Buying theStandard

The asset was never the forty people

Tabular was founded by the engineers who created Apache Iceberg while they were at Netflix. Iceberg is an open table format, the unglamorous plumbing deciding how enormous piles of data get organised so that different tools can read them.

Databricks maintains Delta Lake, which was Iceberg's principal rival. So this was not a capability Databricks lacked and it was not forty engineers it could have hired one at a time for far less. It was the other side of a format war, bought together with the specific people whose names are the reason anybody trusts that format. Call it a talent deal and you have described the receipt instead of the purchase.

Why a standard outruns a product

A product competes every year. A standard competes once, and then it collects. If the industry settles on Iceberg and Databricks sits outside it, every enterprise data decision for the next decade begins with a format Databricks does not influence. That is not a competitive disadvantage you can engineer your way out of, because you cannot out build a default.

Which is why I think two billion is the cheap end of the range rather than the absurd one. You are not paying for revenue or headcount, you are paying to stop competing against your own standard and to put both formats under one roof. Notice too that Databricks has only ever confirmed a floor, that consideration exceeded one billion. A company thrilled with a price tends to name it. A company that would rather you discussed the strategy gives you a floor and lets the number find its own level, and I read that as consistent with a purchase that was never really about the money.

The argument against my own thesis

If standards are the asset, then the value depends entirely on Iceberg staying a standard rather than becoming a product. Open formats are trusted precisely because no single commercial party steers them, and the largest commercial party in the category now employs the creators.

I have seen no evidence of capture and I do not think it was the intent. But my thesis only holds while the format stays genuinely communal, which means the thing I am claiming is valuable is also the thing the acquisition puts at risk.

My call

Cheap if Iceberg remains everybody's, expensive the moment it starts to look like Databricks'. The checkable version is whether the contributor base stays broad over the next few years or quietly narrows to one company, and that is something you can go and count rather than argue about.