The people I am proudest of developing do not work for me anymore. A few now run organizations larger than the ones I led when they reported to me, and none of that showed up in a metric anyone tracked while they were on my team.

Most of us measure our leadership by what we ship. Revenue booked, launches hit, share defended. Those are the numbers boards pay for, and they miss the thing that outlasts every one of them, because it never had a line on the P&L.

I have been watching the OpenAI alumni more closely than the ChatGPT and AGI headlines, because the people who left are the real story. Former OpenAI researchers and leaders have gone on to found or strengthen Anthropic, xAI, Safe Superintelligence, Covariant, and a handful of other companies now defining the frontier, each with its own thesis and its own bet.

All of it traces back to one environment. I wrote the full breakdown of this OpenAI Mafia in a post this week: read it here.

When I ran the Bing Ads revenue engine at Microsoft, the people around me were absorbing something no job description could capture: how to hold a monetization call when the data was incomplete, when to override the model and when to trust it, how to argue a number in front of a room that outranked them. They took that with them, and although the engine we built got rebuilt twice after I left, the judgment those people carried out the door is still operating in companies that have nothing to do with search.

From a board seat I saw the same thing from the other direction. The companies that came through a leadership change with their value intact had one trait in common: the people left behind could keep making good calls without the person who had trained them in the room.

That is why I stopped treating retention as a measure of how well I led. You can hold onto every strong person on your team and develop none of them. The question I care about now is who left me sharper than they arrived, and how far that sharpness has traveled since.

Boards rarely price this in, because it does not appear on a dashboard until it appears as a competitor. The value a leader puts into people stays invisible until the day it walks across the street and starts pitching against the company that paid to develop it. By then it gets logged as market dynamics, when it began as a development decision someone made years earlier.

The OpenAI numbers make the point at a scale most of us will never touch. By current estimates OpenAI sits near 852 billion dollars, Anthropic around 965 billion, xAI roughly 250 billion before its merger with SpaceX, and Safe Superintelligence near 30 billion. Amazon absorbed Covariant for technology it had already proven. Call it well over two trillion dollars in enterprise value, built by people who once shared the same rooms, the same constraints, and the same standard for what good looked like.

For a VP or C-Suite operator, that reframes something no performance review asks. You can quote your delivery numbers and your attrition rate without thinking. I have sat in the rooms where those reviews happen, and the number that would actually predict a leader's legacy is the one no one has figured out how to put on the form:

If your strongest people left tomorrow, what would they be able to build because they spent time with you, and would any of it carry your fingerprints?

That is a governance question about where the durable equity in your organization actually sits, and whether your leadership is compounding it or drawing it down.

The two things I work on most with the executives I coach are exactly those: the judgment your best people take with them when they leave, and the judgment you carry into whatever you build next. Start here.