Blog

The Minerals Beneath the Machine: Who Controls the Data That Controls the Supply Chain?

Jan Gunash · Co-Founder & CGO · March 23, 2026

The best questions are not always the comfortable ones. Like: when we say AI is accelerating the clean energy transition, what is that transition actually built from? When we say the supply chain is the problem, who is measuring it, where does the data come from, and what happens when there is almost none?

These are not rhetorical questions. They have empirical answers, and those answers sit quietly underneath some of the most consequential infrastructure decisions being made right now.

A Physical Problem in a Digital Age

Something has shifted in how we talk about AI and energy. For years, the conversation was about software, algorithms, and compute. Now, it is becoming physical. The materials that make AI run, copper for cooling and power distribution, lithium for backup storage, rare earth elements for the magnets in every motor and transformer, gallium and germanium for semiconductors, are mined, processed, and refined across supply chains that span dozens of countries and carry decades of geopolitical complexity.

The numbers are not small. Demand for critical minerals used to produce clean energy technology could be more than 3.4 times current levels by 2040 (source: IEA via World Economic Forum). AI data centers are themselves becoming one of the most mineral-intensive components of the digital economy, with copper, aluminum, and rare earth elements now consumed at scale by the networking, cooling, and power distribution systems inside every hyperscale facility (source: Global X ETFs). Copper and lithium are among the materials facing potential double-digit supply shortfalls by 2035.

The energy transition and the AI buildout are not competing for these minerals. They are competing together, against a supply chain that was never designed to move this fast.

The Data Problem Nobody Is Naming

Here is the thing about critical minerals that almost no one in the AI and energy conversation is talking about: the data infrastructure for these supply chains barely exists.

Unlike oil and gas markets, where monthly production figures for most countries are available within weeks, production data for critical minerals can lag by a year or more (source: U.S. Energy Information Administration). The reasons are structural: underreporting by small-scale operators, proprietary commercial data held by dominant processors, inconsistent government reporting standards across jurisdictions, and the simple fact that many critical minerals are extracted as byproducts of other mining operations and are not tracked separately at all.

The U.S. Energy Information Administration recently acknowledged this directly, launching a new Energy Minerals Observatory to begin building the kind of transparent, standardized data infrastructure that oil markets have had for decades. The framing of the announcement was telling: this is new. It should not be.

Price benchmarks, which underpin supply contracts and support risk management across energy markets, are largely absent for critical minerals. Reserves data is scarce and often outdated. Recycling production, an increasingly important secondary supply source, is systematically undercounted because standard mining accounting practices exclude materials that amount to less than one percent of a mine's total production value (source: EIA). The result is a market where the signals that investors, governments, and manufacturers need to plan around are missing, delayed, or locked inside proprietary systems.

This is not a technical problem. It is a choice about what gets measured and who benefits from opacity.

A Model Built on Gaps

The parallel to AI weather forecasting is not a stretch. In the March 16 edition of this newsletter, we examined how AI climate models trained on data-sparse regions inherit the biases of the physics models that filled those gaps, creating a model-of-a-model problem where uncertainty compounds in exactly the places it is already highest.

The same dynamic is now visible in critical minerals. Governments and companies making investment decisions about mines, processing facilities, and supply chain diversification are working from data that is incomplete, delayed, and geographically uneven. The regions that most need transparent market signals, producer countries across sub-Saharan Africa, Central Asia, and Latin America, are often the least represented in the datasets that global pricing and procurement decisions are built on.

DARPA recognized this explicitly enough to fund a dedicated program. The Open Price Exploration for National Security initiative is using AI and probabilistic modeling to try to make critical minerals supply and demand legible in real time, because the existing data infrastructure is too fragmented and opaque to support the decisions that national security and energy transition planning require (source: Charles River Analytics / DARPA OPEN). The fact that this requires a defense research program to build from scratch is itself a signal about how far behind the data infrastructure has fallen.

Traceability as the New Frontier

The data problem does not stop at pricing and reserves. It extends into provenance, knowing not just how much of a mineral exists, but where it came from, under what conditions, and through whose hands it passed.

This matters for three converging reasons. Regulatory frameworks in the EU and U.S. are tightening requirements for responsible sourcing disclosure. Defense procurement rules increasingly require verified chain of custody for minerals used in advanced systems. And ESG-linked capital is flowing preferentially toward companies that can demonstrate traceable, compliant supply chains, making traceability a financial differentiator, not just a compliance burden (source: AltaScient Critical Minerals Insights).

The response is emerging but fragmented. Blockchain-based provenance tracking is being deployed for rare earth elements, creating digital records of chain of custody from mining through refining (source: ReElement Technologies). AI-assisted assessment programs at the U.S. Geological Survey are attempting to automate and accelerate mineral evaluation workflows that currently take years. The Department of Energy's CM²US initiative is building an AI-driven digital map of the full critical minerals supply chain, from mining to market, to identify bottlenecks and validate processing technologies earlier in the development cycle (source: U.S. Department of Energy).

Each of these is a genuine step forward. But they are emerging in parallel, without common standards, shared data formats, or interoperability. That is the same fragmentation problem that has limited the value of biodiversity and energy data before them. Traceability without standardization is just a more expensive form of opacity.

The Geopolitical Dimension the Data Makes Visible

There is a reason governments are treating this data problem as a national security issue rather than a market efficiency question. China accounts for 69% of rare earth mining, 91% of refining, and 94% of rare earth magnet manufacturing globally (source: Benchmark Mineral Intelligence via Global X ETFs). When supply chain data is opaque and concentrated, that concentration becomes leverage. And leverage that cannot be measured cannot be managed.

The U.S. recently outlined “Project Vault”, a roughly $12 billion strategic mineral reserve, and signed critical minerals frameworks with eleven countries in a single day at the 2026 Critical Minerals Ministerial (source: U.S. Department of State). The EU's Critical Raw Materials Act is attempting to mandate strategic project selection and supply chain diversification. Governments across South Korea, India, Brazil, and the Gulf states are accelerating their own mineral diplomacy. All of this activity is happening against a backdrop of data that is, at best, months old and, at worst, simply absent.

The arms race for mineral security is being run without reliable maps. That is not hyperbole. It is the data reality.

The Path Forward

The root cause is not the geopolitics. It is the absence of shared data infrastructure that would make transparent, informed decision-making possible at all.

What needs to happen mirrors what the energy sector has been learning, slowly, about clean energy data. Common reporting standards for mineral production, reserves, and processing capacity, the equivalent of what NOAA provides for weather or what the IEA provides for oil, would transform the quality of investment and policy decisions across the entire value chain. Open, regularly updated price benchmarks would reduce the information asymmetry that currently advantages large incumbents and opaque intermediaries. And recycling and secondary supply data needs to be treated as real production data, not as an afterthought.

The good news is that the tools exist. AI-assisted geological assessment, satellite-based monitoring of mining operations, and blockchain-powered provenance tracking are all operational today in at least limited form. The gap is not technology. It is the governance, standards, and institutional will to build the shared data infrastructure that makes these tools useful beyond individual proprietary deployments.

The minerals beneath the AI economy are physical. Their supply chains are real. And the data systems that should make those supply chains legible, plannable, and accountable are still years behind where they need to be.

The algorithm learns from what it is given. Right now, what it is given is not enough. That is not a technical argument. It is a choice about what kind of energy and technology future we are building, and whether the foundations of that future are visible to the people who need to act on them.

Sources:

Want to contribute your thoughts or findings in data and AI with climate, energy, or nature? Message Jan Gunash on LinkedIn or send an email to jgunash@projectcoeus.com.