Five tech giants commit $1 trillion to AI capex as debt risks rise
The five largest technology firms will spend over $1 trillion on artificial intelligence capital expenditure between 2025 and 2026. Speaking at an RBI conference on September 10, 2026, the BIS highlighted rising financial stability risks from debt-financed tech investment.
The trillion-dollar compute race
Global AI investment is projected to expand from $500 billion to nearly $4 trillion by 2030, driven by heavy outlays on data centres, semiconductors and cloud capacity.
Upstream exporters have experienced sharp terms-of-trade gains; in South Korea, five tech firms generated 43 percent of national export earnings in early 2026.
Micro-level studies demonstrate productivity gains between 10 and 65 percent for cognitive tasks, yet median estimates place aggregate total factor productivity growth at roughly 0.5 percentage points annually.
Simultaneously, nearly 80 percent of surveyed firms report active plans for task automation and labour substitution, raising transition pressures across service sectors.
Circular debts and historical echoes
Central banks face heightened uncertainty in estimating potential output and the natural rate of interest.
The BIS identified circular financing—where hardware manufacturers invest equity in clients who buy their compute—alongside opaque private credit as key vulnerabilities.
Should expected returns disappoint, the current spending cycle risks a severe downturn comparable to the 1840s railway mania or the 1990s dotcom collapse, amplified by household equity wealth exposures.
Capex mania meets demand reality
Central banks rightly flag opaque circular financing loops and private credit exposures before balance sheets crack.
Yet theoretical demand bottleneck scenarios provide little practical help for near-term rate setting.
Regulators must prevent excessive tech leverage without choking genuine productivity growth.