Artificial intelligence and the global economy: A framework for investors
We are approaching the AI revolution through an integrated framework that combines bottom-up company and sector evidence with top-down macro analysis. This paper presents the top-down pillar of that framework, which is designed to assess how AI may affect macroeconomic variables such as real GDP growth, inflation and equilibrium interest rates.
At Fidelity International, we are approaching the AI revolution through an integrated framework that combines bottom-up company and sector evidence with top-down macro analysis. The bottom-up work, including insights from our Fidelity Analyst Survey, helps us understand how individual firms are adopting AI and where productivity gains are appearing.
This paper presents the top-down pillar of that framework, which is designed to assess how AI may affect macroeconomic variables such as real GDP growth, inflation and equilibrium interest rates by looking across the entire AI value chain, from energy and critical minerals to semiconductors, data, models, and real-world applications.
As uncertainty is unusually high, this pillar of our framework is scenario-based rather than providing a fixed forecast. We assess both the strengths and weak links that determine whether countries can produce AI models and whether they can exploit it broadly enough to generate economic growth. This includes financing, energy security, talent, sector structure, workforce adaptability, regulation and environmental constraints. Geopolitical forces, including recent export controls on AI models and advanced chips, and inequality are the key underlying forces shaping both AI production and diffusion.
Key findings (base case)
- AI should lift real GDP growth in our base case, but gradually rather than explosively, through stronger productivity, faster innovation and higher investment. However, countries will not capture its full potential if weak links in areas such as energy, capital, skills, adoption or geopolitical fragmentation slow diffusion.
- On inflation, AI is likely to be inflationary before it becomes disinflationary, leaving the net effect broadly neutral in our base case: investment in electricity, data centres and chips can create bottlenecks and add to inflation volatility, while later productivity gains should lower costs. AI should also put modest upward pressure on the equilibrium real rate, R*, with productivity and investment demand acting as upward forces, while ageing, inequality and precautionary saving acting as offsets.
- AI’s macro impact remains scenario-dependent: real GDP growth, inflation and R* could diverge materially across our low- and high-AI cases. We will therefore monitor bottom-up company and sector evidence alongside top-down macro data to judge whether probabilities are shifting between high-, mid-, and low-AI scenarios.
- Each major economy has distinctive strengths and untapped potential, and while the US and China are ahead today - no country is independent. The US has models, capital and cloud; China has scale and industrial depth; Europe has savings and science; Asian countries control robotics, minerals, memory, and fabrication.
- AI has constructive implications for investors. We have already incorporated the implications of higher GDP growth into our latest CMAs through stronger aggregate corporate revenues, while AI-related dynamics and other structural forces should keep rates above the ultra-low post-GFC, pre-Covid regime. Bottom-up, dispersion will be significant: aggregate implementation remains early and some valuations already embed high expectations, but the AI value chain extends beyond the most visible technology leaders. Investors should avoid treating AI as only a US equity, chip or technology-sector story, and instead combine selective diversification, valuation discipline and forward-looking analysis of firms able to turn AI into competitive advantage and durable earnings growth.
We will continue to update our assumptions as new evidence emerges. We will also extend our analysis in a forthcoming paper, focusing on how AI could impact capital market returns through the channels of real rates and corporate earnings, primarily through its impact on labour costs, interest expenses and corporate taxes.
