Method — how much is real data. Every year 2000–2025. Twelve drivers now carry a real annual per-country split — steel (worldsteel, per-country 2001–2025), vehicles & EVs (OICA + IEA), aluminium, cement, drilling (USGS / Baker Hughes), and electricity, population, nuclear, solar, wind, fertilizer (Ember/OWID/FAOSTAT). Three more are level-real — the world total follows a real annual curve while the country split stays a benchmark estimate (“trend real, split estimated”): aerospace (Airbus + Boeing deliveries), semiconductors (WSTS/SIA world sales; country split from SEMI 2025 materials data, Taiwan-led). Lead is now real too (USGS refined-lead production by country, 2004–2022), and glass is proxied by soda-ash production (USGS, 2004–2024) — soda ash being glass’s main raw material — so it is flagged proxy. That leaves only aerospace and semiconductors as level-real. Each material carries a “real” % = the share of its demand from real-split drivers: the steel-, EV-, vehicle-, aluminium-, cement-, drilling-, lead- and glass-driven metals are ~100%; the semiconductor/aerospace specialty metals (gallium, germanium, tantalum, hafnium…) show a real trend but keep an estimated country split. Levels are pinned so 2023 = the calibrated value. Read the movement as the finding; it does not model thrifting or substitution.
Cells are estimated tonnes of contained material consumed in the selected year, shaded within each column. Drag the slider or tap a year. Click a header to sort.
What moves, and why it matters
Two readings carry the page. China across the board: in 2000 the consumption leaders were spread across the industrialised world; by 2023 China leads almost every column, because its steel, power and construction activity outgrew everyone’s — the country that refines the world’s materials also came to use most of them. A new demand category: lithium, cobalt, graphite and rare-earth magnets barely register in 2000 (no EVs, little wind) and become major columns by 2023 — you are watching an entire material economy be born. Both are demand stories, and demand is exactly what trade and production maps leave out.
Method & caveats live on the consumption page; this view feeds the same production-constrained anchor. Reproducible: build_consumption_series.py → consumption_series.json.