Working Paper 08
Time to Power: The Financing Cost of Energy-Delayed Megaprojects
September 2026
Abstract
We assemble an audited dataset on 19 large capital-intensive projects (19 rows, 6 sectors: data centers, semiconductors, mining, LNG, battery manufacturing, industrial processing), nine of which were materially affected by energy-infrastructure constraints, and ask when the cost of carrying financing obligations on a delayed asset exceeds the cost of directly funding the missing power. A 12-mechanism delay-cost model, applied to a stylized 500 MW / $6.0B data center, puts a 12-month energy-driven delay at $857.6M in present value (14.3% of capex, $71.5M per month); the timing value of delayed revenue is the largest term at every horizon (about two-fifths of the total), while financing-only terms are 35% of the 12-month total, so counting only interest during delay understates cost by roughly two-thirds. Five independent energy supply-chain links each run 30 months or more in typical lead time (greenfield transmission 120 months, generator interconnection queues 55 months median, gas-turbine slots 5-7 years quoted, large transformers 30 months average with 48-60 month Tier-1 quotes, large-load interconnection 36 months typical). A wait-vs-solve model over 3-, 6-, 12-, 24-, and 36-month delays finds that no permanent alternative beats waiting at 12 months or less; reciprocating gas engines are the only alternative that beats waiting inside 60 months, from roughly 24-28 months of expected delay, supporting a rational capex premium up to $3,049/kW at 36 months (against $1,600/kW market capex) and a delivered-energy ceiling near $78/MWh. The data support a durable but narrow business thesis: willingness to pay for time-to-power is large (order $140/kW per month of acceleration) yet confined to delays beyond roughly two years, alternatives priced near grid power, and revenue-intense projects. We state the dashboard readings that would prove this wrong. All figures are nominal USD, base year 2026, unless stated.