"Per-seat pricing is dead. Any company still pricing that way in 2027 is implicitly subsidising their own disruption."
Is per-seat SaaS pricing structurally incompatible with AI-augmented software economics?
The mechanics support Slootman: if AI substitutes for the seat (a worker), per-seat pricing literally penalises the customer for adopting the product. Empirically, hybrid models (platform fee + consumption) are winning new logos. Pure per-seat is in slow decline; the death is real but gradual.
Klarna replaced ~$100M of SaaS spend with internal AI agents in 2024.
— Klarna shareholder letter 2024
Hybrid pricing models won 64% of >$100k ACV new logos in 2025.
— OpenView 2025 SaaS Benchmarks
Median per-seat retention (NDR 112%) still exceeds median consumption (NDR 104%).
— Battery Cloud Index 2026
Hybrid wins. Pure per-seat dies. Pure consumption is operationally brutal.
- OpenView 2025 SaaS Benchmarks— OpenView
Slootman is right in direction, too binary in claim. Per-seat doesn't die — it loses share to hybrid until it's a minority motion by ~2028.
Confidence reflects strength of evidence weighted by source quality.
For the Geeks · sources & method▾
Structured expert elicitation (IDEA protocol)
4 independent domain experts, Delphi-style two rounds, aggregated via geometric mean of probabilities.
n = 4 experts · 2 rounds
- Stanford AI Index 2026 ↗Compute, model performance, investment, training cost.
- SEC EDGAR — 10-K / S-1 filings ↗Revenue, segment data, risk factor disclosures.
- Crunchbase Funding Rounds API ↗Live deal flow, investor participation, round sizing.
Last agent run · 2026-06-08