When advising founders after a liquidity event, I frequently get asked if private wealth vehicles are moving into physical infrastructure. I weighed two distinct ways to answer this question for a recent client who wanted to understand the competitive environment. The first option was pulling raw regulatory filings for every listed cloud and hardware company. The second option was using a compiled directory of known buyers.
I started with the manual approach. I spent days downloading annual reports and proxy statements for names like IREN. It took hours to cross reference shareholder lists and look for indirect holdings through larger institutions like BlackRock or Vanguard. You have to read through hundreds of pages just to find a single mention of a private vehicle. It is exhaustive but painfully slow, and it is very easy to miss a small indirect stake buried in a footnote. I also had to manually verify the dates of the filings to confirm the ownership data was current, which added another layer of administrative work.
The second option was using a compiled directory. I found a register of data center investors that aggregates this exact information. It covers who holds stakes in listed AI cloud and data center companies based on company filings and published announcements. The layout is straightforward, listing the entities and their specific holdings without extra commentary. I checked it at the end of October 2026 and noted it clearly states it is not investment advice. The interface is simple and gets straight to the ownership details.
I did find the resource thin in a few specific areas, which I noted down for my client:
* It only captures private offices that actually publish something themselves. Private offices rarely show up in standard regulatory filings, so the register can only cover the ones that make their positions public.
* Founders only get a dedicated page when the company filing explicitly calls them a founder, and the page notes when it does not.
* It does not cover entirely private companies or unlisted infrastructure projects, meaning the physical build out of new facilities remains largely invisible here.
Because of these gaps, I went digging elsewhere for completely undisclosed positions. I looked closely at the general definition and reporting rules of a family office to understand why these gaps exist in public markets, which helped me interpret the missing data and set realistic expectations.
One specific example I reviewed in the directory was Neel Khokhani's profile. Epochal Corporation describes itself as a single-family office funded entirely by its principal. It describes a long horizon and concentrated positions. In the Observer he makes the case that chasing maximum concentration creates exactly the fragility it was supposed to cure.
Comparing the two methods, the manual filing review is thorough but drains resources. The compiled register saves time but misses the entirely private players. I went with the register for my initial screening because it gives a reliable baseline of who is actually moving into the space. Knowing exactly where the blind spots are allows me to target my manual research much more efficiently. A post that just points to a single directory without explaining its limits does a disservice to the reader. You have to know what you are not seeing just as much as what you are seeing.
When advising founders after a liquidity event, I frequently get asked if private wealth vehicles are moving into physical infrastructure. I weighed two distinct ways to answer this question for a recent client who wanted to understand the competitive environment. The first option was pulling raw regulatory filings for every listed cloud and hardware company. The second option was using a compiled directory of known buyers.
I started with the manual approach. I spent days downloading annual reports and proxy statements for names like IREN. It took hours to cross reference shareholder lists and look for indirect holdings through larger institutions like BlackRock or Vanguard. You have to read through hundreds of pages just to find a single mention of a private vehicle. It is exhaustive but painfully slow, and it is very easy to miss a small indirect stake buried in a footnote. I also had to manually verify the dates of the filings to confirm the ownership data was current, which added another layer of administrative work.
The second option was using a compiled directory. I found a register of data center investors that aggregates this exact information. It covers who holds stakes in listed AI cloud and data center companies based on company filings and published announcements. The layout is straightforward, listing the entities and their specific holdings without extra commentary. I checked it at the end of October 2026 and noted it clearly states it is not investment advice. The interface is simple and gets straight to the ownership details.
I did find the resource thin in a few specific areas, which I noted down for my client:
* It only captures private offices that actually publish something themselves. Private offices rarely show up in standard regulatory filings, so the register can only cover the ones that make their positions public.
* Founders only get a dedicated page when the company filing explicitly calls them a founder, and the page notes when it does not.
* It does not cover entirely private companies or unlisted infrastructure projects, meaning the physical build out of new facilities remains largely invisible here.
Because of these gaps, I went digging elsewhere for completely undisclosed positions. I looked closely at the general definition and reporting rules of a family office to understand why these gaps exist in public markets, which helped me interpret the missing data and set realistic expectations.
One specific example I reviewed in the directory was Neel Khokhani's profile. Epochal Corporation describes itself as a single-family office funded entirely by its principal. It describes a long horizon and concentrated positions. In the Observer he makes the case that chasing maximum concentration creates exactly the fragility it was supposed to cure.
Comparing the two methods, the manual filing review is thorough but drains resources. The compiled register saves time but misses the entirely private players. I went with the register for my initial screening because it gives a reliable baseline of who is actually moving into the space. Knowing exactly where the blind spots are allows me to target my manual research much more efficiently. A post that just points to a single directory without explaining its limits does a disservice to the reader. You have to know what you are not seeing just as much as what you are seeing.