More on convertibles – wording matters

Recently during a discussion I had with our young associates, I realized that not every one of them digs into the past executed investment agreements when building a cap table simulation.

This can be quite dangerous especially when there were convertible notes present. Usually people only worry about (1) cap and (2) discount, which in principle is okay. But sometimes the details in the contract could be surprising when it's time to convert.

A typical example is how to calculate the company's fully diluted capitalization immediately prior to the priced round. This number is used to calculate a potential conversion price by dividing the conversion cap (say, $10M) to this number. The higher this number is, the lower the conversion price, which results in more shares converted.

Most convertible bond contracts only account issued common shares and optional pools as well as other issued preferred shares when doing this calculation. In other words, un-converted convertible bonds are not taken into account. This is a logical choice.

On the other hand, the SAFE instrument popularized by YC has a different definition for this number:

Excerpts of YC's (in-)famous SAFE notes.

As one can see, the definition of capitalization here includes "other convertible securities" but excludes "convertible promissory notes". While "convertible promissory notes" are exactly the most common convertible bonds that we all know so in this sense SAFE is no different. However, don't be surprised if lawyers representing the holders of those "other convertible securities" spend tons of time bickering with the startup's advocate at the ensuing priced round, whatever kind of instruments they really are.

Also, SAFE puts itself in priority to any potential new option pool issued at the ensuing priced round. This is yet another difference to most normal convertible bonds.

Bottom line is, not all convertibles are ranked equal. Investors have to be diligent in chasing down the executed documents of all convertible-like instruments.

On raising from VCs

The bottleneck for AI progress continues to be hardware