AI Investment Boom Raises Financial Stability Concerns, BIS Chief Warns

The rapid expansion of artificial intelligence is creating emerging risks for global financial stability, Bank for International Settlements (BIS) General Manager Pablo Hernandez de Cos said on September 10, 2026, warning that the scale of investment in AI infrastructure is becoming large enough to influence broader economic conditions.

Speaking at a conference hosted by India’s central bank, Hernandez de Cos said AI was transforming demand, supply and financial markets at the same time. While the technology does not alter the mandates of central banks, its effects could make economic conditions more difficult for policymakers to assess.

The BIS estimates that the world’s five largest technology companies will spend more than $1 trillion on AI during 2025 and 2026. Industry forecasts indicate that worldwide AI investment could rise from roughly $500 billion currently to as much as $4 trillion by 2030.

Hernandez de Cos said the potential benefits of AI were substantial, but stressed that its longer-term economic impact would depend on government policies, investment in infrastructure and skills, and how broadly the gains from the technology are distributed.

At the same time, he called for closer examination of how the AI expansion is being financed. Increasingly, AI-related investment is being supported by debt and private credit rather than corporate earnings. Hernandez de Cos said this was a concern because significant portions of the financing network remain “opaque and interconnected.”

Such financial structures could become more vulnerable if companies fail to generate the profits investors currently expect.

## AI Reshapes Global Trade and Productivity

The AI boom is also affecting international trade. Countries with strong links to the technology supply chain, including South Korea, Singapore, Malaysia and Taiwan, have benefited from higher export prices for AI chips and related equipment.

Beyond trade, AI has demonstrated the potential to improve productivity. Hernandez de Cos cited studies showing that generative AI can increase productivity by between 10% and 65% for particular tasks. The strongest gains have been reported in areas such as coding, consulting and professional writing.

However, the BIS chief said the more important economic question was whether these improvements at the individual task level would translate into stronger productivity across entire economies.

Current estimates indicate that AI could increase total factor productivity growth by around half a percentage point annually. The eventual impact will depend on how quickly businesses adopt the technology and how effectively workers and capital are shifted toward more productive uses.

Advanced economies are expected to see benefits earlier because they have larger service sectors and are generally better prepared to deploy AI technologies. Emerging economies face more diverse outcomes.

India, however, could be positioned to benefit significantly. Hernandez de Cos said the country had a “genuine opportunity” to reduce the productivity gap with more advanced economies, supported in part by its digital public infrastructure.

## Employment Risks and Market Vulnerabilities

The economic effects of AI are not limited to productivity gains. Hernandez de Cos also warned that the technology could replace workers performing routine cognitive tasks.

So far, job losses linked directly to AI have remained limited, but early signs of disruption are appearing in customer service, programming and administrative work. This increases the importance of retraining and reskilling workers as businesses adopt AI more widely.

Financial markets present another area of concern. Hernandez de Cos highlighted elevated valuations, concentration among major technology companies and complex financing arrangements as potential sources of vulnerability.

If corporate profits fail to meet the expectations built into current valuations, those weaknesses could become more pronounced. He cautioned that the outcome was not predetermined, but said the scale and speed of today’s AI investment cycle justified careful attention.

Hernandez de Cos compared the situation with earlier periods of rapid economic expansion, including the railway boom and the dotcom surge. The comparison was not intended to suggest that AI would necessarily produce the same outcome, but rather to underline the risks that can emerge when investment grows rapidly alongside expectations of large future commercial returns.

“The promise of AI is real,” Hernandez de Cos said, while emphasizing that policymakers and investors must remain alert to the financial risks accompanying its extraordinary expansion.