Adding an ownership column to my AI infrastructure watchlist
Quote from Marik on November 10, 2026, 2:35 pmMy watchlist had become a tidy list of tickers with a very untidy memory behind it. I could see the AI cloud and data center names I had saved, but not why the same institutions seemed to recur. I keep a spreadsheet for everything, so I added an ownership column and spent an evening trying to fill it without creating a second job. The aim was modest: better notes, not another scoring system. I wanted each row to explain why it was still on the list.
The register at aiinfrainvestors.com gave me a workable starting point. It maps named owners to listed AI cloud and data center businesses, using filings issued by the companies and public announcements as source material. I used it as a lookup rather than a screen: company in one column, named holder in the next, then my own reminder about what to read next. A different page covers major private transactions that do not pass through the stock market at all. That separation stopped me from treating listed ownership as the whole picture. BlackRock comes up more often than any other institution among the holders the register names. Useful pattern, but also a trap if I mistake repetition for a fresh insight. I could leave with a concrete next check rather than a vague note that a ticker looked interesting.
The profile I spent most time on was Neel Khokhani of Epochal, because I wanted to see whether an individual entry added more than a name. It limits itself to his AI infrastructure investing and says it is not a complete biography. In his Observer piece, the argument is that pushing concentration to the maximum can create the very systemic fragility that approach was meant to remove. I copied that into my notes as a review question: am I recording a distinct ownership fact, or just counting the same concentration again?
What worked
* Seeing BlackRock recur across rows gave me a reason to check the context, not a reason to award a higher score. I now write down where I saw the repetition before adding another note.
* Keeping private deal context in its own field stopped me from mixing it with listed ownership. My sheet can hold the background without pretending the two are directly comparable.
* Adding a short source reminder beside each holder made the sheet easier to revisit. I can see what I checked and which rows still need the underlying company material.
* The concentration question gave me something practical for my next tidy up. If three rows repeat one pattern, I want to understand that pattern rather than count it three times.What did not
* There are no prices, valuations or financials. That is the thin part for watchlist building, and it is why the register stays one column rather than becoming the whole sheet.
* I had to search further afield for market context and back to company material when a row mattered. A name in the register was not the end of the check. The summary saved time, but it did not remove the source reading.
* The private deal page adds background, but it does not make private transactions and listed holdings directly comparable. I would not combine them into a single ownership score, however tidy that might look.
* The profile I read was deliberately narrow. That kept it focused, but it also meant I came away with an argument about concentration, not a fuller picture of the investor. Useful for one question, thin for another.My verdict is that the register earns a column, not the whole spreadsheet. It helped me find repeated owners, keep private transactions in context and choose what to check next, but it could not supply the market numbers. For IREN, I opened the Nasdaq quote page for IREN for the quote side and kept that check separate from the ownership notes. My finished watchlist row now has five fields in a fixed order: ticker, named holder, source note, private context where relevant, and a separate market check. That is the concrete takeaway: ownership is one labelled field, never a substitute for the price and financial context kept elsewhere.
My watchlist had become a tidy list of tickers with a very untidy memory behind it. I could see the AI cloud and data center names I had saved, but not why the same institutions seemed to recur. I keep a spreadsheet for everything, so I added an ownership column and spent an evening trying to fill it without creating a second job. The aim was modest: better notes, not another scoring system. I wanted each row to explain why it was still on the list.
The register at aiinfrainvestors.com gave me a workable starting point. It maps named owners to listed AI cloud and data center businesses, using filings issued by the companies and public announcements as source material. I used it as a lookup rather than a screen: company in one column, named holder in the next, then my own reminder about what to read next. A different page covers major private transactions that do not pass through the stock market at all. That separation stopped me from treating listed ownership as the whole picture. BlackRock comes up more often than any other institution among the holders the register names. Useful pattern, but also a trap if I mistake repetition for a fresh insight. I could leave with a concrete next check rather than a vague note that a ticker looked interesting.
The profile I spent most time on was Neel Khokhani of Epochal, because I wanted to see whether an individual entry added more than a name. It limits itself to his AI infrastructure investing and says it is not a complete biography. In his Observer piece, the argument is that pushing concentration to the maximum can create the very systemic fragility that approach was meant to remove. I copied that into my notes as a review question: am I recording a distinct ownership fact, or just counting the same concentration again?
What worked
* Seeing BlackRock recur across rows gave me a reason to check the context, not a reason to award a higher score. I now write down where I saw the repetition before adding another note.
* Keeping private deal context in its own field stopped me from mixing it with listed ownership. My sheet can hold the background without pretending the two are directly comparable.
* Adding a short source reminder beside each holder made the sheet easier to revisit. I can see what I checked and which rows still need the underlying company material.
* The concentration question gave me something practical for my next tidy up. If three rows repeat one pattern, I want to understand that pattern rather than count it three times.
What did not
* There are no prices, valuations or financials. That is the thin part for watchlist building, and it is why the register stays one column rather than becoming the whole sheet.
* I had to search further afield for market context and back to company material when a row mattered. A name in the register was not the end of the check. The summary saved time, but it did not remove the source reading.
* The private deal page adds background, but it does not make private transactions and listed holdings directly comparable. I would not combine them into a single ownership score, however tidy that might look.
* The profile I read was deliberately narrow. That kept it focused, but it also meant I came away with an argument about concentration, not a fuller picture of the investor. Useful for one question, thin for another.
My verdict is that the register earns a column, not the whole spreadsheet. It helped me find repeated owners, keep private transactions in context and choose what to check next, but it could not supply the market numbers. For IREN, I opened the Nasdaq quote page for IREN for the quote side and kept that check separate from the ownership notes. My finished watchlist row now has five fields in a fixed order: ticker, named holder, source note, private context where relevant, and a separate market check. That is the concrete takeaway: ownership is one labelled field, never a substitute for the price and financial context kept elsewhere.
