The next energy crisis is in the supply chain, not the oil field

The report covers more than 140 countries, 77 demand segments and 76 energy categories. The methodology shift is itself a signal: the energy question is moving from “which source” to “can we secure it reliably”.

1. Demand is not a trade-off—it is additive

In 2025, oil, gas, coal and renewables expanded simultaneously, lifting global energy consumption by 2.7%. Coal reached a new high, gas demand grew across major regions, oil consumption hit its highest level since 2019, and low-carbon investment exceeded fossil fuel investment for a second year.

Renewables did not “subtract” an equal share from fossil fuels; they added a second growth curve alongside the existing system. For materials, this means both markets press on the same supply table—the same capacity, the same certification resources and the same skilled labour.

2. The measuring stick has changed: from LCOE to system cost

Cheap generation does not mean cheap energy. The metric is shifting from standalone LCOE to system cost: generation, grids, storage, firm capacity, flexibility and integration.

Global transmission and distribution investment is expected to rise from USD 443 billion to USD 970 billion between 2025 and 2050. Pressure comes from both ends: system cost pushes owners and EPC contractors to squeeze materials, while grid delays magnify the cost of any material delay.

3. Supply security: from “can I buy it” to system architecture

The report warns that the next energy crisis is more likely to come from equipment, labour, batteries or critical minerals than from oil shortages. China accounts for about 80% of global solar module output, 75% of battery cell manufacturing capacity, more than 70% of refined lithium, cobalt and graphite, and about 90% of the rare earths needed for wind turbine permanent magnets.

In project language, “structural driver” means specifications will be continuously rewritten for de-risking: multi-source and non-single-origin clauses, more AVL second-source requests, and substitution as a supply-security imperative rather than a cost tactic.

Three questions test whether “diversified supply” is real: how many approvable sources exist today; how many weeks does a new source need from sample to approval; and which performance boundaries change when the source changes?

4. AI: the biggest demand variable, and a different bill of materials

Under the “continued momentum” scenario, data-centre electricity demand grows at a 24% CAGR to 2030. Data centres are not IT projects; they are heavy engineering: cooling water systems, firestop, cable trays and distribution, UPS and generator packages, and large volumes of structural steel.

There is also a mismatch: drawings get faster, certification does not. Design iterations shrink from two weeks to three days and the material list changes three times; approval of a new material still takes months. Design speed has overtaken compliance speed.

5. The grid is the ultimate limit—and the test of lead times

A new grid takes 5–15 years from planning to completion; key equipment procurement takes 2–4 years; offshore wind approval averages 9 years. Together they show that project schedule is no longer decided on site—it is locked when long-lead items are fixed.

Long-lead items must be the first schedule, not the third procurement table. Grid connection conditions must be clarified at feasibility stage. Offshore material strategies must be designed across the approval cycle, not just the supply cycle.

6. Three actions no scenario overturns

The report models five scenarios with very different outcomes, but only one line matters for materials: none of them reduces material demand. What matters is the capability every scenario needs.

First, build a certification map by category: approvable sources, missing documents and time to close. Second, move long-lead items into FEED. Third, prepare an “equivalence boundary” note for each key category: what can change, what cannot, and which parameters are only close rather than equivalent.