The AI native mandate is a change management job wearing a technical costume

Gergely Orosz talked with almost twenty engineering leaders who are on a career break or seriously considering one, and reports that six of the ten CTO level people he asked privately are on the way out. That is his sample and his framing, so take it as a direction rather than a measurement.

The reasons are less about the tooling than the headline suggests. Equity sitting behind a steep investor preference stack. Startups that stopped growing. Teams small enough that a VP of Engineering is not needed yet. But the reason he puts first is a mandate: transform the company to be AI native, cut engineering costs by twenty to fifty percent, and do it without anybody deciding how the work itself should change.

Claire Vo gives the harsh version. Most of these organizations will never go AI native, she says, because most engineering leaders are not good enough at change management to pull it off. She is not being generous, and she may be right.

The vault’s career page leans on Eugene Yan for its leadership spine, and his framing fits this week uncomfortably well. Match the leader to the stage. Ambiguous zero to one work wants fast iteration and clarity making. Scale up wants expansion discipline. Mature systems want standards, stability, and steady improvement. Most of the leaders now being graded on the first job were hired to do the third one.

There is one detail in the piece I cannot stop turning over. Anthropic’s head of Claude Platform says a project there usually cannot hold more than two people, because each engineer is already running several agents that step on each other.

If that shape spreads, the shrinking is not a cost decision anyone made. It is just what fits.