The investment euphoria around artificial intelligence, which served as a key driver of the global economy in 2025, is itself beginning to generate macro-financial threats. The Bank for International Settlements (BIS) analysis for 2026 has identified alarming signals related to debt financing of AI infrastructure, inflated market valuations, and a rise in opaque deals between hyperscalers, chip manufacturers, and labs.

Four Points of Tension in 2026

In 2025, the global economy demonstrated resilience to tariff and geopolitical shocks, largely due to optimism surrounding AI. However, already in 2026, BIS analysts highlighted four key risk areas: the threat of persistent inflation, doubts about the long-term sustainability of AI investments, growing financial vulnerabilities, and deteriorating fiscal positions. Optimism around AI may not persist, despite the promise of future productivity growth — this conclusion sounds like a warning for the market.

Capital Expenditure: A Giant $1 Trillion Bubble

According to BIS estimates, 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 some companies, forcing them to actively raise debt financing. Disappointment in returns could trigger a sudden withdrawal of funding and turn the capex boom into a prolonged investment downturn with potential consequences for financial conditions.

Historical parallels are obvious: the BIS compares the current cycle to the canal mania of the 1830s, the railway mania of the 1840s, the electrification euphoria of the 1920s, and the dot-com bubble of the 1990s. The common feature is a real technological breakthrough that attracted more capital than commercial results could later justify.

Circular Financing and Hidden Risks

Particularly alarming is the opacity of private deals within the sector. This refers to circular financing, where hyperscalers or chip manufacturers acquire stakes in AI labs and cloud providers, which in turn take on multi-year commitments to purchase chips or computing resources. The terms of such deals are usually poorly disclosed, creating a risk of rehypothecation of the same asset. If hyperscalers slow down aggressive capex, borrowers across the supply chain may lose the revenue needed to service their debt.

Furthermore, a correction in the US, where stocks account for about 64% of the MSCI Global index, could quickly spread worldwide, affecting household wealth, consumption, and global financial conditions.

Microchips: Demand Outstrips Supply

A separate indicator of tension is the situation in the memory chip market. On June 24, Micron reported record revenue for the third quarter of 2026 — $41.46 billion. The company signed 16 strategic customer agreements with a total backlog of commitments (RPO) of approximately $100 billion. We currently have no clarity on when memory supply will be able to catch up with growing demand — admitted Micron CEO Sanjay Mehrotra. DRAM brought in a record $31.3 billion, or 76% of total revenue.

The rise in chip prices has already become the subject of a legal dispute: on June 25, a class-action antitrust lawsuit was filed in the Northern District of California against Samsung, SK Hynix, and Micron. The plaintiffs allege that the three largest DRAM manufacturers coordinated to restrict supply and raise prices, shifting priorities towards more expensive AI products, including HBM memory.

My expert conclusion: The AI boom of 2025 was undoubtedly a powerful economic driver, but we are now seeing classic signs of overheating. Debt financing, opaque deals, and the disconnect between capex and actual profits create ideal conditions for a correction. Investors should prepare for increased volatility, especially in the semiconductor and data center segments.