Open-Source AI Stack Maps Full Pipeline

This is a more useful response to AI hype than another vendor landscape: it frames AI as a layered system with public infrastructure gaps, not merely a model leaderboard.

• –Ten layers show where lock-in and sovereignty accumulate across the production pipeline
• –Cross-cutting evaluation, safety, identity, and governance layers prevent critical concerns from becoming afterthoughts
• –Project catalogs and funding data give developers a practical starting point for architecture and ecosystem research
• –The taxonomy’s biggest challenge is defining “open” consistently across code, data, weights, and hosted services
• –Treat it as a decision framework, not a guarantee of production maturity or support