Blackstone and Pimco warn AI debt risks mount as spending heads toward $10 trillion

Financing the global artificial intelligence boom will require significantly more equity funding rather than relying too heavily on borrowing, according to top investment executives. Speaking at a finance summit in Hong Kong, leaders from Blackstone and Pimco warned that lenders face growing exposure risks as data centers and infrastructure expand rapidly.

Limits of borrowing for AI growth

Blackstone President Jon Gray explained that the enormous amounts of money needed for AI projects create concentration risks for debt investors. Lenders take on downside risk if projects fail, but they face capped upside because their returns are limited to interest and loan repayments. In contrast, equity investors can share in the long-term gains of corporate growth.

Gray noted that funding this infrastructure will require a mix of public markets, commercial banks, private capital, and internal cash from large cloud providers, also known as hyperscalers. Companies are also using structured deals to keep borrowing within safer bounds.

A massive funding requirement

Pimco Chief Executive Emmanuel Roman warned that the sheer scale of borrowing could overwhelm financial markets. Roman cited estimates that AI-related capital spending could hit $10 trillion over the next five years. For comparison, the entire global corporate bond market is worth around $20 trillion.

Around $550 billion in AI-related debt has already been issued this year. Investment bank Goldman Sachs separately estimated that data centers alone will demand about $8.5 trillion in capital spending over the coming five years.

Rising demand faces physical limits

Despite the financing hurdles, executives say demand for computing power continues to outrun available supply. Within Blackstone’s own portfolio companies, spending on AI models grew 20-fold over the past year for tasks such as coding, content creation, and customer service.

However, builders face practical bottlenecks beyond raising money. Gray pointed out that securing land, government permits, microchips, electrical power, and funding are all growing more difficult at the same time.