04 — STAX
STartup Architecture indeX
Live tech stack signals across 6,000+ venture-backed startups. Detecting what they build on, where they're hiring, and what's changing — in near-real-time.
Every view shares the same underlying dataset. Every number updates daily from live scans.
Real-time tech adoption across cloud, CDN, AI models, frameworks, frontend, and 15+ more categories. Ranked and updated daily.
Open → ModelsOpenAI leads. Anthropic growing. Plus self-hosted LLMs, Mistral, Groq, Cohere — whatever their sites and GitHub orgs actually call.
Open → InfraH100 at $2.35/hr on 1-year contract. $12.29/hr on-demand. GPU cloud pricing, memory, inference trends.
Open → HiringRAG leads AI role keywords. DPO and Cursor close behind. Structured signals pulled from live job boards.
Open → CohortCompare stacks across 79+ VC portfolios — YC vs Seedcamp vs a16z vs BoxGroup. See which funds bet on which stack.
Open →Claude-generated analysis of the signal data: what's diverging, what's consolidating, what the numbers are actually saying.
Open → Weekly BriefAI infrastructure adoption lags at ~10% while hyperscaler dominance crystallizes. Published weekly.
Read →Coming next: Trends · Charts · Cross-Portfolio Contamination Map · Detection Sprint additions
STAX detects signals from public sources only: company websites, job board platforms, GitHub organizations, job descriptions, DNS patterns, CSP headers, and more. No scraping of private systems. No access to non-public data.
Signal detection rates are not adoption rates — absence of a signal doesn't mean absence of use. But the signal that does leak through tells a consistent story, and it shows up weeks before it lands in earnings calls.
Read the full methodology →Why
STAX is the upstream sensor network I always wanted. It runs every day, tracks every company, and feeds the thinking behind everything I write at From the Porch.
This is a working tool, not a data product. The more you see here, the more you're seeing what I'm seeing.
— David Levy, Porch Capital
Google Workspace's 80% adoption leads Microsoft 365's 35% by a commanding margin, signaling entrenched productivity monopolization rather than competitive switching. AWS leads Google Cloud 51% to 38%, but the split reveals workload specialization — not the winner-take-all dynamic VCs expected. Cloudflare's 49% penetration approaches parity with hyperscalers, marking edge infrastructure as table stakes for AI-native products…
Read the full brief →