When Belief Becomes Price: A Three-Stage Valuation Discipline for Today’s AI Markets
The V4G House View
In this cycle, the binding constraint is not model performance, but the share of revenue that survives independence and retention testing.
Private AI has produced real scale - roughly $1.2 trillion in reported value and around $20 billion in annualised revenue at the top end of the market - but the evidence base remains thinner than the prices imply. Retention is often undisclosed. Revenue can circulate between equity counterparties. Margins may exclude compute costs that are effectively paid through dilution.
Mid-market QoE disciplines - built in sectors where circular flows do not exist - are now the cleanest way to separate durable AI revenue from belief-weighted revenue.
This note introduces the V4G 4-I Framework - four lenses on AI revenue quality - maps where the $1.2 trillion currently sits, and outlines three scenarios for how the gap closes.
This note accompanies When Belief Becomes Price: A Valuation Discipline Check for 2026.
In One Minute
Scale: 15 private AI companies sum to ~$1.2T in reported valuations; OpenAI ($500B) and Anthropic ($183B) represent ~57% of the basket (two-company share); adding xAI brings Top 3 to 73%
The Gap: Revenue multiples of 25–80× are applied to run-rate figures that have not passed the four-eyes test - Independence, Institutionalisation, Inertia, and Input Pricing
Concentration: The highest-valued companies are often the least transparent on the metrics that matter most
Forward view: Three scenarios determine who absorbs the adjustment - late-stage investors, strategic partners, or employees via dilution
The V4G 4-I Framework
Standard QoE asks whether reported revenue is real, recurring, and arms-length. In AI private markets, those questions map to four testable pillars:
| Pillar | What It Tests | Key Question |
|---|---|---|
Independence | Related-party exposure, ecosystem circularity | What share of revenue comes from investor-customers or equity-aligned counterparties? |
Institutionalisation | Customer concentration, contract depth | Is revenue dependent on a small number of strategic partners with non-market terms? |
Inertia | Cohort retention, switching costs | Is NRR disclosed? What is churn on contracts older than 12 months? |
Input Pricing | Fully-loaded margin, equity-for-compute | What is gross margin when compute is priced at market rates, including dilution? |
Where the $1.2T sits today:
Most of the USD 1.2T private AI value currently concentrates in a "High Price, Low Observable Proof" zone. The 4-I Framework is a map for moving assets out of that zone - or for pricing the risk of staying in it.
