Positioning for AI Disruption: Investor Feedback & Top Debates

研报英文原文证据摘录

Positioning for AI Disruption: Investor Feedback & Top Debates

to execution and timing. This aligns with our AI Analyst

Sentiment Tracker, where many infrastructure-exposed sectors show limited near-term

disruption risk and are already recognizing revenue from the build-out today. At the same time,

analyst commentary consistently points to power availability, permitting, and data center

delivery timelines as constraints, suggesting that the pace of earnings realization will be defined

by what can physically be built, rather than by demand strength alone.

2) Is digital infrastructure a bubble or a durable investment cycle?

In our conversations with investors, we heard debate on the duration of the infrastructure

cycle and whether current valuations already reflect a peak.

Our view: AI is better characterized as a durable investment cycle than a

speculative bubble.

We see the AI build-out underpinned by real demand, while supply-side constraints could result

in a less peaky, but longer-duration investment cycle, as noted in Powering AI: GEV's Power

Commentary Has Us Thinking about the 'Powering AI' Cycle. AI adoption is still early in its

deployment phase, and looking ahead, we expect demand to be supported by the shift toward

inference. As we've previously argued, lower costs can democratize AI and accelerate activity,

per Jevons Paradox, which holds that lower costs related to technology efficiencies tend to lead

to higher resource consumption. The deployment of agentic AI and physical AI will also drive

significant increases in inference demand. Importantly, AI inference demand is not about AI

adoption alone, but rather the compounding exponential growth across (i) AI tasks x (ii) tokens

per task x (iii) compute per token due to longer token context windows.