The artificial-intelligence boom is reaching the economy in an awkward order. Companies must buy accelerators, erect data centres, secure electricity, hire engineers and finance long contracts before the rest of the economy knows how much useful output the technology will create. The spending is immediate. The productivity dividend is delayed and uncertain.[1][2]
That timing helps explain why central bankers are discussing AI as a reason interest rates may stay higher—not lower. The International Monetary Fund describes the technology boom as a positive demand shock: it supports activity, trade and investment, but it also adds to price pressure. The October outlook projects global growth of 3.1% in 2026 and 3.4% in 2027 while warning that the resilience is uneven and borrowing costs are already constraining governments.[2]
Three numbers behind the monetary-policy question
| Measure | Value | Context |
|---|---|---|
| Global growth | 3.1% in 2026 | The IMF’s October projection; growth is forecast to rise to 3.4% in 2027.Source 2 |
| Business AI use | 17% of US firms | Share reporting AI use in the December 2025 Census business survey cited by Federal Reserve Governor Michael Barr; use was higher at large and technology-intensive firms.Source 3 |
| Potential productivity contribution | 0.3–0.9 percentage point a year | Range cited by Barr from one sectoral estimate of AI’s contribution to annual total-factor-productivity growth over the next decade.Source 3 |
Why a technology boom can raise the neutral rate
A central bank normally distinguishes demand from supply. If AI makes workers and capital persistently more productive, the economy may produce more without generating the same inflation. But reaching that future requires a large stock of new capital. Competition for savings can raise the return required to fund projects, while expectations of higher future income may reduce household saving. Barr argues that both channels can push up the equilibrium real interest rate, often called r-star.[3]
The two clocks of the AI economy
| What happens first | What may happen later | Monetary consequence |
|---|---|---|
| Data-centre construction, chip orders and power contracts | Faster production or better products if firms integrate the technology successfully | Demand can rise before supply expands, adding near-term inflation pressureSource 1Source 3 |
| Heavy borrowing and long infrastructure commitments | Higher earnings if customers pay enough for AI services | Capital demand can lift market yields and expose weak projects to refinancing riskSource 1Source 3 |
| Electricity demand concentrated in particular regions | Grid investment and new generation capacity | Local bottlenecks can raise prices before new supply arrivesSource 3 |
The productivity J-curve matters
General-purpose technologies rarely produce their full gains when the equipment is first installed. Firms must redesign workflows, retrain workers, clean data and discover which tasks should be automated. Barr points to evidence of a J-shaped adjustment: productivity can initially weaken because the organisational changes are costly, then improve after successful adoption. That creates a period in which an economy pays for AI while still learning how to use it.[3]
What central banks have to separate
- A temporary investment surge from a durable rise in the economy’s productive capacity.
- Local power and construction bottlenecks from economy-wide inflation.
- Higher neutral rates caused by stronger investment demand from higher policy rates used to cool an overheated cycle.
- Profitable infrastructure from projects whose expected AI demand never materialises.
The AI boom can therefore be good for growth and still make monetary policy harder. If productivity arrives, economies may support faster wage and output growth. Until then, central banks are pricing the gap between the factories, power lines and debt being built today and the economic capacity those investments are expected to create tomorrow.[1][2][3]
Sources & further reading
- Central bankers eye higher rates as AI boom fuels scramble for cash Reuters · 16 Oct 2026 · Accessed 16 Oct 2026
- IMF World Economic Output Press Briefing International Monetary Fund · 14 Oct 2026 · Accessed 16 Oct 2026
- What Will Artificial Intelligence Mean for the Labor Market and the Economy? Federal Reserve Board · 17 Feb 2026 · Accessed 16 Oct 2026