I posted earlier because I saw someone say they held a stock, then could not find their name in the company's proxy. My original plan was to use the proxy as a complete list of owners. I have since worked out that it is not quite that simple.
I started with AI infrastructure investors (https://aiinfrainvestors.com/), which brings together ownership information for publicly listed data center and AI cloud businesses from filings and public announcements. I found it useful for getting from a company to the investors associated with it, then tracing the information back to its source.
The first thing I had wrong was what a proxy is for. The investor.gov definition of a proxy statement (https://www.investor.gov/introduction-investing/investing-basics/glossary/proxy-statement) explains it as information a company gives shareholders before a vote. It is not meant to name every person who owns shares.
A proxy will generally show certain large beneficial owners, directors and named executives. Someone with a smaller position can own shares without appearing in that table. That alone does not prove or disprove what they said.
"Beneficial owner" also took me a minute to understand. The name on a report can reflect who has voting or investment control, rather than the name I expected to see from a public statement. Shares can also be held through a brokerage or another entity. I would not assume that every missing individual is hiding behind an entity, though. There may simply be no requirement to name them.
The date matters just as much as the name. A proxy table describes ownership as of a specified date, not necessarily the day I happen to read it. If an investor made a statement after that date, the table cannot settle what they owned when they made it. The reverse is true too: an older statement does not establish what they own now.
The AI infrastructure register makes another date problem clear. A proxy can repeat information from an older 13G filing, so even a freshly published proxy may contain a row based on much older information. I had been looking at the publication date and treating every row as equally current. Now I would check the date attached to the row and find the underlying filing.
I also looked closely at the profile for Neel Khokhani (https://aiinfrainvestors.com/family-offices/neel-khokhani/) as one example. The page says it covers his AI infrastructure investing rather than his full biography. It also describes Epochal as a single-family office investing on one principal's behalf, built around long holding periods and concentrated bets. That context is useful, but it does not turn a public statement about a position into a filing I can verify.
That is the limit I would keep in mind with the resource. If a position is self-reported and there is no filing that checks it, the gap remains open. The site can point me toward what has been disclosed, but it cannot fill in an unreported holding just because I want a definite answer.
I noticed a similar care with founder labels. The site ties those labels to the company's own filing and notes when that wording is absent. That helped me separate what a company has formally said from a description I might have picked up elsewhere.
One other filing type I mixed up with the proxy is the 13F. Certain institutional investment managers file these reports, which show reportable holdings at a quarter end. The manager's name on a 13F is not necessarily the name of each person whose money is invested. It is also a snapshot, not a live account balance.
If I were starting over, I would write down three things before comparing claims: the name or entity being discussed, the date of the claim, and the date and source behind the ownership row. Then I would check whether the document is actually supposed to identify that kind of holder.
My takeaway is that a missing name in a proxy means only that the name is missing from that table, not that the person necessarily owns no shares.