The author leaves the punchline a little vague, possibly in deference to a16z portfolio interests. My rendition is this. Value capture for any tech almost requ

The author leaves the punchline a little vague, possibly in deference to a16z portfolio interests. My rendition is this. Value capture for any tech almost requires owning the customer relationship. Customers (non-enterprise anyway) buy must-have solutions, not fancy bolt-ons. Ergo, inventors must sometimes build the mundane parts of an app/platform in order to fully cash in on the fancy new parts. This need goes against founder/inventor philosophies. It's wasteful and boring. That's what makes it so hard. It's also more difficult to fund building the broad solution. I think that many of today's AI companies are going to come up against this dilemma soon. Can a tool that analyzes CRM data make as much money as one that handles the whole problem? (Think of how many times CRM tools have been reincarnated in the last 30 years. Building yet another is not that crazy.)