Spending on artificial intelligence has surged to levels exceeding those seen during the late 1990s dotcom boom, according to Nathan Sheets, Citi’s chief economist. Speaking ahead of his keynote at the Citi Australia Conference, Sheets highlighted that AI investment currently accounts for approximately 1.5% of global GDP, roughly double the share that technology investment held during the comparable stage of the earlier tech cycle.
Sheets, a former official at the US Treasury and Federal Reserve who has observed multiple market cycles, attributes the sustained nature of the AI spending boom more to the businesses adopting the technology than to the technology providers themselves. He noted that many firms feel compelled to invest in AI to remain competitive, making demand for AI relatively insensitive to economic conditions or interest rates. This demand creates a feedback loop, with robust enterprise investment driving continued supply-side increases among AI producers, helping to bolster global economic resilience amid recent shocks such as tariffs in 2025 and rising oil prices in 2026.
Despite these large investments, Sheets cautions that significant productivity gains from AI have yet to materialize broadly. He described this dynamic as a “productivity J-curve,” common with disruptive, general-purpose technologies, where initial periods of heavy investment and adoption precede visible productivity improvements. While some sectors may be beginning to see benefits, he indicated that overall, measurable productivity growth remains limited at this stage.
The widespread adoption of AI is also contributing to an increasingly “K-shaped” economic landscape. Sheets pointed out a divergence between the AI-driven parts of the economy, which are experiencing strong growth and are less sensitive to interest rates, and sectors such as housing, real estate, and lower-income consumer markets, which remain weak. This split presents a challenge to the Federal Reserve as it attempts to manage inflation; raising interest rates to slow AI-related inflation risks further weakening other already fragile economic sectors without necessarily dampening AI-driven price pressures.
Sheets emphasized the importance of a strong US labor market as a key factor sustaining overall spending and economic stability. He suggested that any significant weakening in employment could exacerbate economic softness across both high- and lower-income groups.
Looking ahead, Sheets identified geopolitical developments, particularly the situation in Iran, as a significant near-term factor influencing US inflation and monetary policy decisions. A resolution of tensions there could allow the Fed to adopt a more measured approach. While a December interest rate increase is widely anticipated, future hikes will depend on inflation trajectories.
On the financial markets front, Sheets observed recent volatility in US Treasury yields, interpreting the fluctuations as a response to inflation data and equity market shifts. He also highlighted a structural risk in credit markets linked to AI-related borrowers, noting that credit investors face limited upside but considerable downside risk compared to equity holders, making credit markets especially sensitive to changing risk perceptions.
A broader concern raised by Sheets is a looming global savings shortage. He projects that US Treasury debt issuance could reach roughly $2.5 trillion by 2030, with additional sovereign borrowing and AI investment demands potentially pushing the total needed funding to around $5 trillion—far exceeding past levels. This contrasts with the “global savings glut” previously discussed in economic literature, now shifting toward a scarcity of savings available to meet financing needs.
Despite these headwinds, Sheets expressed cautious optimism about the future. He suggested that if economic growth driven by AI is effectively harnessed and invested rather than spent unwisely, it could help reduce the US deficit over the coming decade. “AI is the sunshine,” he remarked, though he acknowledged that several obstacles remain. On balance, he predicted that economic conditions would “be okay” moving forward.