1. The Circular Investment Problem and the Revenue Gap

A defining feature of the current cycle is the prevalence of circular AI investment structures:

• Silicon vendors finance or invest in AI labs.
• Labs, in turn, announce long-term infrastructure ambitions that depend on future financing, power, and regulatory approvals.
• Hyperscalers commit capacity to those labs.

While these structures signal confidence, they inflate headline demand without guaranteeing near-term revenue.

Key Observations

Only 10–25% of today’s announced AI infrastructure “commitments” are likely to be recognized as revenue by the end of 2026.

A majority of the capital is explicitly scheduled to land beyond 2028, reflecting data-center build times of 20–36 months.

Many widely cited figures represent letters of intent (LOIs), framework agreements, or “up to” commitments, not non-cancellable purchase obligations.

Investor Implication

Near-term earnings risk in AI Cloud is not a demand mirage, but a timing mismatch between financial markets and physical, regulatory, and organizational realities. The market will increasingly differentiate between:

Hard revenue: non-cancellable POs, minimum cloud consumption, live production workloads.

Soft signals: aspirational capacity plans, contingent financing, and long-dated sovereign projects.