Investor, entrepreneur, author of Amplianism: The Economics of Error
Can Layoffs Fuel Israel’s Next Startup Wave?
Intel's development center in Haifa, 2014. Photo: xiquinhosilva, via Wikimedia Commons (CC BY 2.0).
AI is cutting some jobs. It is also cutting the cost of building a company. What happens next depends on the price of trying.
In September 1957, eight young scientists walked out of Shockley Semiconductor in California. Their boss, William Shockley, had just shared a Nobel Prize for the transistor. The group went down in history as the “traitorous eight.” With outside backing they founded Fairchild Semiconductor, and in 1968 two of them, Robert Noyce and Gordon Moore, left again to start Intel [1].
In 1974, an engineer named Dov Frohman, who had invented one of Intel’s early memory chips, persuaded the company to open a small design center in Haifa. It was Intel’s first outside the United States, and it started with five people [2].
That story is usually told as a story about genius. As an investor, I read it as a story about price. California does not enforce contracts that stop a worker from leaving to compete, so leaving was cheap and the best people kept recombining. The legal scholar Ronald Gilson argued in 1999 that this rule may help explain why Silicon Valley kept reinventing itself while Boston’s Route 128, with comparable talent, did not [3].
Before I put money into any founder, I run a short test. I call it the Amplianism test, and it has three questions. Can this person live while they try? Are they free to compete? If it fails, does it end? Since May, I have been running that test on Israel.
The restructuring:
In May, Wix announced it would cut about 1,000 people, roughly a fifth of a company where more than 60 percent of employees work in Israel. It cited AI, but also the strong shekel, which makes Israeli salaries more expensive in dollars [4]. In July, monday.com said it would let go of about 620 employees, 20 percent of the company, around 350 of them in Tel Aviv, as it reorganized around its AI platform [5].
An Israel Innovation Authority survey of tech employers in June 2026 found software companies laying people off at a rate of 6.6 percent and hardware companies at 1.1 percent. Only about 7 percent of companies said AI was the main reason for cuts, but half of those planning layoffs said AI was shaping the decision. Total tech employment barely moved, because others kept hiring [6].
So this is not an AI apocalypse. It is a restructuring that AI is accelerating, and it is releasing senior engineers and product managers who know exactly how their products were built and what customers still lack.
Here is the part I find most interesting.
AI makes some jobs less necessary. It also makes many companies cheaper to challenge.
The tools that are making some of these roles redundant are lowering the cost of building software, allowing small teams to test ideas that once required larger budgets.
That does not guarantee a startup wave. Israel founded about 775 new startups in 2025, well below the more than 1,400 of 2014 [7]. Most people who are laid off will, and should, take another job, and a few months of savings will not fund a company that needs heavy computing power.
But for the ones with a real idea, the decision will turn on the price of trying.
Three questions before trying:
In my first post here, I introduced Amplianism [8], an idea built on two laws. One of them fits this moment exactly: whoever pays decides who tries. The test turns that law into three questions.
1. Can I live while I try? Israeli law has required since 1963 that a dismissed employee receive roughly one month’s salary for each year worked [9]. In practice, much of that money sits in the employee’s pension fund, under the arrangement known as Section 14. It belongs to the worker, but withdrawing it to live on may carry tax consequences and will reduce future retirement savings. It is a cushion, not free venture capital.
2. Am I free to compete? In 1999, Israel’s National Labor Court heard the case of Dan Frumer, a security software engineer who had signed a 22-month non-compete with a company called Radguard and then went to work for a competitor, Check Point. The court refused to enforce the clause. Its starting point, still applied today, is that a restriction on someone’s freedom to work is not enforceable unless the employer shows a legitimate interest, such as a real trade secret [10]. Secrets stay protected. Experience goes with the person.
3. If I fail, does it end? Israel’s Insolvency and Economic Rehabilitation Law, in force since September 2019, gives individuals a standard payment period of three years after a rehabilitation order, which the court can shorten or extend, before qualifying debts are discharged [11]. A failed company does not automatically ruin its founder, but when it does, there is a road back.
On all three, Israel’s answers are stronger than any single law suggests.
When one answer gets expensive:
Michigan ran the experiment by accident. In 1985, an antitrust reform apparently made non-competes enforceable again without anyone intending it, and researchers Matt Marx, Deborah Strumsky and Lee Fleming measured what followed. Job mobility among Michigan’s inventors fell by about 8 percent compared with states that still refused to enforce these clauses, and by about 15 percent for inventors whose skills were most specific to their employer [12].
The study measured job changes, not startups, but the direction is clear: raise the price of leaving, and the most specialized people stay put.
Brazil answers differently:
In my country, workers have the FGTS, a fund into which the employer deposits 8 percent of salary every month; on a dismissal without cause, the employer adds a penalty equal to 40 percent of those deposits [13]. Brazilian labor courts generally accept a non-compete only if it is limited in time and scope and the employer compensates the former employee for the period of restriction [14]. Whoever wants to stop you from trying pays for it. That is exactly the right order.
The third question is where Brazil struggles. Small-business owners are often asked to guarantee loans personally, and founders who do can remain liable long after their companies close [8]. A Brazilian engineer can leave with savings and freedom and still decide not to try.
What is worth protecting:
If Israel wants this restructuring to produce companies and not only CVs, each answer deserves protection. Keep non-competes narrow: the 1999 principle, which protects secrets but not experience, is one quiet reason Israeli talent recombines so quickly. Make clear what severance can be withdrawn, and at what cost to retirement. And keep the road back open for founders who signed personally for a failed company, not only for consumers.
Companies can gain too. Fairchild did not survive as a giant, but its alumni built Intel and AMD [1]. A firm that treats the people it lets go as future partners or founders of acquisition targets may benefit from the wave.
The question to ask before trying:
If you received a severance letter this year, the decision to build something is yours, and it should rest on the idea, the market and the money you can raise. But run the test too. If this fails, who pays, and for how long?
When the answer is “I do, and it ends,” the choice can be made on the merits of the idea instead of on fear of the bill. That is how an ecosystem renews itself, and it is what Amplianism asks of any country: whoever decides pays first, nobody above error, nobody below a fresh start.
Sources
[1] Computer History Museum, “The Traitorous Eight” (2017). ; PBS American Experience, “Silicon Valley” timeline.
[2] Intel Israel, company profile. ; Calcalist, “Audacity and Innovation: This is How Intel Israel Development Center became a Global Chip Development Center”.
[5] Globes, “monday.com to lay off 620, as it adapts to AI”, July 22, 2026.
[6] Israel Innovation Authority and Zviran, High-Tech Employment Survey, August 2026. ; The Times of Israel, “Survey finds AI adoption spurring shift in tech hiring from software to hardware”.
[7] Israel Innovation Authority, Annual Report 2026. ; Ynet, “High-tech brain drain grows as thousands of developers leave Israel”.
[10] National Labor Court of Israel, Labor Appeal 164/99, Frumer and Check Point Software Technologies v. Radguard (1999). Case brief: ; Afik & Co., “The enforceability of an excessive employee non-compete clause”.
[12] Matt Marx, Deborah Strumsky and Lee Fleming, “Mobility, Skills, and the Michigan Non-Compete Experiment”, Management Science 55(6), 2009, 875-889. ; author summary:
Related Topics
Related Posts
