The investment boom around artificial intelligence, which served as one of the main drivers of the global economy in 2025, is itself turning into a source of serious macro-financial risks. This conclusion was reached by experts from the Bank for International Settlements (BIS) in their annual report published at the end of June.
The analysts' focus is on debt financing of AI infrastructure, inflated company valuations, rapid growth in private lending, and opaque deals between hyperscalers (operators of large data centers), chip manufacturers, and AI labs.
Four Points of Tension
The BIS notes that in 2025, the global economy showed surprising resilience to tariff and geopolitical shocks. A key role was played by optimism around AI, which supported capital expenditures, semiconductor trade, and soft financial conditions. However, by 2026, the picture changed. Analysts identified four main risks: the threat of sustained inflation, doubts about the sustainability of AI investments themselves, growing financial vulnerabilities, and deteriorating fiscal positions of states.
"Optimism around AI may not persist, despite the promise of future productivity growth," the document emphasizes. According to the BIS, the current growth in capital expenditure risks being unsustainable if production hits bottlenecks: shortages of electricity, advanced semiconductors, and network equipment.
Trillion-Dollar Investments in Question
The scale of the debt overhang is particularly alarming. According to the BIS, the total capital expenditures of the five largest hyperscalers on AI infrastructure in 2025–2026 will exceed $1 trillion. The problem is that these commitments already outpace the profits and free cash flow of several companies, forcing them to actively raise debt.
"Disappointment in returns could trigger a sudden pullback in financing and turn the capex boom into a prolonged investment slump with potential consequences for financial conditions," analysts warn.
The BIS draws direct historical parallels: the current cycle is compared to the canal mania of the 1830s, the railway fever of the 1840s, the electrification boom of the late 1920s, and the dot-com crash of the late 1990s. In all these cases, a real technological breakthrough attracted significantly more capital than commercial results later justified.
Circular Financing and Hidden Risks
Another alarming signal is opaque private deals within the sector. This refers to so-called circular financing, where hyperscalers or chip manufacturers acquire stakes in AI labs, which in turn take on multi-year commitments to purchase chips or computing power. "The terms of such deals are usually poorly disclosed, creating a risk of rehypothecation of the same asset," the report notes.
Not only the giants themselves may be vulnerable, but also credit markets, private lending, and contractors building data centers. If hyperscalers slow down aggressive capex, borrowers across the supply chain will lose the revenue needed to service their debt.
Furthermore, the BIS reminds that a correction in the US, where stocks account for about 64% of the global MSCI index, could quickly spread worldwide. A revaluation of AI companies would directly hit household wealth, consumption, and global financial conditions.
Simultaneously, the memory market demonstrates the scale of demand and, at the same time, signs of strain. Micron reported record revenue of $41.46 billion for the third quarter of 2026, signing 16 strategic contracts worth about $100 billion (RPO). At the same time, the company's CEO admitted: "Right now, we have no clarity on when memory chip supply will be able to catch up with growing demand."
The rise in chip prices has already become the subject of a collective antitrust lawsuit against Samsung, SK Hynix, and Micron. The plaintiffs allege that the three largest DRAM manufacturers coordinated supply restrictions and artificially inflated prices, shifting priorities towards more expensive AI products, including HBM memory.
My comment: The situation described by the BIS is a classic example of how technological euphoria can lead the financial system into a trap. AI is undoubtedly a breakthrough, but the market is already pricing in future revenues that may not materialize within the stated timeframe. Investors should pay close attention to the debt burden of hyperscalers and the transparency of their off-balance-sheet obligations. History teaches us that booms fueled by easy money and excessive optimism rarely end in a soft landing.