A pro-build operating doctrine for AI data centers
Technical position paper by Sasan Salmanzadeh.
AI data centers are strategic infrastructure. If hyperscalers let water, grid, noise, construction, tax, land, security, cultural, and community issues collapse into one public anti-AI story, they will lose legitimacy before the buildout can mature.
The answer is not PR. The answer is proof, category discipline, redesign gates, durable local benefit, and a build philosophy that treats legitimacy as infrastructure.
- Designed dark-mode paper
- One-page executive memo
- Category Discipline excerpt
- Canonical Markdown
- PDF snapshot
- Source notes
The paper argues that AI data centers should be treated less like opaque compounds and more like civic infrastructure whose legitimacy has to be designed, verified, maintained, and improved in public.
The goal is not anti-build delay. The goal is the opposite: build faster by proving more, separating categories before discourse collapses them, and over-delivering on the local systems that make frontier compute politically and operationally durable.
paper.md is the source of truth. paper.html and paper.pdf are generated snapshots for easier reading and sharing.
To regenerate the designed artifacts locally:
python3 tools/render_dark_paper.py --paper paper.md --html paper.html --index index.html --pdf paper.pdf
python3 tools/render_dark_paper.py --paper executive-memo.md --html executive-memo.html --canonical-url https://saucony.github.io/legitimacy-is-infrastructure/executive-memo.html --eyebrow "Executive Memo" --description "A one-page executive entry point for the paper's argument: build faster by showing more work."
python3 tools/render_dark_paper.py --paper category-discipline-excerpt.md --html category-discipline-excerpt.html --canonical-url https://saucony.github.io/legitimacy-is-infrastructure/category-discipline-excerpt.html --eyebrow "Paper Excerpt" --description "A short entry point explaining why category discipline is the core operating idea of the paper."
python3 tools/render_social_preview.py --output assets/social-preview.pngThe renderer uses only Python standard library plus local Chrome/Chromium for PDF export.