Everything here is open to anyone. No account needed.
The SAVRN public site: home, solutions, platform and company
savrn.com is where SAVRN explains what it builds and how to start a conversation. SAVRN designs, builds and delivers AI factories: modular compute, on-site power, water-free cooling, an operating platform, and a training institute that prepares local people to run them. This chapter walks you through the home page, the navigation, the Solutions pages, the Atom drawings, the Platform and the guided tour, then the case-study register, the contact forms and the legal pages. The case-study register is the place to check what stage each design is at, because it labels every case by its evidence status.
- 1
Start on the home page
Open the home page. The top section explains the idea in one line: an AI factory, a training institute and a community learning center, designed as one campus. Below that are three doors:
- Inside the AI factory opens the interactive Atom drawings.
- The Training Institute opens the Institute.
- Community learning center opens the campus concept.
Further down are short sections on private data and models, the operating platform, the AI workforce, campus financing, and SAVRN Research. Each one has a link to its own page. Every Plan an AI campus button opens the Engage form.
- 2
Find your way with the header and footer
The header has nine menus: SAVRN Index, Models, Cloud, Intelligence, Trackers and Maps, AI Solutions, The Institute, Research and Blog, and About. Each one opens a panel of links. Press Command K (Ctrl K on Windows), or use the search box, to search the public pages. On a phone, tap Menu.
Two buttons look alike but go to different places. The Contact SAVRN button at the right of the header opens the Engage form, for a specific project or site. The Contact link in the footer opens the short general contact form. Step 10 explains when to use each.
The footer also carries Privacy, Terms, Legal, Accessibility and a complete site map.
- 3
Walk the Solutions hub
Solutions describes the delivered AI factory one piece at a time. The delivered system explorer has a tab for each piece, and each tab has highlights, key details and an Explore link to its own page. The pieces are:
- The SAVRN Atom (compute)
- Proton fuel-cell power
- Off-grid power and the energy island
- Water-free liquid cooling
- Private compute on SAVRN hardware, where SAVRN runs its own model inside your boundary
- The Training Institute
- The two partner models, Boxcar and Sidecar
Lower down, the page sorts the offer by audience (universities, communities, network owners, operations that need power and compute, investors) and ends with common questions.
- 4
Explore the Atom drawings
The Atom page shows the factory block drawn from its engineering drawings, credited "Built by IntelliFlex". The headline numbers are 13.2 MW gross and 8.8 MW of IT load per Atom, and four Atoms on one shared spine make 52.8 MW. Choose a view (North, South, East or West elevation, Plan, or Power path). Then select a module in the legend or in the drawing to zoom in and read its specifications. Power path walks through eight steps from generation to the rack. On a phone, pinch or use the plus and minus buttons to zoom.
- 5
Read the water-free explainer
Water-Free AI Data Center is a plain-language explainer of what "water-free" should mean and how to check the claim. It separates chip cooling from heat rejection, asks for a whole-site water balance (cooling, fire protection, commissioning fills, maintenance and emergency modes), and lists what a site should meter and publish once it opens. It ends with common questions. Use Explore the evidence to jump to the water-free cooling Solutions page and to SAVRN's source register.
- 6
See the university and data pages
University AI Infrastructure is written for research universities. It walks through an example research workflow from question to reviewed deliverable, then covers research compute, on-site power, water-free cooling, governed workflows, the Institute and the region.
Data to Intelligence sets out SAVRN's data principle: a dataset becomes useful for AI when it records decisions and outcomes, not only measurements. It has three companion pages: the method, an athlete-development design case and a numbered source list.
- 7
Tour the SAVRN Platform
The Platform page explains the software SAVRN runs every day as eight centers:
- Command
- Sales
- AI
- Grid
- Institute
- Capital
- Asset
- Security
Use the numbered strip to jump between centers. Press Start the audio tour, or Listen on any section, to hear it narrated. Three ideas apply to every center: one login scoped to your role, everything logged, and humans approving anything that ships, signs or spends.
- 8
Take the guided tour, then open SAVRN Grid
The guided tour walks six stops: site intelligence, a data room, proposals, work orders, the AI workforce and the build lifecycle. It is marked Sample data at the top, and nothing in it is live.
When you want to read a real site, go straight to the SAVRN Grid page, which is also linked from the Solutions hub. It explains what you get (power readiness, fiber and water, risk, incentives, and a one-page site read you can print). Press Enter the Grid, create a free account with your name and work email, and confirm the code SAVRN emails you. Then drop a pin on any address.
- 9
Check the case-study register
Case Studies labels each case by its evidence status: a reference design (self-powered AI factory campus), an operating-model case (university research work center) and a program model (community workforce campus). Each case says what evidence it would need before anyone could claim it is operating. Read these to understand how SAVRN separates design from proof.
- 10
Get in touch or request platform access
There are three ways in:
- Contact (the footer link) is one short form for universities, communities, investors, partners and press. Add your name and work email, pick a topic, write a message and send.
- Engage (the header's Contact SAVRN button, and every Plan an AI campus button) is for discussing a specific project or site. It takes you into the guided tour after you submit.
- Request access is for teams and investors who want a platform login.
Research emails are an optional checkbox on the forms. If you already have an account, sign in on the platform login page.
- 11
Legal and privacy
The footer links to the Privacy Policy, Terms of Use, Legal Notice and the Platform IP and Licensed Use Notice. The Privacy Policy explains the first-party analytics cookie (savrn_visitor), the third-party analytics tools the site also uses, how to clear the cookie, and how to ask by email to opt out of analytics tracking.
Read the narration
This chapter covers the front door of SAVRN. It's the public website, where we explain what we build and how to start working with us. We design, build and deliver AI factories. Each one brings together modular compute, power made on site, cooling that draws no community water, the software that runs it, and a school that trains local people to operate it. Start on the home page. Three doors take you inside the factory, into the training institute, or into the community learning center. The menus across the top reach everything else, and search finds pages fast. Next, open Solutions. It lays out the factory piece by piece. There's the Atom, our compute block. There's Proton fuel-cell power and the energy island. There's water-free cooling, and private compute on SAVRN hardware. There's the training institute. And there are two partner models, called Boxcar and Sidecar. Each piece has its own page. If you want to see the hardware, open the Atom drawings. Pick a view, tap a module, and read its specifications. One view follows power all the way from generation to the rack. The water-free explainer sets out what that claim should mean, and how anyone can check it. The Platform page shows the software we run every day, in eight centers. They cover the business, sales, the AI workforce, site power, the institute, capital, assets and site security. Every section can read itself to you. Just press listen. The guided tour walks the same story with sample data, from a pin on a map to a running factory. When you're ready to read a real site, open SAVRN Grid. Create a free account, confirm the code we email you, and drop a pin. Read the case studies with care. The register labels every case by its evidence. A reference design is a design. A program model is a plan. Each case says what proof it would need before anyone calls it operating. Then reach out. Use the contact form for general questions, and the engage form for a specific project or site. If your team or fund needs the platform, ask for access. That's the public site. Start with the home page, and follow whichever door fits your work.
Trackers and Maps
SAVRN's trackers and maps are free, public reference pages on where data centers are, what is slowing them down, and who is paying for them. They are built from public records such as county permit systems, legislature pages, grid operator filings and company reports, and every row links to its source so you can check it yourself. Start at the Research Hub, then open the tracker that matches your question: location, policy, permits, water, grid, delays, capital, or AI models. Most trackers also have one page per state and a sources page listing every citation. Each tracker is maintained on its own, so headline counts can differ from one tracker to the next. When you need the latest figure on a topic, use the tracker that covers it directly.
- 1
Start at the Research Hub
The Research Hub lists every tracker, map and research tool in one place. You can also open any tracker from the Trackers & Maps menu at the top of every savrn.com page. For a side-by-side view of states, open State Connections.
- 2
U.S. Data Center Map: where the buildings are
The map places every U.S. data center SAVRN can name from a primary source. Type in the search box to find a data center, operator, city or county. Use Shade states by to color states by facilities identified, reference count, grid operator, or moratorium status. Below the map, click a state tile or Open page for that state's full facility list, or switch to the Operators tab to open an operator page.
The map also has tracker layers for delayed projects, Capital Atlas sites, blocked projects and local restrictions. Those layers are a snapshot from an earlier build, so their counts run behind the trackers themselves. For the latest restrictions, delays or capital records, open the Moratorium Tracker, Delay Watchlist or Capital Atlas directly.
- 3
Moratorium Tracker: state and local restrictions
This tracker maps statewide bills, executive orders, tax-incentive pauses, and city and county moratoria. Filter with the status chips (In force, Paused, Proposed, Failed / vetoed). Click a state on the map to see its measures, each with a Primary source button. Scroll to Every state, one page each to open a state page. Further down is the federal bills table. The full citation list is on the sources page.
Tip: the measure entries and their sources are the most reliable part of the page. Read the individual entries rather than relying on the summary counts at the top.
- 4
Permits and Power Development Tracker
This tracker follows data center and substation building permits from city and county permit systems, interconnection queue positions, planned generators reported to the EIA, NRC reactor projects, and EPA-registered data center facilities. Use the category chips and the State dropdown to filter permits. Each section below (Queue positions, Planned generators, air permits) has its own filters and pages through results 50 at a time. Every row has a Source link, so check the date on the source when timing matters. State pages give the same view for one state.
- 5
Water Tracker: water rules and use
The Water Tracker opens with where data center water actually goes: direct cooling at the building versus water consumed at the power plants. The Fifty states, one taxonomy section shows each state's withdrawal by category from the USGS. Use the filters (All 50, Has a measure, Basin rules only, Nothing located), then click a state to open its file with every water measure that applies there.
- 6
Grid Operator Watchlist: who runs the grid where you build
The watchlist covers ERCOT, PJM, MISO, SPP, CAISO, NYISO, ISO-NE, FERC and NERC. Use the operator chips above the map, then click a state to see which operators serve it and the large-load rules each one has published. Each operator section lists its rulings and filings with a Source button. Check the Last updated date at the top of the page and the date on each source, since some deadlines listed on the page may already have passed.
- 7
Delay Watchlist: named projects that slipped
The Delay Watchlist records named projects that were delayed, cancelled or put under review, along with jurisdictional actions, lawsuits, utility and federal dockets, public-company exposure and securities filings. Use the section tabs (Projects, Local restrictions, Lawsuits, Utility actions, Interconnection queue, Public companies, Filings) and the state filter. To take the data with you, download the whole dataset as JSON from the link at the top of the sections.
- 8
Capital Atlas: who is funding the buildout
The Capital Atlas records capital providers, named sites, financing transactions, and blocked or delayed projects. On the capital map, use the chips (Committed capital, Blocked or cancelled, Delayed or paused, Courts) and click any circle for the record and its sources. The section tabs move you through the numbers, capital at risk, queues, findings and method. Open a state page or a company page for a narrower view. This is a large page, so give it a moment to load on a phone.
- 9
Model Tracker: what changed in AI models
The Model Tracker follows open AI model repositories and records observed changes with the exact revision and a link to the original source. Use search and the Update type filter to narrow the feed. A change is shown as a fact, not as proof of an improvement. Read the methodology page to see how evidence is labeled.
- 10
Get updates and take the data with you
The Moratorium Tracker, Delay Watchlist, Grid Operator Watchlist and Capital Atlas each have a Yes, sign me up box. Enter your email address to get tracker updates. The Permits Tracker, Water Tracker and Data Center Map do not have a sign-up box, so sign up on one of the four trackers above. For raw data, the Public Data Catalog lists the downloadable datasets. If you have a specific location in mind, Site Intelligence pulls these sources together for one site.
Read the narration
This chapter covers the SAVRN trackers and maps. They are free, public reference pages about where data centers are, what slows them down, and who pays for them. Every row links back to its source, so you can check our work. Start at the Research Hub. It lists every tracker in one place. The Trackers and Maps menu at the top of the site gets you to any of them. The U.S. Data Center Map shows the data centers we can name from a primary source. Search for an operator, a city, or a county. Click a state to open its full list. The map's tracker layers are an earlier snapshot. For the latest numbers, open the tracker itself. The Moratorium Tracker covers the policy side. It holds state bills, governors' orders, tax incentive pauses, and city and county measures. Click a state on the map, and every measure shows its source. Read the entries themselves, not just the totals at the top. The Permits Tracker follows building permits for data centers and substations, interconnection queue positions, and planned power plants. Filter by category and state, and page through the rows. The Water Tracker shows where data center water really goes, and which water rules apply in each state. The Grid Operator Watchlist explains who runs the grid where you want to build, and the large load rules each operator has published. Check the date on each source. The Delay Watchlist tracks named projects that slipped, plus the lawsuits, dockets, and filings behind them. You can download the whole dataset. The Capital Atlas shows who is funding the buildout, by state and by company, and where the money stopped. The Model Tracker follows changes to open AI models and ties each change to its source. A change is a fact. It is not proof that a model got better. Each tracker is kept on its own, so totals can differ from one to the next. Use the tracker that covers your question. Finally, you can sign up for email updates on the Moratorium Tracker, the Delay Watchlist, the Grid Operator Watchlist, or the Capital Atlas. If you have a specific site in mind, use Site Intelligence to pull all of this together for one location.
The SAVRN Index, Model Hub and Datasets
This part of savrn.com is SAVRN's public research library for AI infrastructure and open models. The SAVRN Index tracks the companies that fund, build and run AI capacity: what they have disclosed about sites, megawatts and capital, and what they charge for GPUs and tokens. The Model Hub lists open-weight models, datasets, papers and publishers, each read from its own public repository. For each model it shows the license, the memory it needs, and the cheapest GPU setup to run it at today's tracked prices. Around those two sit the SAVRN Standard, the Scarcity Tracker, the AI Workforce book, the System overview, Sentinel Guard and a hands-on dataset explainer. Every Index and Model Hub page also has a plain markdown copy at the same address with .md on the end.
- 1
Start at the SAVRN Index overview
Open the Index overview. The header shows how many companies are profiled, how many trade publicly, and how many prices are tracked. Below that are the seven principles of the SAVRN Standard and links into the seven SAVRN references (Capital Atlas, Delay Watchlist, Moratorium Tracker, Water Tracker, Grid Operator Watchlist, Permits and Power Development, Scarcity Tracker). The menu at the top of every Index page switches between Overview, AI pricing, Companies, Categories, Scarcity Tracker and the SAVRN Standard. For the freshest data date, look at the subpages: each one carries its own date.
- 2
Look up a company
The Companies directory has one page per company, with official links, live delayed market data for the listed ones, named sites from the Capital Atlas, disclosed capacity and capital, and tracked GPU prices. Use the search box to find a company by name or ticker, then click its row to open the company page.
- 3
Read the rankings by category
Categories ranks companies on disclosed facts: top neocloud GPU providers, the largest developers by megawatts, self-powered and off-grid sites, fuel cell powered sites, modular vendors, and sovereign AI offerings scored against the Standard. Each page shows a data date and links every figure back to its source record. On the sovereign AI page, SAVRN's own row is rated from its own design, and the page says so.
- 4
Compare companies side by side, or find alternatives
Compare pages put two or three companies in one table: listing, named sites, disclosed capacity, the operating or building share, site capital, states, and their lowest published H100 and B200 prices. Each ends with plain answers to common questions. Alternatives pages take one company and rank every other neocloud and GPU provider the Index tracks against it.
- 5
Check what AI costs
AI pricing follows one unit of AI capacity from building rent to finished compute to the accelerator-hour to the served and sold token. It shows how many prices are tracked and when they were last checked. From there, open GPU cloud prices by chip, frontier and open model token prices, wholesale and colocation, and the log of price changes.
- 6
Read the Scarcity Tracker
The Scarcity Tracker scores eleven sourced indicators of supply, equipment and power lead time, and demand, each from 0 to 100, and rolls them into one composite score with a dated reading. Each indicator row shows its value, date, weight and source. Further down, a set of price cards reads a market token index, and the price the scarcity score implies, against a retail median, a published owner's-cost estimate for the newest rack-scale GPU system, and the highest tracked frontier list price.
- 7
Browse open-weight models in the Model Hub
The Models page lists open-weight models you can download and run on your own hardware. Filter by task, library, modality or license. Click a model to open its page: it shows the license in plain terms, the memory needed at 16, 8 and 4 bits, and the cheapest accelerator setup that holds it, priced from the SAVRN Index. Use By Task or By License to start from the job or from the terms your counsel will accept. The live counter on each page is the current count.
- 8
Compare two models, or start from a collection
Model comparisons pair models that do the same job at a similar size. Click any pair on the comparisons page. Each one shows parameters, context length, architecture, license, files, the memory needed at 16 and 4 bits, and the cheapest GPUs to run each model, with SAVRN's notes underneath. The example below pairs two small sentence-embedding models. Collections are short, curated lists that answer one practical question, such as which models fit on a single accelerator.
- 9
Trace the data, the papers and the publishers
Datasets lists open training and test sets, sorted by most downloaded unless you pick Likes, Updated or Name. Each dataset page shows its structure (splits, row counts, column types), files, license, and the models that say they trained on it. Some popular repositories have no card or license, so skip those and read the license before training anything you plan to sell. Datasets by task is a good place to start from the job. Papers holds the research behind the models. Model Publishers sorts the whole catalog by the lab or person who released it. The Model Tracker logs observed model changes back to official sources and has an RSS feed.
- 10
Check how the library is built
The Methodology page lists the sources the hub reads, what each evidence label means, how relationships are decided, how often records refresh, the recent import runs with their results, and what is not built yet. A JSON status feed gives live coverage and freshness for anyone who wants to check it.
- 11
Screen a site with SAVRN Intelligence
Intelligence is a browser-based workspace for site decisions. Enter a US address on Sites to open a cited site report. Then save it and work it through Portfolio, Scenarios, Diligence, Monitoring and Reports. Your work stays in your own browser, and you can export a copy.
- 12
Meet the AI Workforce
The AI Workforce page explains how SAVRN runs on governed software agents. It shows how many desks exist, how many run today and which project phase each later desk belongs to. Read the full book, open a division volume, or read how a single agent is built and governed.
- 13
See the whole system, and Sentinel Guard
The System page shows how a SAVRN AI factory fits together: the Intelligence Center and Training Institute, the Atom compute block, Electron on-site power, the operating software, and the two partner models, Boxcar and Sidecar. Press Listen for a narrated walk-through. The Sentinel Guard page describes SAVRN's drone program for site security: thermal patrols, perimeter checks and safety watch. It is designed to be detection-first, with every flight logged and every alert sent to a named person, and it is scoped site by site.
- 14
See how your own documents become training data
The research-to-AI page walks through the path from your files to a fine-tuned model. Paste text, upload a PDF, or pick a sample. The page splits it into passages and builds one example training row in front of you. Today the question-and-answer step is a demo: the answer is built inside your browser from a template, no model is called, and nothing is stored.
Read the narration
This chapter covers SAVRN's public research library. It has two halves. The SAVRN Index covers the companies building AI capacity. The Model Hub covers the open models you can run yourself. Start at the Index overview. Up top you'll see how many companies we track and how many prices we follow. Below that are the seven principles of the SAVRN Standard. That is the yardstick we hold every offering to, our own included. From there, open Companies. Every company gets its own page. It shows named sites, disclosed megawatts, disclosed capital and published GPU prices, and every figure is tied to a source. Categories ranks them. Who are the biggest neoclouds? Who has the most megawatts? Who is self-powered? Who is building with fuel cells? The compare pages put two or three companies in one table, so you can see who is bigger and who charges less for an H100. The alternatives pages rank everyone else against the one you're looking at. Next is pricing. It follows one unit of AI capacity from rent on the building, to finished compute, to the GPU hour, to the token. The Scarcity Tracker scores eleven sourced indicators and rolls them into one number for how tight the market is right now. Then the Model Hub. These are open-weight models, which means you can download the weights and run them on your own hardware. Every model page tells you three things. What the model does. What the license lets you do with it. And how much memory and GPU it takes to run, priced from the Index. You can browse by task or by license, compare two models that do the same job, or start from a short curated collection. Datasets, papers and publishers let you trace a model back to its data and to the people who built it. When you browse datasets, read the card and the license before you use anything. The methodology page shows exactly where all of this comes from and how fresh it is. A few more stops. Intelligence lets you screen a real address and keep your work in your own browser. The AI Workforce shows the agents that run SAVRN. The System page shows how a SAVRN factory fits together. The Sentinel Guard page describes our drone program for site security, which is scoped site by site. Last, the research-to-AI page shows how your own documents turn into training data, in a demo that runs in your browser. One more tip. Every Index and Model Hub page has a plain text copy for machines. Add dot m d to the end of the address.
Research, the blog and the SAVRN Institute
This part of savrn.com is where SAVRN publishes what it knows. The blog, called SAVRN Insights, carries essays, research briefs and the regular tracker updates. The Research section holds our long-form reports. Major essays and briefs have a sources page so you can check the receipts, and the methodology page explains how we label a claim. The Papers directory lists the research behind the open-weight models we track. The SAVRN Institute publishes our workforce curriculum: departments, courses, proposed certificates and how a partner can plan a program with us. Everything here is public. You do not need an account.
- 1
Start at SAVRN Insights
Open the blog. The top card is the newest piece, and the complete archive runs below it, newest first. Each card shows a category tag, such as Power, Water Tracker, Capital Atlas, Moratorium Tracker or Model Tracker, along with the author and the date. Click any card to open the full piece. Use the search box in the header to find a topic.
- 2
Read an essay
Essays are long, signed pieces with the evidence built in. Open the example below. At the top you see the byline, the date and the reading time, then a short row of headline numbers. Source links sit right inside the text, so you can click through to the record behind a figure as you read. At the bottom, a Read next section points you to related essays, including the other parts of the series.
- 3
Follow the tracker updates
Every tracker publishes a dated update post that says what moved since the last edition: counts at the top, then the new entries, each with its source link. Each update links back to the live tracker. The sign-up box on the update posts lets you choose which updates you want by email. To learn why the trackers exist, read the origin piece.
- 4
Read the long research report
The Superintelligence Transition is our largest research report. It has eleven chapters, from school and first jobs through health, money, civic life and the data centers underneath. Each chapter has its own page and its own audio edition. Open the evidence register to see every source in one place, along with a glossary.
- 5
Open the Research library
The Research page holds our reports, each built around one question. Two are published today. The first is a source-cited map of the data center supply chain, with lead times for transformers, turbines and generators. The second, The Enclave and the Gate, is a reference design for running AI privately inside a university data center. Click a report card to open it. The page also links to the Community Value Calculator and the trackers.
- 6
Check the receipts on the Sources pages
Major essays and briefs have a Sources and Citations page. Open the example below: it shows how many primary sources sit behind the essay, grouped by section, with each claim next to a View source link. A link back to the essay sits at the top. At the bottom, if you have a better source or spot an error, click Tell us and we will correct it. The All research sources link takes you to the hub, which lists every topic with its source count.
- 7
Understand how we label a claim
The methodology page explains how we rank sources and how we label each claim: fact supported by a source, calculation or estimate, design assumption, proposal or planned work, operating evidence, or unresolved question. It also explains how to send a correction.
- 8
Read a Perspective piece
Perspective pieces are signed arguments. The current one looks at the Texas poll showing most voters oppose a data center in their community, and makes the case that people object to the terms, not the building.
- 9
Browse the Papers directory
Papers lists the research papers behind the open-weight models in the SAVRN Model Hub. Filter by year, search by title, or sort by date or name. Click a paper to open its page, which shows the authors, the abstract, a link to the full paper, and which models in the hub cite it.
- 10
Meet the SAVRN Institute
The Institute is our workforce curriculum for the people who run AI infrastructure. The home page shows who it serves (employers, universities and colleges, communities) and the published catalog: 10 departments, 43 courses and 4,668 hours, with 9 proposed certificate programs. Location, start dates and credentials are set with each local partner. Each audience has its own page.
- 11
Explore programs, departments and courses
Programs lists the nine proposed Level 1 certificates with their hours, from Service Delivery Coordinator at 576 hours to AI Compute Operations Technician at 956. Departments breaks the curriculum into ten areas. Click a department to see its courses, then click a course. Each course page shows total hours, the lecture and lab split, prerequisites and what a learner will be able to do. The catalog presents the whole curriculum as a book.
- 12
See the campus, funding and partnership pages
The campus page shows the labs and learning spaces in our campus concept. The Workforce Pell page compares the federal rule with each certificate, hour for hour. The university partnerships page explains how a college can adopt the curriculum.
- 13
Plan a program with us
How It Works lays out the five stages: assess, design, prepare, deliver and measure. When you are ready, open the Plan a Program form. Only your name and email are required. You can also name a program of interest and an existing facility. Send the form and SAVRN will follow up on your inquiry.
Read the narration
This is the part of SAVRN where we publish what we know. Start with SAVRN Insights, our blog. The newest piece is always at the top, and the full archive runs below it, newest first. Each card tells you the topic, the author and the date. Open an essay and you will see the evidence built right into the text. Click a source link and you land on the record behind the number. Every tracker we run posts a dated update on the blog. Each one tells you what moved, shows the new entries, and links every entry to its source. If you want those updates by email, the sign-up box is right there on the update posts. For deeper reading, open the Research page. Each report is built around one question. One maps the data center supply chain and how long equipment takes to arrive. Another lays out a design for running AI privately inside a university data center. Our largest report, The Superintelligence Transition, follows AI through a human life in eleven chapters. Each chapter has its own audio edition. We want you to check our work. That is why our major essays and briefs have a Sources page listing the primary records behind them. The methodology page explains how we label every claim. A fact, an estimate, a design assumption and a plan are not the same thing. We tell you which one you are reading. If you find a mistake, tell us and we will fix it. The Papers directory lists the research behind the open models we track, with abstracts and links to the full papers. Then there is the SAVRN Institute. This is our curriculum for the people who run AI infrastructure. It has ten departments, forty-three courses and nine proposed certificates. Every course page shows its hours, its labs, and what a learner will be able to do. The Workforce Pell page compares the federal rule with each certificate, hour for hour. Employers, colleges and communities each have their own page. When you are ready to build a program, open the plan form. It needs only your name and your email, and SAVRN will follow up.