What Dario Amodei Understands About Executive Time That Most CEOs Miss
Moving away from span of control toward span of attention in the AI era.
In 2017, I was running a product portfolio at GoDaddy that touched around $600M in revenue. My calendar that quarter held seven direct reports, each with weekly business reviews, plus four cross-functional steering committees, plus the executive forums that came with sitting at the table where capital decisions got made.
I remember closing my laptop one Friday afternoon in early November and noticing something I had been avoiding for weeks. The three or four strategic questions that only I could answer, the directional calls the portfolio actually needed from me, had not moved an inch.
I assumed at the time it was a discipline problem. I needed to protect time better, push delegation harder, get more ruthless about my calendar. It would be years before I understood the discipline was a symptom. The structure was the cause.
When I started coaching senior tech leaders after leaving corporate roles in 2021, the same pattern surfaced in nearly every engagement. A CTO at one of the large platform companies had fourteen direct reports and could not name a single strategic decision he had personally moved that quarter. A VP at Microsoft told me she had started measuring her own importance by how full her calendar was. A senior leader at Apple admitted, three months into our work together, that he had not had a clean two-hour thinking block since taking the role.
The architecture of their jobs optimized for visibility and coordination. The thing the company actually needed from them, executive judgment on the small number of decisions only they could make, had no protected slot. It was whatever was left over after the operational reviews finished, which is to say, almost nothing.
What I had missed at GoDaddy was that the number of people reporting to me reflected the company’s structural inability to let me do the work only I could do, not the value I was adding.
I thought about that pattern again when I read the recent reporting on Dario Amodei. The Anthropic CEO has exactly one direct report, his chief of staff Avital Balwit. His sister and co-founder Daniela serves as President. Every other executive, including the CFO and CCO, reports through her. In a June 2026 Bloomberg interview on The Circuit, Amodei described the separation between strategic long-term thinking and short-term operations as “incredibly freeing.”
The split runs counter to almost everything Silicon Valley currently rewards. The trend has been toward flattening, broadening executive control, scaling span of management. Jensen Huang manages roughly 60 direct reports at Nvidia. Sam Altman has about half a dozen at OpenAI. The conventional read is that more reports means more leverage. Amodei has gone the other direction, deliberately. He spends up to 40% of his time on culture, with the remainder on AI research direction and the long-form policy essays that have become part of how Anthropic shapes the broader conversation. Daniela runs the organization. He shapes the ideas.
Underneath the structural choice sits a question most senior leaders refuse to ask honestly. What is the small number of things only I can do, and what would it take to protect them? Most of the executives I have coached cannot answer the first half of that question with any precision. They have not separated the work that requires their judgment from the work that requires their presence. The two get conflated, and presence becomes the proxy for contribution.
The conflation gets more expensive, not less, as the company gets more consequential. Anthropic is shaping infrastructure that will outlast its current leadership. The marginal cost of Amodei spending a Tuesday on a budget reconciliation someone else could run is no longer measured in hours. It is measured in the policy essays that did not get written, the research directions that did not get set, the cultural signals that did not get reinforced.
The Amodei structure reads as eccentric from the outside. From inside a senior seat, after enough years, it reads as something else. He has done what most senior tech leaders never get around to doing. He has named, in public, the small number of things that constitute the actual job, and then designed everything else out of his calendar.
The metric most senior leaders optimize is span of control. The metric that actually matters in AI-era leadership is span of attention, and the two are often inversely correlated.
Looking back at that 2017 calendar, the part I find hardest to forgive in myself is not that the strategic questions went unanswered. It is that I assumed the calendar was a reflection of how important my work was, rather than what it actually was, which was a record of which work was easiest to schedule.
For the structured framework version of this argument, I send a private intelligence brief to senior tech leaders twice a week in Executive Navigator. It is short, calm, and built for people navigating real inflection points.