Opaque “circular” financing deals among big artificial intelligence companies could pose “systemic spillover” risks to the global financial system, the International Monetary Fund has warned. 

The organisation said that growing interdependencies between companies across the AI and technology industries have increased the risk that an isolated shock could be amplified through the system. 

“Greater interconnectedness owing to such circular financing structures heightens concerns of systemic spillovers should an adverse shock affect even a single entity,” the IMF said in its global financial stability report. 

Business newsletter

The business editor’s exclusive analysis of all the latest financial and economic news.

Sign up with one click

Circular financing deals are arrangements where an investor provides funding to a company, and that funding is then used to purchase products from the initial investor. 

Since last year, these deals have become increasingly common, with start-ups including OpenAI and Anthropic receiving tens of billions in investment from companies such as Microsoft, Nvidia and Amazon, often alongside long-term agreements to spend heavily on cloud infrastructure and AI chips. Some analysts have warned that the deals mirrored developments during the dotcom bubble in the 1990s. 

The IMF warned that this has made it difficult to track the underlying fundamentals of the market, and could create ripple effects from shocks. 

“Such interconnectedness raises opacity risks. AI-related firms within the circle may simultaneously report inflated revenues, potentially detached from fundamentals, increasing the likelihood of stretched equity valuations and affecting market pricing more generally.

“The complex multidirectional channels of shock transmission between circle firms created due to the circular structure could result in nonlinear amplification of adverse shocks, which could propagate to the wider market given centrality of these particular firms in driving AI-related investment, at a time when concentration risk is historically elevated.”

Staff at the IMF attempted to estimate which firms in the AI supply chain were vulnerable from the sheer scale of spending on AI infrastructure, which is set to hit $1.4 trillion this year, according to Gartner, a research firm. 

While core “hyperscalers” and chipmakers such as Google, Microsoft and Nvidia had strong enough balance sheets to finance the spending, “less-systemic segments, such as energisers [power producers for data centres] and operators, appear to have balance sheets more sensitive to the cost of rapid expansion”. 

However, even the biggest hyperscalers, the term for the giant cloud computing and “platform” service providers such as Amazon, could come under pressure in the coming years. “Earnings and cash buffers of hyperscalers could prove insufficient given the estimated $3.4 trillion in AI-related capital expenditure through 2029, potentially creating balance sheet pressures in the coming years,” the report said. 

Hyperscalers have increasingly turned to debt to finance their capital spending. The IMF noted that financial markets have been willing to absorb the debt so far, suggesting that risks through this channel “remain contained” for the moment.