About
Telling AI moats from traps, in public
Investing in AI is the home of Rob May’s book and weekly newsletter, and of the AI Signal Gauge, a prototype that checks whether a company’s AI is earning its cost using only what the company files with the SEC.
The book gives investors six mental models for judging AI businesses. The newsletter keeps those models current as the market moves. The gauge applies the same thinking to public companies, one filing at a time, and shows its evidence.
What we publish
The book
Investing in AI: Six Mental Models to Tell Moats from Traps (Bat House Press, 2026). Ten years and more than 135 AI investments turned into a framework a general investing audience can use, no technical background required.
The newsletter
Rob’s weekly take on where value is moving in AI, on Substack. The book grew out of it; chapter excerpts and the ideas that did not fit in the book appear there first.
The AI Signal Gauge
A filings-only check on whether a company’s AI is earning its cost and whether what it says matches what it reports. The demo walks through six fictional companies; the evidence pages show what it reads in real filings, quoted and linked.
The AI stack map
Who gets paid when AI gets built: the companies supplying the AI build-out in eight layers, from the cloud buyers to chips, memory, networking, power and the foundries, read only from their SEC filings. Deep dives follow the capex money, map the financing loops between suppliers and customers, and track whether the build-out is running ahead of itself.
Company pages
Nvidia, Microsoft, TSMC and the other most-watched AI companies, each read from its own filings: what it does in the stack, its quarterly numbers, customer concentration, financing links, backlog and commitments, and what to watch next.
The quarterly AI Stack Report
After the five largest buyers file each quarter, a dated report on who spent, who got paid, who kept it and what changed. Each report is frozen when it is published, so it keeps saying what it said.
The glossary
Plain-English explanations of the terms in AI companies’ filings, from hyperscalers and neoclouds to high-bandwidth memory, remaining performance obligations and vendor financing, each tied to the numbers.
Press and data
Citable figures, embeddable charts and CSV downloads for journalists and researchers, free for editorial use with credit.
The assessment
Eighteen questions that place you in one of 30 investor profiles and point you to the mental models to focus on first.
How we work
- Economics over excitement
- The same AI product can be a moat, a trap or a lottery ticket. We look at what it costs to run, who pays for it, and what becomes scarce, not at how good the demo is.
- Primary sources, quoted
- The gauge reads 10-Ks, 10-Qs and earnings releases, and every reading it shows links to the filing and quotes the sentence it came from. If we can’t point to it, we don’t claim it.
- Test before we trust
- A signal earns a place in the gauge only after it is written down in advance and tested against history. Most ideas fail that test, and we say so rather than quietly dropping them.
- “Not disclosed” is an answer
- When a company doesn’t report enough to judge, the gauge says so instead of guessing. Missing evidence is information in its own right.
- Fictional where it must be
- The gauge demo uses made-up companies with tickers that belong to no real listing, so a demonstration rating is never mistaken for a call on a real stock.
- Dated, and left alone
- Each quarterly report is frozen on the day it is published, so what we said stays visible next to what happened. Live pages update with the data; reports don’t.
- Corrections in public
- When we get something wrong, we fix it and say so on our editorial standards page, rather than quietly changing the number. Readers can tell us about errors through the work-with-us form.
- Not investment advice
- Nothing here tells you to buy or sell anything. The book and the gauge are tools for sharper judgment; the decisions stay yours.
The founders
Co-founder and CEO of Neurometric, which optimizes inference for multi-model AI systems. He co-founded HalfCourt Capital, which has invested only in AI since 2015, built Backupify with Vik before its acquisition by Datto, and is the author of the book Investing in AI.
Founder and CEO of Scalable Ventures, a venture lab building AI-native B2B software. He co-founded UnifyCX in 2002, growing it to more than 6,000 employees, later co-founded Backupify with Rob, invests as a general partner at Unbridled Ventures, and sits on the board of VentureFirst.
Why we built Investing in AI
We have known each other since Backupify, the cloud-backup company we built together and sold to Datto. Since then Rob has spent a decade investing exclusively in AI, and Vik has spent it building software companies and investing in others. Between us we have seen AI from the founder’s side, the operator’s side and the investor’s side.
What we kept seeing was a gap. Everyone now agrees AI matters, but most investors still judge AI companies with habits formed in the dot-com and SaaS eras: wide, scattered bets, comparisons to whoever looks similar, and early growth taken at face value. Those habits fail in AI, where adoption and abandonment move faster, inference costs can eat margins, and a single technical result can reprice a whole layer of the market.
Rob’s book is the framework for closing that gap. The AI Signal Gauge is the attempt to make it practical for public markets: to ask of every company the questions the book asks, and to answer them from filings rather than from press coverage or earnings-call enthusiasm.
Because the AI story a company tells and the AI economics it reports are not always the same thing.
Rob May
Co-Founder, Investing in AI
Vik Chadha
Co-Founder, Investing in AI
Who it’s for
- Individual investors
- People who own or are thinking about owning AI companies, directly or through funds, and want to know what the companies themselves report rather than what is said about them. Everything is written for a general investing audience; no technical or accounting background is assumed.
- Advisers and analysts
- Professionals who need a fast, sourced view of how the AI supply chain fits together: who depends on whom, where the money is concentrated and which disclosures matter. Every figure links back to the filing it came from, so it can be checked and cited.
- Journalists and researchers
- Writers covering AI spending, chips, data centers and the financing behind them. The press room has citable figures, charts and datasets that can be used with credit, and we answer questions about how any number was built.
- Boards and leadership teams
- Directors and executives deciding how to invest in AI, or how to oversee those decisions. The same questions the site asks of public companies apply inside any company spending heavily on AI, and we brief boards and teams on them.
By the numbers today: 91 companies on the AI stack map, 53 with their own company page, 25 financing links quoted from filings on the money-loop map, 8 full glossary entries and 7 downloadable datasets, all rebuilt from SEC filings each time the data is refreshed.
Read Rob every week
The book grew out of years of the Investing in AI newsletter. Subscribe for Rob's weekly take on where value is moving in AI, including chapter excerpts and the ideas that did not fit in the book.
Subscribe on SubstackQuestions about us
Who runs Investing in AI?
Rob May and Vik Chadha, co-founders. Rob wrote the book and writes the newsletter; together they build the AI Signal Gauge. They previously co-founded Backupify, which was acquired by Datto.
Is the AI Signal Gauge investment advice?
No. It is a prototype research tool. Demo ratings use fictional companies and illustrative figures, and the real-company pages show filing evidence without a rating. Nothing on this site recommends buying or selling a security.
Where does the gauge get its data?
Only from what companies file with the SEC: the annual 10-K, the quarterly 10-Q, and the earnings release furnished on Form 8-K. Every reading links back to the filing it came from.
Where does the AI stack map get its numbers?
From the companies’ own SEC filings: structured XBRL data for revenue, margins, capex and backlog, and the text of 10-K, 10-Q and 20-F filings for financing links, customers and commitments, which we quote with a link to the document. Nothing comes from estimates, analyst forecasts or press coverage.
How often is the site updated?
The data is refreshed after each earnings season, and a dated quarterly AI Stack Report is published once the five largest buyers have filed. Figures in the text are computed from the same data as the charts, so they update together.
Do you rate or recommend stocks?
No. The AI stack map and company pages describe what companies report and where they sit in the build-out. They don’t value shares or say whether to buy or sell. The AI Signal Gauge’s demonstration ratings use fictional companies.
Can I use your charts and data?
Yes, for editorial use such as news, commentary, research and teaching, with credit to Investing in AI and a link. The data is licensed CC BY-NC 4.0 with an explicit grant for editorial use; commercial products need a licence. The press page has the details.
Do Rob and Vik own the stocks they write about?
Our editorial standards page sets out our roles and how we handle conflicts, and will list holdings as we publish them. Assume we may own shares in large AI companies; nothing on the site is a recommendation.
Can Rob or Vik speak to my group or brief my board?
Yes. We give keynotes, board and investment-committee briefings and custom research on AI investing and the AI build-out. The work-with-us page describes the options and has an inquiry form.
What is the difference between the AI stack map and the AI Signal Gauge?
The AI stack map describes the whole AI supply chain from public filings, with no ratings: who pays whom and who keeps the money. The AI Signal Gauge is a prototype that asks whether an individual company’s AI is earning its cost; its public demo uses fictional companies, and its real-company evidence pages show filing evidence without a rating.
How do you choose which companies to cover?
A company joins the AI stack map if it files with the SEC and AI is material to its revenue. Company pages started with the most-searched AI names and were extended so that every layer of the stack has at least three, and then to every company in the software layer. Readers can suggest companies through the work-with-us form.
Why only SEC filings?
Because filings are the one source companies can be held to. Press releases, interviews and earnings calls describe what a company hopes; a 10-K or 10-Q reports what happened, under rules and with legal liability. Using only filings also means every reader can check every number we show.
How do I follow along?
Subscribe to the free Investing in AI newsletter on Substack. It carries Rob’s weekly analysis, chapter excerpts from the book, and updates as the gauge develops.