What Happens If AI’s Trillion-Dollar Bet Falls Short?

A surge in artificial intelligence investment is supporting growth, boosting demand for chips and data infrastructure, and lifting market sentiment. However, the Bank for International Settlements warns that the same boom could become a financial vulnerability if supply bottlenecks, overinvestment or disappointing returns trigger a sharp reversal.

Artificial intelligence (AI) is becoming a macroeconomic force, lifting investment, boosting chip demand, driving data centre construction and shaping financial markets. However, the same surge that has made AI one of the strongest engines of recent market optimism is now raising a harder question: what happens if the returns do not arrive fast enough?

The Bank for International Settlements (BIS), in its Annual Economic Report June 2026, places the AI boom at the centre of a wider debate over growth, financial stability and market risk. The report notes that AI-related investment helped support global growth and financial conditions, but also warns that the sustainability of that investment is now one of the major pressure points facing the global economy.

The AI Boom Has Moved Beyond Silicon Valley

The current AI surge is not limited to software companies or start-ups. It is reshaping demand for semiconductors, cloud computing, electricity, data centres, grid equipment and digital infrastructure. BIS notes that capital expenditure in semiconductor purchases, data centre construction and power infrastructure surged in the United States, driven by large hyperscalers. The spending supported aggregate investment in the US and created spillovers for Asia through AI-linked supply chains, including semiconductors, data storage and digital infrastructure.

A technology cycle that began with excitement over generative AI is increasingly tied to growth forecasts, corporate earnings, credit markets and energy demand. Strong profits among US chipmakers reflect the scale of demand for AI technologies, but the investment wave is also creating shortages and bottlenecks in the physical systems needed to run AI at scale.

For bullish investors, the current spending cycle looks like the beginning of a long technology buildout. AI could raise productivity, reshape business processes and create new demand for hardware, software, cloud infrastructure and energy systems. The BIS also acknowledges that AI has the potential to raise productivity significantly over the coming decade, with task-level studies reporting large efficiency gains, although aggregate productivity estimates remain more conservative.

The Trillion-Dollar Bet Is Running Into Supply Limits

According to the BIS report, the five largest hyperscalers are set to spend more than US$1 trillion on AI-related capital expenditure from 2025 through 2026. These commitments are outpacing earnings and free cash flow, pushing some firms to issue debt to raise additional financing.

Infrastructure-heavy technology transitions often require large upfront investment, but the risk rises when spending decisions are shaped by a winner-takes-most race, where firms believe only a few players will dominate future markets. BIS warns that intense competition may lead firms to overcommit resources to projects whose returns remain uncertain. If AI payoffs disappoint, the capital spending boom could turn into a prolonged investment bust.

Supply bottlenecks make that risk more complicated. The AI buildout is already facing constraints in electricity, advanced semiconductors and grid equipment. Fast-growing demand for computing power is pressuring electricity prices and input costs, with possible spillovers to inflation. Shortages may also encourage companies to lock in future capacity through long-term contracts, increasing their exposure if demand later falls short.

This is the paradox of the AI investment race: shortages can make companies spend even more aggressively, not less. It may help secure capacity in the near term, but it can also increase the cost of a correction if demand, pricing or productivity gains fail to match expectations.

The Bubble Risk Is Really a Financial Stability Risk

BIS warns that AI-linked optimism is becoming embedded in wider financial conditions. AI optimism helped ease financial conditions in 2025, lifted global stocks and supported household consumption through wealth effects. It makes the downside more consequential. Equity valuations are elevated, especially for firms at the centre of AI development. BIS says implied long-term earnings growth for the largest corporations sits above recent historical benchmarks, while investor sentiment has played a major role in current valuations.

The risks extend into credit markets. If hyperscalers slow or halt aggressive capital spending, borrowers across the AI supply chain could struggle to replace revenue and service debt. BIS identifies fixed income markets as a vulnerability because of high volumes of debt issued by hyperscalers, AI labs and engineering, procurement and construction firms.

Opacity adds another layer. BIS describes a complex web of private arrangements linking hyperscalers, chipmakers and AI labs, including circular financing, where firms take equity stakes in AI labs or neocloud providers that then commit to buying chips or computing power. Data centre construction is also increasingly outsourced to third parties under long-dated contracts, with deal terms often poorly disclosed.

A sharp AI repricing could therefore spread beyond the technology sector. BIS warns that a reassessment of equity risk could lead to tighter credit conditions and, if triggered by higher rates or an AI bust, could create a corporate credit freeze with wider implications for investment.

The Real Test Is Whether AI Can Deliver Fast Enough

The key issue is whether the financial, energy and industrial systems being built around AI are scaling at a pace that future returns can support. If AI delivers broad productivity gains, today's investment may be seen as the early infrastructure phase of a major technological transition. If productivity gains remain narrow, delayed or uneven, the same spending could leave companies, lenders and investors exposed to a painful correction.

There is also a labour market dimension. BIS warns that as more capable AI tools spread across tasks and occupations, labour displacement could intensify. It remains uncertain whether AI will create enough new jobs, or expand demand for existing ones, to offset displacement.

What matters next is whether the AI boom can prove its economics. Investors and regulators will be watching hyperscaler spending, chip supply, grid capacity, data centre financing, private credit exposure, valuations, credit spreads and hard evidence of productivity gains. They will also need to track whether leverage, opaque deal structures, circular financing and frontier-AI cyber risks are building faster than safeguards.