiconductors. The broader consequence may prove far more significant: AI is becoming constrained not only by compute capacity, but increasingly by access to critical minerals.

The Reshoring Illusion

🟦 Could the next bottleneck for artificial intelligence begin not inside the fab, but inside the mine?

The United States and Europe are investing hundreds of billions of dollars in new semiconductor fabrication plants. Industrial policy increasingly focuses on rebuilding manufacturing capacity and reducing technological dependence.

Yet fabrication plants cannot manufacture chips without secure access to the materials required for production. Semiconductor sovereignty therefore extends beyond factories alone.

China’s Quiet Leverage

🟦 Can semiconductor sovereignty ever exist without mineral sovereignty?

Much of the geopolitical debate has centred on export controls affecting advanced AI chips and lithography systems.

China has increasingly focused on another layer of the value chain.

By controlling significant portions of the global supply of critical minerals such as tungsten, Beijing influences technologies long before manufacturing even begins.

The Physical Reality of AI

🟦 Has the geopolitical debate focused too heavily on chips while overlooking the raw materials from which those chips are built?

Artificial intelligence is often discussed as software. Its foundations are anything but virtual.

Every AI accelerator depends upon mining, refining, metallurgy, manufacturing, electricity and logistics before the first computation is ever performed.

The cloud begins underground.

The Economics of Scarcity

🟦 Have we become so focused on digital intelligence that we overlook the physical systems making it possible?

For decades, the semiconductor industry optimised for efficiency through globalisation.

Critical minerals introduce another equation.

Scarcity raises prices. Higher prices influence investment. Investment determines industrial capacity.

The economics of AI may increasingly be shaped by commodity markets rather than software alone.

🟦 Could commodity markets become as strategically important as semiconductor design itself?

Beyond Tungsten

Tungsten is unlikely to remain an isolated case.

Gallium. Germanium. Graphite. Rare earth elements.

Increasingly, the competition surrounding advanced technologies is extending deeper into the geological foundations of industrial society.

SIGNIFY

🟦 Are we entering an era in which the race for AI is increasingly becoming a race for materials?

For years, technological leadership was measured through better software, faster processors and larger data centres. The tungsten crisis points towards another reality.

The future of artificial intelligence may no longer be determined solely by who develops the most capable models.

It may increasingly depend upon who controls the materials from which those models ultimately emerge.

The future of AI may not only be written in code. It may first be dug out of the ground.

---

Credit

Image: AI-generated illustration / Altair Media

Caption

A raw tungsten ore beside a semiconductor wafer illustrates the growing importance of critical minerals in the AI economy. As tungsten prices reach record levels, the race for artificial intelligence increasingly begins long before the chip factory—inside the mine.