AI is simultaneously everywhere and nowhere. It dominates boardroom conversations, conference agendas, LinkedIn feeds, earning calls. And yet, for most organizations, it's like a toy sitting in its box. Everyone got the gift. Few know what to do with it. In the frenzy to be seen doing something with AI, companies are shoehorning different innovations into existing structures and hoping something sticks.

Jack Dorsey is trying something different. Rather than fitting AI to Block, he's reshaping Block to fit AI, and it started with the elimination of roughly 40% of its 10,000-person workforce. The magnitude of the cuts and the bluntness with which they were executed have generated headlines and attention. Whether you found the move bold or chillingly dystopian, the layoffs force a question: what does AI and the pace of its improvement mean for the current standard of company org structure?

The traditional employee hierarchy, with layers of management, specialized roles, and linear reporting structures, is starting to look less like an org chart and more like a legacy system.

For the uninitiated, legacy systems are outdated technologies so deeply entrenched in core processes that they become nearly impossible to replace. Think COBOL is still running the payments infrastructure, or that ancient ERP system holding your operations hostage. Leadership knows these systems are a liability, but the cost and disruption of updating them always seem to outweigh the theoretical gains. So they persist.

As AI reshapes what individual jobs can do, it will inevitably reshape the structures that connect them.

This isn't unprecedented. Our Thesis 3.0 explores how the collapsing cost of connectivity reshaped how we understand and interact with information. To put it into context, the cost of bandwidth dropped nearly 100,000x over 25 years. Those of us born during the internet era can’t even begin to appreciate what the collapsing cost of connectivity looked or felt like in the workplace (or much less have the experience of management pushing computer use on us).

Today, with cheap bandwidth being the norm, email, instant messaging, e-commerce and the lack of typist pool are unremarkable. Yes, the dotcom era was a bubble but the internet fundamentally was not. What felt like chaos eventually became infrastructure.

Signs point to the fact that we’re at the same inflection point in the collapse of the cost of intelligence. The cost of using LLMs dropped by a factor of over 280x in less than two years. These unit economics are impossible for businesses to ignore. Organizations that treat their current structure as fixed will find themselves in the same position as businesses that treated their technology stack as fixed.

The shape of that future is still blurry, but the outlines are starting to emerge. If the trajectory holds, the eight-layer org chart may compress to three. AI-native firms could run at revenue-per-employee ratios that make today's benchmarks look quaint. Management may move from supervision and directing people to coordination and systems orchestration. The new leverage point may no longer be headcount, but model quality. Startups growing today have a structural advantage in that they can shape their teams around AI-driven efficiency before hierarchy hardens. Nobody can say for certain what the other side looks like, but the organizations - enterprise or startup - asking these questions now will likely be the ones who get to find out.

That is a genuinely difficult thing to sit with, especially for those whose livelihoods are caught in the middle of this shift. The people behind the headcount numbers are real, and the disruption being felt right now is real.

But, if history is any indication, the overall demand will almost certainly go up as the cost collapses. And that's a good thing.