The AI Innovation Cycle: Scenarios and Signposts

The AI investment cycle is still in its breakthrough and speculative phases, with delayed payoff and inflationary boom currently the most likely outcomes.

• AI infrastructure spending is rising rapidly and could rank among history's largest investment waves, although growth may slow in 2027 because of permitting delays, power shortages, equipment constraints, labor shortages and financing pressure.
• Hyperscalers initially funded expansion through operating cash flow, but heavy capex has reduced free cash flow and pushed companies toward debt, equity, leases and structured financing; debt and equity issuance has risen sharply as yields and credit scrutiny increase.
• Adoption remains concentrated among early users. Ramp data covering 85,000 firms shows a wide spending gap between the top 1% of corporate adopters and the median company, while business token spending has recently leveled off.
• Four scenarios frame the cycle: a productivity boom with broad adoption and higher output per hour, a delayed payoff with slow monetization, an inflationary boom with strong nominal growth and higher real rates, and a deflationary bust caused by excess capacity and falling pricing.
• The Bloomberg AI Value Chain Equal Weight Index has fallen 22% so far this quarter, while U.S. midterm election uncertainty and opposition to data-center construction have increased the risk of slower investment and adoption.

Takeaway: Investors should monitor capex, financing conditions, adoption, utilization and monetization while maintaining diversification, selectivity and flexibility, since delayed returns or excess capacity could change market leadership and increase the probability of a deflationary bust.