(UBS)
The capex cycle has advanced faster than anticipated. CIO's current estimate of global AI capex this year stands at USD 423bn, more than 25% above its initial 2027 projections. In fact, CIO has said that the race to artificial general intelligence (AGI) could trigger a capex cycle where the capex of the enabling layer is dissociated from the near-term monetization potential of the application layer. Nevertheless, it believes this pattern is consistent with previous innovation cycles, and we do not see evidence of an investment bubble—key indicators, particularly those related to company fundamentals, remain reasonable compared to historical bubble peaks.
User growth and the pace of capability upgrades have exceeded expectations. The strong demand for AI products and services has supported the rising capex. ChatGPT, for example, now has over 800 million weekly active users, a fourfold increase from its substantial user base just one year ago. Adoption rates among US businesses have also jumped, according to the Ramp AI index—nearly 45% of US companies now have paid subscriptions to AI models, platforms, and tools, up from around 25% at the start of this year. Meanwhile, the outperformance of recently released AI models points to non-linear gains in capabilities, suggesting that increased compute is still translating into better model behaviors.
Evidence of monetization points to large return potential. The scaling of AI investment has made the question of return on investment (ROI) more prominent, with productivity gains being necessary to justify the level of spending. But while revenue diffusion at the application layer has lagged the rapid pace of infrastructure buildout, our estimates indicate that the AI business is already attractive not only for infrastructure manufacturers, but also further down the value chain. Additionally, a Microsoft-commissioned IDC study finds that early adopters of AI at the application layer are generating around USD 3.70 in returns for every USD 1 invested. CIO's conviction in the structural return on AI investment remains strong, and it has raised its estimate for the AI total addressable market to USD 3.1tr in revenue potential by 2030.
So, CIO continues to believe that the long-term potential of AI remains underestimated. Investors underallocated to the theme should consider diversified exposure across the AI value chain. CIO's strategic focus will increasingly favor the application layer, as it anticipates that companies operating within this segment will benefit most from ongoing AI-related capital expenditures.
Original report - AI should continue to power equity performance, 9 December 2025.
