For twenty years, the formula for software companies was simple: charge the customer per user. A growing organization needs more licenses, and every new employee is another subscription. Some of the most successful companies to come out of Israel were built on that formula. In recent weeks, during earnings season, it became clear how fragile it is in an era when part of the work moves to AI agents.
The problem in one line
If an AI agent does part of the work of a support agent, a project manager or a designer, the customer needs fewer people. Fewer people means fewer licenses. Wall Street already has a name for it: the SaaSpocalypse. It is not happening everywhere at once yet, but investors are pricing it in advance, and that is enough to move stocks by tens of percent.
monday.com: beat expectations, got punished
monday.com reported 19% growth and revenue above forecasts, yet the stock fell about 6% in early trading, because guidance for next quarter reflects growth of only about 14%. AI revenue, about $36 million a year, is still too small to change the picture. monday.com saw the problem early: in July it announced cutting about 20% of its workforce and a move to an AI agent platform. But such a shift takes time, and meanwhile the old model keeps weakening.
NICE: selling the agents that replace its customers
At NICE the dilemma is even sharper. The company built its business selling software to contact centers, priced by the number of agents. It bought Germany's Cognigy, which develops AI agents for customer service, in a deal of about $1 billion according to reports, because the logic says whoever sells to contact centers must also sell the agents that replace the human agents. This week it announced about 500 layoffs, about 5% of its workforce. According to reports, NICE is worth about $6.5 billion today, far from its peak of about $20 billion.
Fiverr: when small jobs disappear
Fiverr may be the clearest example. In the third quarter revenue fell 15.8%, and active buyers dropped 25.8% to 2.4 million. The buyers who remain spend more, meaning the ones who vanished are the small customers, those who ordered a logo, a short translation or marketing copy. Most of those tasks can now be done yourself with a chatbot. Fiverr is profitable with $316 million in cash, but expects a sharper decline in traffic through 2027.
The other side: who is winning
Not all software companies are hurting. Wix grew only 13%, but the stock rose, because Base44, the vibe-coding tool it bought for about $80 million, is approaching $250 million in annual recurring revenue. JFrog reported 31% growth and cloud revenue up 60%, and the stock jumped about 22% after hours. Riskified accelerated to 41% growth. What they share: they don't charge by the number of employees at the customer, but by usage, volume or outcome.
Why that model works
JFrog manages the "parts" software is built from. The more code AI agents write, the more parts there are to manage and secure, and the customer pays more. Riskified charges a percentage of the transactions it approves, and the more purchases happen online, perhaps by AI agents too, the more transactions there are. Base44 sells the ability to build an app without a programmer. In these cases, AI doesn't shrink the base the company charges on. It grows it.
It's not just Israeli
ServiceNow, one of the world's strongest enterprise software companies, was punished too: the stock plunged about 25% the day after earnings, and the company confirmed its third round of layoffs this year. Investors fear exactly the same scenario: fewer employees at customers, fewer licenses. When the giants are hit too, it is clear this is not one company's problem but a change in how the market values an entire industry.
What companies can do
The obvious answer is to move from per-user pricing to outcome-based pricing: charging per task completed, ticket resolved or deal closed. It sounds simple, but it is risky. The new pricing is less predictable, customers are used to the old model, and revenue may fall during the transition. In addition, running AI agents costs money, so gross margin, one of software's great attractions, erodes. Wix, for example, lowered its cash flow guidance because of its investment in Base44.
What it means for employees
The effect is already felt in the job market. Veteran software companies are cutting, often citing AI, while chip and infrastructure companies are hiring: Tower, for example, plans to add about 150 employees in Israel. Israeli high-tech is splitting in two, and anyone working at a software company that charges per user should ask a simple question: does AI grow what our customer buys, or shrink it?
The bottom line
The previous generation of Israeli tech built large companies by selling tools to people. The next generation will have to sell work that got done. The companies that make the transition, and prove in numbers that AI brings them money and doesn't just cost them money, will keep growing. The others will find that even a good quarter isn't enough when the market no longer believes in the model.
