2.7x Productivity Surge Needed to Secure Returns on AI Investment

According to research ("What Investment Data Implies about the AI Transition") published by NBER (National Bureau of Economic Research) and the Wharton School's Jessica A. Wachter and Jonathan D. Wachter, given the current scale of AI-related capital expenditure (CAPEX) by Big Tech companies, an extremely high level of productivity improvement is required.

Key points are as follows:

• Investment Scale and Profitability: In 2025, CAPEX by the top 5 US tech companies reached approximately $380 billion, and is expected to double in 2026. For this investment to be justified on a net present value (NPV) basis, productivity in the AI sector would need to rise by about 2.7x from current levels.
• Risk Factors: If revenue growth fails to keep pace with these expectations, the companies in question could face insolvency or bankruptcy risk.
• Economic Impact: The research finds that if the AI boom materializes, cumulative GDP could grow by 5-58% by 2030, but it also presents risks such as upward pressure on interest rates and volatility in the equity premium.
• Historical Context: In the past, productivity gains from general-purpose technologies (GPT) such as electricity, the internal combustion engine, and ICT unfolded gradually over decades, but the 2.7x short-term productivity leap now required in the AI industry is historically an extremely unusual pace.