SAVRN Infrastructure Platform
SAVRN Research Sources & Citations
Last updated: May 15, 2026
Master Index · Sources · Citations · Receipts
Master index of the primary research, dockets, filings, and regulatory documents cited across every SAVRN published piece. Each article has its own dedicated sources page with the full audit-grade citation record. Click through to verify any specific claim against the primary publisher.
Editorial standard. SAVRN’s research operates on a named-primary-source standard. Every quantitative claim in every published piece traces to a named, verified primary source. URLs are validated at the time of publication and re-validated quarterly. Trade-press summaries are never load-bearing. Where a URL has rotated, the document title, docket number, and publication date are sufficient to retrieve the record at the issuing publisher directly. If you find a citation you can disprove, the receipt is on the author.
PIECE 16 · 2026-05-15
AI Data Center Construction Cost: The 2026 Capex Math Guide
Writes the 2026 capex math for AI data center operators, CFOs, and sovereign-program buyers. Hyperscale construction cost is up roughly 38 percent since 2020; AI-ready high-density campuses now run $15 million to $25 million per megawatt. The brief decomposes the capex stack by line item, maps the regional cost spread across California, Texas, Colorado, Nebraska, Panama, and Barbados, and shows why timeline compression is the dominant capex efficiency lever for any AI project with a first-token target inside 2028.
PIECE 14 · 2026-05-13
AI Inference Infrastructure: The 2026 Operator’s Playbook
Operator brief defining inference as a distinct workload class with its own power profile, latency budget, hardware mix, and unit economics. Worked-example unit-economic walk for a 20 MW deployment. Six-dimension procurement scorecard.
FEATURE · 2026-05-11
The Quiet Rewrite: Every US Grid Operator Is Rebuilding the Rules for AI Data Centers
News-pegged feature documenting the simultaneous rewriting of large-load interconnection rules across every US grid operator, FERC, NERC, and eight state regulatory commissions. Every docket, board decision, PUC order, and OCP hardware spec named in the article is linked to its primary source.
PIECE 13 · 2026-05-12
AI Factory Infrastructure: The 2026 Operator’s Playbook
Operator playbook on the AI factory infrastructure pattern — electrons-to-tokens transformation stack with on-site power, modular compute pods, sovereign control plane. Sized for the next-generation rack densities NVIDIA has telegraphed.
PIECE 12 · 2026-05-11
Liquid Cooling for AI: The 2026 Operator’s Density Playbook
Operator density playbook for liquid cooling in AI infrastructure. Direct-to-chip vs immersion, the 50-60 kW air-cooling cliff per ASHRAE, closed-loop designs, and the Dell’Oro market tracking.
PIECE 11 · 2026-05-10
Behind the Meter AI Power: The 2026 Operator Field Guide
Operator field guide on behind-the-meter power for AI infrastructure. Why on-site generation is the only architecture that removes interconnection-queue dependency. Covers FERC orders, hyperscaler precedents (Microsoft/Constellation, AWS/Talen, Google/Kairos), SMR programs, EPA Title V permitting.
PIECE 10 · 2026-05-09
Tokens Per Watt Per Dollar: The 2026 AI Efficiency Metric
Doctrine piece introducing tokens-per-watt-per-dollar as the unifying efficiency metric for AI infrastructure: ties silicon, cooling, network, and power source into one comparable number.
PIECE 9 · 2026-05-08
Build vs Buy AI Infrastructure: A 2026 Decision Guide
CFO/CTO decision-guide on the build-versus-buy choice for AI infrastructure in 2026. Total cost of ownership, deployment-timeline risk, sovereignty, and Intelliflex as the integrated ‘buy’ option.
PIECE 8 · 2026-05-08
AI Campus Colocation: The Sovereign Integrated Model
Doctrine piece on AI campus colocation: the sovereign integrated model where compute, power, and cooling are owned and operated as one stack, versus the conventional separate-vendor lease arrangement.
PIECE 7 · 2026-05-08
Modular AI Campus: How 6 Months Beats Hyperscale Builds
Doctrine piece on the modular AI campus model — pre-built liquid-cooled compute pods manufactured by Intelliflex on a fixed cadence, deployed on pre-identified sites with behind-the-meter generation.
PIECE 6 · 2026-05-05
Data Center Moratorium: 12 States, 2026 Map, The Fix
Doctrine piece documenting the 12 U.S. states with data center moratoria or restrictive AI-load bills, with state-by-state map and operator framing on the regulatory response.
PIECE 5 · 2026-05-05
From Site to First Token: SAVRN’s 6-Month AI Campus Deployment Timeline
Doctrine piece on compressing AI infrastructure deployment from the 24-48 month industry standard to a 6-12 month sovereign campus timeline. Three architectural decisions: pre-identified sites, behind-the-meter generation, modular liquid-cooled compute pods.
PIECE 4 · 2026-05-04
The 49 Billion Gallon Mirage: AI Data Center Water, Verified
Adjudication of a widely-cited industry claim: that AI data centers will consume 49 billion gallons of water annually. Bottoms-up water model verified against Texas Water Development Board and LBNL primary data.
PIECE 3 · 2026-05-04
Data Center Land Requirements for a Sovereign AI Campus
Operator brief on the acreage, zoning, soil, water-rights, and proximity factors that determine site viability for a sovereign AI campus.
PIECE 2 · 2026-05-03
Defense-Grade AI Infrastructure: Sovereign, Air-Gap Ready
Operator brief on building AI infrastructure that meets U.S. Department of Defense, NDAA, CMMC, and NIST SP 800-171 requirements. Air-gap architectures, FedRAMP-eligible postures, JWCC integration.
PIECE 1 · 2026-05-01
Sovereign AI Infrastructure: The Operator’s Complete Guide
Foundational doctrine piece on the off-grid sovereign AI infrastructure campus model. Defines ownership of power, compute, cooling, and control plane as a single integrated stack — versus the conventional lease-everything posture.
Questions?
Reach the team directly.
Legal and compliance inquiries can be directed to our general channel.