In 1993, IBM cut 60,000 jobs. It was the largest layoff in U.S. corporate history at the time, and it shocked the country. IBM had gone seven decades without a layoff. The company’s culture was built on the promise that if you did your job, you had a place there for life. That promise ended in a single year.
The headlines focused on the human cost. The deeper story was something else entirely.
IBM wasn’t dying. It was restructuring during a paradigm shift. The mainframe era was ending. The client/server era was beginning. And the people IBM cut weren’t surplus to the economy. They were surplus to a particular architecture of computing that was on its way out.
Where they went next is the part of the story most people don’t tell.
And IBM wasn’t alone. Wang Laboratories filed for bankruptcy in 1992. Unisys, formed from the merger of Burroughs and Sperry, nearly collapsed under $2.5 billion in losses. Digital Equipment Corporation ousted its founder the same year as it bled out. The mainframe era didn’t end at one company. It ended across the entire generation of companies that had defined it.
The decade gap
Here’s the thing about paradigm shifts. The talent released in the cuts doesn’t immediately found the companies that come to define the next era. There’s a gap. A long one.
Twitter wasn’t founded until 2006. Facebook in 2004. The web 2.0 wave that the 1993 cohort is sometimes credited with seeding came a full decade after the layoffs. The people who built it had to get there first.
Where they went in the meantime is the more interesting question.
They went to Sun. To Oracle. To Cisco. To Netscape. To the early enterprise software companies and the systems integrators. They went to the companies that were doing the unglamorous structural work of the new paradigm: building the distributed databases, the network protocols, the application servers, the integration layers. The companies that were actively rebuilding what the mainframe used to do, in client/server form.
That work wasn’t the next wave. It was the substrate the next wave would later be built on.
By the time Twitter and Facebook arrived, the senior architects who built them had ten years of distributed systems experience earned at companies most people have stopped naming. The Xerox PARC diaspora is the well-known version of this story: PARC’s people seeded 3Com, Adobe, Cisco, Apple, and Microsoft, but rarely directly and almost never quickly. They went somewhere first. They retooled. Then they built what came next.
What’s actually happening in 2026
In 2026, the same structural pattern is playing out across the giants of an entire era.
More than 125,000 tech workers have been cut in the first four months of the year, on pace to exceed the 246,000 cut in all of 2025. The cuts are concentrated in three corridors: the San Francisco Bay Area, Seattle, and New York. Office vacancy in San Francisco hit 36.7% in Q1 2026, up from 33.9% a year earlier. Seattle’s tech corridor is showing the same pattern. The geographic story is as concentrated as the industry one.
The cuts are a smaller share of a much larger industry than what hit IBM and its peers in 1993. But the structural logic is the same. The incumbents of one era are restructuring during the shift to the next, and the capital that ran the current paradigm is being redirected to build what comes after it.
The companies running those reductions are reporting record revenues at the same time. Amazon’s Q1 2026 capex came in at $44.2 billion, up 77% year over year, with AWS growing at its fastest pace in fifteen quarters. Alphabet’s Q1 capex was $36 billion, up 107%. Together, the four hyperscalers, Amazon, Microsoft, Alphabet, and Meta, will spend roughly $725 billion on AI infrastructure in 2026, up from $410 billion last year.
Meta’s projected 2026 AI capex runs $125 to $145 billion. The company’s total human compensation, every salary, every benefit, every stock grant, comes to roughly $27 billion. The AI capex line is four to five times the entire payroll line.
The layoffs are not the cost-cutting story. They are the capital reallocation story. The cuts could be twice their current size and still wouldn’t meaningfully fund the AI infrastructure buildout. What’s funding it is the redirection of capital itself, away from the work of running the current era and toward the work of building the next one.
The 125,000 are not surplus to the economy. They are surplus to a particular shape of company that is on its way out.
Where they’re already going
Most of the 125,000 won’t found the next wave’s defining companies. The 1993 cohort didn’t either. They went somewhere to retool. The companies they chose are the ones that defined the next era.
The same will be true of where this talent lands now. And the redistribution is already visible.
Recruiting firms tracking placement patterns report that roughly 40% of displaced senior engineers are landing at mid-market or PE-backed companies, often within weeks rather than months. Senior engineers with eight or more years of experience are placing in a median of 17 days through specialized recruiters, against a market average of 45 to 90 days. The fastest-moving slots are senior full-stack roles at mid-market SaaS companies running real cloud migrations, and director-level technology leadership at companies under 500 employees. Total compensation is coming down from hyperscaler peaks. Base is holding. Candidates are trading brand-name equity for cash, stability, and the chance to operate inside a function rather than manage six layers of one.
The redistribution is also geographic. The talent leaving the Bay Area, Seattle, and New York is not all staying in the Bay Area, Seattle, and New York. Mid-size companies in Denver, Austin, Miami, Salt Lake City, and Raleigh have a window they did not have eighteen months ago.
The question worth asking
If you are one of the 125,000 and counting, the question is not “what should I build?”
The question is: where do I go to retool, so that when the next wave really lands in five to ten years, I am the one architecting it?
The answer in 1993 was not the obvious one. It was not the giants that survived the shift. They were busy preserving what they had. The answer was the companies that were actively rebuilding around the new paradigm. Companies most people would have struggled to name at the time. Companies that were doing the hard, unglamorous, foundational work of assembling the building blocks of what came next.
The same is true now.
The biggest, shiniest landing spots are not the ones doing the most foundational work. The companies adopting AI tools at the surface layer are not the same as the companies rewiring how they actually operate. The first group will use AI to do current work faster. The second group will rebuild what work even is. Only one of those is the place to retool.
For the talent reading this, the question to bring into every conversation about a next move is not “what’s the comp” or “what’s the title.” It is this:
How deep does the AI rebuild actually go in this company? Surface tools that make current work faster, or architectural decisions that change what the work even is?
If the rebuild is deep, the architectural decisions being made in the next eighteen months will define what gets built on top of them for the next decade. Being inside that work is what positions you to be the senior architect, founding CTO, or early VP of the companies that emerge in 2032 and 2035.
If the rebuild is shallow, you will spend the next five years doing optimization work on a fading paradigm. That is the wrong place to retool.
What this means for mid-size CEOs
The 1993 diaspora was not absorbed by the giants. It was absorbed by the companies that were rebuilding around the new paradigm and that needed senior capability to do it. The capability didn’t disappear. It redistributed. And the companies it redistributed to were reshaped by it.
The same redistribution is starting now, and the data already shows it. Mid-market SaaS companies are absorbing senior engineers within weeks. PE-backed mid-size firms are absorbing director-level technology leadership at a pace they could not match two years ago. The talent is moving. The question is whether the operating systems on the receiving end are mature enough to put that capability to work on foundational rebuild, or whether new senior hires will land inside organizations that are still patching what already exists.
That is a different kind of readiness than most companies are paying attention to right now. Talent acquisition strategy is the surface answer. The deeper question is whether the foundations of the company can metabolize what’s about to arrive. A senior engineer joining a company whose decision rhythms, architectural standards, and operating cadence are still organized around 2022 will burn out inside a year. A senior engineer joining a company that is actively rebuilding its operating system around AI will compound. The talent knows the difference. The market is now sorting on it.
The real signal
More than 125,000 people are not entering a job market. They are entering a paradigm shift. The decisions they make in the next twelve months will, ten years from now, look like the decisions IBM’s 1993 cohort made when they chose Sun and Oracle and Cisco over the safer-looking giants of the day.
The companies they choose now will be the companies they reshape. And the companies they reshape will be the ones that define the next decade.
The capital reallocation is already happening at a scale the headlines have not yet caught up to. $725 billion is moving toward AI infrastructure this year alone. The talent is already moving toward the companies that are rebuilding around it. The only real question is which side of the reallocation your company is positioned for.
That is the part of 1993 the headlines missed. It is also the part of 2026 worth paying attention to.
Kathy Keating is a Board-Certified Director and systems strategist who works with CEOs and Boards on decision architecture, operating rhythms, and AI readiness. She is the co-author of Liquid: How CEOs and CTOs Unlock Flow & Momentum in Complex Systems.
