The October answer
An AI revenue claim is credible only when usage, cost and customer outcomes reconcile. Founders most often select usage-based pricing, while fewer than half of AI Leaders respondents report full real-time cost visibility. That combination can produce revenue growth without a reliable view of gross margin.
What this edition adds
The September report covered founder pricing. This edition adds buyer-side evidence on planned spend, production-agent counts, inference cost, compute constraints and cost visibility.
Revenue quality
Usage pricing works best when activity is tied to both customer value and delivery cost. Without that link, greater usage can increase revenue and erode margin at the same time. Outcome pricing requires an auditable result, while a flat subscription offers predictability but can conceal costly consumption. The appropriate model depends on what the vendor and customer can measure reliably.
Required issuer disclosures
An issuer claiming AI leverage should explain which workflows are in production, how customers adopt them, what they cost to serve, how they are priced, how margins respond to usage and what customers can measure. Agent count alone says little about utilization, quality or return.
The IPO and M&A questions
For an IPO or M&A process, test who signs, which budget pays, how the pricing unit scales, how much inference and compute cost is visible, which data or integration dependency limits growth, and what result drives renewal.
Investment banking transaction matrix
| Transaction question | October evidence | Base | Banking implication |
|---|---|---|---|
| How do AI companies price? | Usage 43%; subscription 24%; outcome 24%; per seat 20% | 148 | Reconcile consumption, commitments and variable cost. |
| How large are buyer plans? | 43% plan at least $100K; 21% at least $1M | 76 | Treat bands as appetite, not TAM. |
| Is run cost visible? | 44% full visibility; 39% partial; 17% none | 77 | Test gross-margin sensitivity and cost allocation. |
| What constrains production? | Compute 29%; inference cost 25%; data 39%; integration 28% | 75 | Separate model economics from enterprise implementation. |
| Who signs? | CEO 49%; finance 31% | 467 | Map sponsor, validator and economic buyer separately. |
Founder pricing components
| Answer | Count | Share |
|---|---|---|
| Usage-based | 63 | 43% |
| Flat subscription | 36 | 24% |
| Outcome-based | 35 | 24% |
| Per seat | 29 | 20% |
| Not charging yet | 12 | 8% |
Base 148; any mention, so shares can sum above 100 percent.
Buyer planned spend
| Answer | Count | Share |
|---|---|---|
| Under $100K | 31 | 41% |
| $100K to $1M | 17 | 22% |
| $1M to $10M | 9 | 12% |
| $10M to $100M | 2 | 3% |
| Over $100M | 5 | 7% |
| Not sure | 12 | 16% |
Base 76.
Together, the exhibits show why usage must be reconciled with cost. Usage pricing can expand with adoption, but its economics depend on consumption, committed spend and inference or tool costs. The spend bands describe respondents' plans; they do not establish realized revenue or total addressable market.
Comparison with external research
McKinsey's 2026 State of AI finds that AI operating costs constrain some organizations and that broad adoption has not produced equally broad enterprise-level financial impact. For issuers, that makes the reconciliation among usage, cost and measured customer or operating value especially important.
What this means for bankers
Build the transaction narrative around reconciled unit economics. Test management's cost and productivity claims against a workflow-level ledger that a buyer or public-market investor can examine.
Questions this report answers
Which AI pricing model do founders use most?
Usage-based pricing leads at 43 percent (base 148, any mention).
How much do AI leaders plan to spend?
Forty-three percent plan at least $100,000 over the next 12 months, and 21 percent plan at least $1 million (base 76).
Can operators see AI operating costs in real time?
Forty-four percent report full visibility, 39 percent partial visibility and 17 percent none (base 77).
Key citable facts
- Usage pricing is named by 43 percent of founders (base 148).
- Twenty-one percent of AI Leaders respondents plan at least $1 million of AI spend over the next 12 months (base 76).
- Only 44 percent report full real-time AI cost visibility (base 77).
Methodology and limitations
Founder pricing, common-instrument sign-off and AI Leaders responses come from separate populations. This report does not estimate valuation multiples, gross margins or market size. Planned spend is reported in bands without midpoint assumptions.
Related reading
Citation
Suggested citation: Newlands, M. (2026). Investment Banking AI Report, October 2026. Open Future Forum.
Work These Questions with Dealmaking Peers
Forum Select meets through small, off-the-record gatherings. Membership is by application and referral.