In the first half of 2026, more than 100 tech companies announced layoffs, totaling more than 100,000 jobs (at time of writing). While many cited AI as a driver, it was clear that some were typical cost reductions or over-hiring corrections. The problem with this approach is that executives are simply hoping the organization's structure can survive a reduction in employees. But oftentimes, this kind of layoff is a half-measure that usually forces another round six months later.

If tech companies want to actually improve performance after an AI-driven reduction, they should start by redesigning their organizational structure, then deciding what roles still make sense inside it.​

Org Charts Were Built For A Different Era

Tech leader Molly Graham's metaphor about "giving away your Legos" has shaped how a generation of operators thought about scaling. It frames what it feels like when your company begins growing faster than you can keep up with. There comes a time when you have to hand some of your responsibilities to your teammates, then find new pieces to play with so you can build a bigger tower.

However, with each step change in size, your org chart introduces new layers of complexity. For example, what started out as a single role responsible for doing the strategizing, executing and analyzing may become three separate roles. Execution can get bottlenecked as a result because the work that needs to be done requires people to do it.

With the rise of AI adoption, complex organizational structures are becoming dated. The layers responsible for execution and analysis are no longer as load-bearing as they once were, and their perception as enablers of growth might now be the very thing preventing it.​ But in these scenarios, a standard layoff won't make the kind of difference that executives want. What needs to take place is a structural redesign.

Before You Make Cuts, Sort The Work

During a reorganization, instinct is to look at the people and determine who's expensive, who's redundant and who's underperforming. But that can undermine the growth you're looking for because people aren't actually the unit you're sorting. The work is.

‍

Read full article here