Anthropic’s CEO Has 1 Direct Report. Nvidia’s Jensen Huang Has 60. I’ve Tried Both. Here’s Which One Founders Should Copy

EXPERT OPINION BY DAVE KERPEN, CEO, KERPEN VENTURES @DAVEKERPEN

Dario Amodei and Jensen Huang. Photos: Getty Images; Adobe Stock

Dario Amodei, the CEO of Anthropic, told Bloomberg he has exactly one direct report. “It’s incredibly freeing,” he said. “It lets me do all the things that I do much more easily than I would otherwise.” Across the industry, Nvidia’s Jensen Huang has around 60. Satya Nadella has about 16, Sundar Pichai about 18. Inc. laid the contrast out in June, and the question every founder reading it should ask is: which one am I, and is it on purpose?

I’ve written that a CEO should do only three things and delegate everything else, about the 5 C’s of delegation, and about the day I stopped being my company’s bottleneck. I’ve also run companies both ways: as a Huang, with what felt like everyone reporting to me at Likeable Media in the early years, and closer to an Amodei, with a small number of leaders between me and everything else. Here’s what each one costs.

The Huang model buys accountability and sells your calendar.

Huang’s line is: “We present a problem, and all of us attack it.” Sixty direct reports means nothing hides. Every problem gets to the CEO fast, and the CEO’s judgment gets applied everywhere. In year two at Likeable, that’s exactly what we needed; the company was too young for a middle layer, and I was the only one who’d seen the whole picture. The cost was that I was the whole picture. When I finally handed the CEO job to Carrie in 2013, it was in part because a company with one brain at the center can’t grow past that brain.

The Amodei model buys focus and requires trust you may not have earned yet.

Amodei’s argument is that “it’s very hard to pay attention to the strategic picture if there’s, like, a zillion things you have to handle tomorrow.” That’s the three-things CEO I’ve written about: vision, key relationships, and the culture. But it only works if the one or two people beneath you are people you’d hand the keys to. Most founders adopt the Amodei structure before they’ve hired the Amodei-caliber layer, and then they’ve simply added a bottleneck below the old one.

Both leaders are protecting the same thing: their attention.

That’s the part the org-chart debate misses. Huang uses 60 reports to put his attention on every problem at once. Amodei uses one to put his attention on the biggest problem only. The failure mode isn’t picking the wrong number. It’s picking no number, which is what most founders do: eight reports by accident, three of whom should really report to someone else, a calendar that reflects who asked first rather than what matters.

The founder’s answer changes with the stage.

For most startups I invest in, the honest answer is Huang until roughly 20 people, then a deliberate migration toward Amodei, one leader at a time, as you find people who don’t need you. The mistake is treating either model as a personality. It’s a tool, and the tool should change as the company does.

Amodei is right that one direct report is freeing. Huang is right that 60 keeps everyone honest. The question for you is simpler: does your org chart look the way it does because you designed it, or because you’ve never looked?