The Great Talent Release

A bet, and the evidence for and against it

I want to make a prediction and be honest about how far the evidence goes.

My prediction: as companies built for Evolution 4 struggle to adapt to Evolution 5, they will release more skilled people than at any point in history. Many of those people will not go looking for the same job again. They will build the companies that Evolution 5 needs.

That is a bet, not a fact. Here is what supports it, what does not, and what I am watching.

The pattern: when the old structure lets talent go

In 1957 eight engineers left Shockley's laboratory and started Fairchild Semiconductor. Within a generation their departure had seeded more than 100 companies. I told that story in my first post, The Traitorous Eight.

It was not a one-off. A 2009 study by the Kauffman Foundation found that more than half of the companies on that year's Fortune 500 list were launched during a recession or bear market. Downturns slow business formation for a while. Even so, the study found they do not appear to significantly harm the formation or survival of new businesses. Its author called the data far from conclusive, and I read it the same way: a pattern worth taking seriously, not proof.

What is happening now

• Crunchbase counts at least 95,667 layoffs at US tech companies in 2024 and around 127,000 in 2025.
• Employers announced more than 1.2 million job cuts in 2025, the most since 2020, according to Challenger, Gray and Christmas. About 55,000 of those, or 4.5%, were attributed to AI.
• Built In reports a Stanford finding that employment among early-career workers in AI-exposed jobs fell 16% after ChatGPT launched in 2022.
• The Harvard Gazette reported in September 2026 that entry-level job postings have fallen by up to 73% over four years by some counts. Joseph Fuller of Harvard Business School estimates that 41% of work tasks could be automated or augmented by AI today.

What the evidence does not show

It does not show that AI is the cause. Many layoffs attributed to AI look like financial cuts with a better story. In a survey of more than 1,000 executives reported by Harvard Business Review, 60% had cut jobs in anticipation of AI gains. Only 2% had made large cuts because of actual AI implementation.

It does not show lasting mass unemployment. Doug Elmendorf of the Harvard Kennedy School expects significant job losses over the next couple of decades, yet expects most displaced workers to find new work. His scenario work with colleagues assumes about 3 million people out of work at any time, around 2% of the labour force.

It does not show that laid-off people are starting companies. US business applications reached 3,232,301 in the first half of 2026, the highest first half since 2005 and about 12% above a year earlier, according to US Census Bureau figures reported by Sovereign Magazine. However, the narrower measure of applications likely to become employers rose only about 1.7% and remains below 2021 and 2023. The Census Bureau also changed that series in January 2026, which lowers the count. Nothing in the data says who is filing.

I found no study that links layoffs to founding at scale. If you see a claim that does, ask for the source.

Why I still make the bet

This part is my reasoning, not evidence, and you should weigh it that way.

Evolution 4 companies are slow to adapt. Fuller notes that only about a third of corporate AI experiments succeed, partly because employees get little formal training. Slow adaptation creates pressure, and pressure pushes people out in clusters. Clusters are how the Eight happened.

The cost of starting has fallen. A person with deep domain knowledge once needed a team before they could build a company around it. With Agentic Citizens carrying the execution, one person can cover functions that once needed hires. That is the SynAgentic thesis, so discount it accordingly. I have an interest in it being true.

The people most likely to be released carry what matters most. Domain knowledge, taste, judgment and relationships do not automate. Those are the assets a new company is built on.

What would change my mind

I will revise this view if the narrower measure of business applications stays flat while total applications keep rising, which would suggest more side projects than companies. I will revise it if displaced workers return to employment at similar rates to before, and if the share of layoffs truly attributable to AI stays in the low single digits. I will report back when the numbers move.

If you are inside an Evolution 4 company

The strongest position is to become the person who knows how to run work with Agentic Citizens, whether you stay or go. If you are thinking about building, the advantage of this moment is that a small team with clear vision can now reach further than it could before.

This is analysis and opinion. It is not financial or career advice.

Sources

• Harvard Gazette: Why AI hasn't triggered mass layoffs, yet (29 September 2026)
• Built In: Did AI take your job? The truth about AI washing
• Crunchbase News: Tech layoffs tracker
• Kauffman Foundation: The Economic Future Just Happened (2009)
• Sovereign Magazine: Record US business applications, first half 2026