Brinley Institute
← Early Consequence Intelligence
Future Pillar

Material Trace

Phase A · Doctrine Registered·APEX RATIFIED 2026-06-17

The physical supply chain behind AI infrastructure — from chip fabrication to e-waste — tracked as public consequence in the making.

Every data center is the visible end of a supply chain that begins in semiconductor fabs, runs through rare-earth extraction, and ends in e-waste streams that rarely surface in the public record. Material Trace is the proposed seventh observatory pillar: tracking the physical-layer consequence of AI infrastructure build-out.

Public-record sources exist for most chain segments. Where they do, BEI surfaces the verbatim finding with full attribution. Where the public record infrastructure itself is the gap, BEI names that gap.

Supply chain segments

Each segment surfaces public-record evidence when a sanctioned, verified source exists. Segments without wired data show their disclosure status.

01

Chip fabrication

Semiconductor design, fab, and packaging — geographic concentration, export controls, energy intensity

Curation pending

SEC EDGAR, EPA Toxics Release Inventory, and ITAR/EAR export control disclosures are the sanctioned sources for this segment. Curated anchor entries are pending REESE verification against primary documents.

02

Server assembly

ODM manufacturing, component sourcing, lead times, and procurement concentration

Curation pending

SEC supply chain risk disclosures (SCDP) and publicly filed corporate ESG reports are the sanctioned sources. Curated anchor entries pending REESE verification. Customs import data surfaces at commodity/HS-code level only — not individual shipment records.

03

Cooling systems

Air cooling, liquid cooling, water consumption per MW, and regional water stress

Data wired

A large data center can use up to approximately 5 million gallons of water per day for cooling purposes. Water Use Effectiveness (WUE), measured in liters per kilowatt-hour, varies significantly by cooling method and climate.

Tier 3·Environmental and Energy Study Institute — "Data Centers and Water Consumption" (2024)·Primary source →

Google's water use at its data center campus in The Dalles, Oregon was disclosed only after a public-records lawsuit. The water consumption figures were legally public but practically hidden — assembled into meaning only through litigation, not routine disclosure.

Tier 3·The Oregonian / Associated Press — Google data-center water use disclosed after public-records lawsuit, The Dalles (2022)·Primary source →

Cooling accounts for a substantial share of data center water consumption. Air-cooled facilities typically consume water only indirectly (through grid generation); liquid-cooled and evaporative cooling systems consume water directly on-site. As AI workloads drive higher rack densities, direct liquid cooling is expanding, shifting water consumption from grid-distributed to facility-local.

Tier 2·International Energy Agency — Energy and AI: World Energy Outlook Special Report (2025)·Primary source →

Water stress by data center market

Phoenix

USA

Critical

Acute desert water scarcity; cooling-water sourcing is the central question

Dubai / Abu Dhabi

UAE

Critical

Desert siting; desalination-linked water/energy nexus

Santiago

Chile

Critical

Google’s campus water permit was revoked/revised amid drought

Meta · El Paso, TX

USA

Critical

Acute Chihuahuan-desert water scarcity

Meta · Mesa, AZ

USA

Critical

Acute Arizona desert water scarcity

Silicon Valley (Santa Clara)

USA

High

Arid California water context

Water risk · DCES dimension · DCES band assessments are qualitative, not audited scores. Source: BEI observatory data layer.

04

Rare earth inputs

Neodymium, dysprosium, and other critical minerals in motors, power systems, and magnets

Data wired

Global mine production was estimated to have increased to 390,000 tons of REO equivalent largely owing to increased mining and processing in China, Nigeria, and Thailand.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Rare Earths (Cordier)·Primary source →

Import Sources (2020–23) — rare-earth compounds and metals: China, 70%; Malaysia, 13%; Japan, 6%; Estonia, 5%; and other, 6%. Compounds and metals imported from Estonia, Japan, and Malaysia were derived from mineral concentrates and chemical intermediates produced in Australia, China, and elsewhere.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Rare Earths (Cordier)·Primary source →

Net import reliance as a percentage of apparent consumption, compounds and metals: 80 [2024, estimated]. U.S. domestic mine production of mineral concentrates: 45,000 metric tons REO (2024, estimated). An estimated 45,000 tons of REO in mineral concentrates were mined and processed domestically in 2024.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Rare Earths (Cordier)·Primary source →
05

Battery minerals

Cobalt, lithium, and nickel supply chains — extraction geography, labor conditions, and ESG disclosure

Data wired

The increase in mine production was mainly in Congo (Kinshasa), the world's leading source of mined cobalt, which accounted for an estimated 76% of world cobalt mine production, followed by Indonesia, which accounted for 10%. World total cobalt mine production: 290,000 metric tons (2024, estimated).

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Cobalt (Ewing)·Primary source →

China was the world's leading producer of refined cobalt and increased metal refining capacity throughout the year. China was the world's leading consumer of cobalt, with the majority used by the lithium-ion battery industry. U.S. net import reliance as a percentage of apparent consumption: 76 [2024, estimated].

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Cobalt (Ewing)·Primary source →

Excluding U.S. production, worldwide lithium production in 2024 increased by 18% to approximately 240,000 tons from 204,000 tons in 2023 in response to strong demand from the lithium-ion battery market. Global consumption of lithium in 2024 was estimated to be 220,000 tons, a 29% increase from revised consumption of 170,000 tons in 2023.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Lithium (Jaskula)·Primary source →

Although lithium uses vary by location, global end uses were estimated as follows: batteries, 87%; ceramics and glass, 5%; lubricating greases, 2%; air treatment, 1%; continuous casting mold flux powders, 1%; medical, 1%; and other uses, 3%. Lithium consumption for batteries increased significantly owing to the use of rechargeable lithium batteries in the growing market for electric vehicles (EVs), portable electronic devices, electric tools, and energy grid storage applications.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Lithium (Jaskula)·Primary source →

Import Sources (2020–23): Chile, 50%; Argentina, 47%; and other, 3%. Net import reliance as a percentage of apparent consumption: greater than 50 [2024, estimated]. World total mine production: 240,000 metric tons lithium content (2024, estimated). World reserves: 30,000,000 metric tons.

Tier 1·U.S. Geological Survey — Mineral Commodity Summaries, January 2025: Lithium (Jaskula)·Primary source →
06

E-waste and refresh

Replacement cycles, decommissioning practices, and end-of-life disposal — publicly disclosed or not

Structural inaccessibility

No unified public-record feed for e-waste disposition exists with permissible aggregation rights. Fragmented state-level programs, no national API, no sanctioned structured source. This is a public records infrastructure gap, not a BEI gap. The absence of a public e-waste tracking feed is itself a material fact about the supply chain's accountability infrastructure.

ECI connection

Material Trace is registered in the Early Consequence Intelligence doctrine as a future module — the physical-chain consequence that is knowable before a facility's full lifecycle impact becomes visible. It does not belong inside any of the six current pillars; it is a seventh domain when the data architecture is ready.

→ Early Consequence Intelligence — module vocabulary

BEI does not surface supply-chain data for this pillar until a sanctioned, ToS-compliant public-record feed is verified for each segment. No bot-wall evasion. No fabricated data. Segments without verified sources display an honest empty state with their disclosure status.

EASTON-LMC-2026-06-17-ECI-DOCTRINE · APEX RATIFIED 2026-06-17 · PAT-035 / Humanity Ledger §2.6 · Zero Protocol · CALLAN wire-time clearance CLRR-MATERIAL-TRACE-TOS-2026-06-18