Originally published by Trace Cohen @Trace_Cohen on X.
View original article
· View original post
· 2026-05-09T01:30:05.000Z
They also weren’t built for the internet, or the cloud either but that kind of worked out well for most.
Large organizations were designed for a slower era of technology where information moved slower, decisions moved slower, software changed slower, and scale itself often hid inefficiency. The systems, org charts, approval chains, budgeting cycles, and management structures all evolved around stability, risk reduction, and predictability.
That worked reasonably well for decades.
But AI fundamentally changes the economics and speed of execution inside companies.
At my last Fortune 50 role, I was helping modernize internal onboarding infrastructure by replacing spreadsheets, email chains, fragmented approvals, and manual workflows with APIs, digital forms, and centralized systems. Nothing particularly groundbreaking either. Just operational improvements that obviously should have existed already to save countless amount of time for everyone involved.
The interesting part was not the technology.
Everyone agreed with the direction almost immediately. Engineering agreed. Product agreed. Operations agreed. Leadership conceptually understood the business case because faster onboarding meant faster revenue, fewer errors, lower operational burden, and a better partner experience.
The problem was that enterprise systems are often optimized around process more than execution.
We reached one relatively straightforward architectural decision that required approval before rollout. What followed was weeks of decks, meetings, reviews, steering committee updates, and management chain presentations so the decision could slowly move upward through layers of organizational process to my boss’ boss’ boss.
The actual decision itself probably took fifteen minutes.
The process surrounding the decision took over a month.
Meanwhile timelines slipped, rollout slowed, partners waited, frustration increased, and leadership still wanted to understand why execution lagged behind expectations.
This is part of why so many companies are struggling through the current transition.
During the COVID ZIRP era, demand was massively pulled forward across software, ecommerce, fintech, cloud infrastructure, and enterprise technology. Companies staffed aggressively for a world where growth rates, cheap capital, and digital acceleration would continue indefinitely. Entire org structures were built around sustaining that pace.
Then the environment changed.
Rates rose. Growth normalized. Efficiency mattered again. And now AI is simultaneously increasing the output potential of smaller teams while exposing how much organizational overhead many companies accumulated during the last cycle.
A fully loaded enterprise middle manager can easily cost a company between $180k and $350k annually once salary, benefits, management overhead, software, payroll taxes, and office costs are included (not including stock, RSUs etc). At scale, reducing headcount by 1,000 employees can translate into hundreds of millions in annual savings before even accounting for reduced operational drag and faster execution cycles.
That is why so many companies are simultaneously laying people off, flattening org charts, increasing AI infrastructure spending, and asking remaining employees to move faster with fewer resources.
According to Layoffs.fyi and related industry tracking, tech layoffs in 2026 have already surpassed roughly 90,000 workers globally, while many companies openly discuss AI-driven efficiency gains alongside workforce reductions.
The uncomfortable reality is that AI is not simply replacing workers.
It is forcing enterprises to confront the fact that many of their systems, workflows, and organizational structures were never designed for a world where execution speed, automation, and real-time iteration matter this much.
Some companies will adapt.
Some will slowly restructure over years.
Some may decide they cannot or do not want to change.
But over the next decade, the organizations that win will likely be the ones capable of removing friction between idea and execution faster than everyone else.
