SCAs as engineering instruments, not press releases
Most large partner agreements optimize for headline dollars and burn through MDF on activity that produces slides, not shipped software.
Structured the Esri stack to prove the opposite: a multi-year private pricing agreement anchored consumption, and a separate AI-focused SCA was deliberately scoped to fund engineering artifacts — two Foundation Models for geospatial AI and an ArcGIS Model Server POC — with named owners, milestones, and re:Invent as the forcing function.
Both sides' engineers leaned in because the contract required it. ~$1M ARR estimated for 2026 from the SCA alone, on top of ~$10M ARR from the PPA. Thesis: an SCA with defined engineering deliverables buys product; MDF alone just buys activity.