People are deeply anxious and uncertain about AI replacing jobs. How can we move beyond vague warnings and exaggerated predictions and use data to answer this question? A good approach is to examine a profession in which AI capabilities are among the most mature and adoption has progressed exceptionally quickly: software engineering.
In this article, we argue that the existing evidence is already sufficient to refute the narrative that once AI capabilities reach a certain threshold, they will lead to mass layoffs. If this holds true even in an industry with very few regulatory barriers, then most other professions are likely to have even stronger buffers.
We also have a fairly clear understanding of why. We can view many forms of knowledge work, including software development, as a “decide-execute-deliver sandwich.” AI compresses the “execute” layer—the middle of the sandwich—but the other two layers resist automation in ways that cannot be overcome through capability improvements alone.
Ultimately, we are cautiously optimistic about the future trajectory of demand for software engineering. This article is the first in a series. The next will explore why the careers of individual software engineers may still be highly turbulent even if overall demand remains healthy. This series draws on published literature in economics and software engineering, our evaluations and observations of AI agents, and the reflections of many software engineers on AI’s current and future impact on their careers—views expressed both in published articles and in our conversations with relevant communities.
Claims that AI is causing mass layoffs in the software industry appear to be a classic case of “AI washing”
Let’s look at three stories that made headlines and the contrast between them and reality:
-
In February, fintech company Block—the parent company of Cash App, Square, Afterpay, and other applications—announced that it would lay off 4,000 employees. According to founder Jack Dorsey, AI is “ushering in a new way of working,” bringing “smaller, flatter teams.” He specifically cited the improvements in model capabilities at the end of 2025.
But subsequent reporting revealed a very different picture. After more than tripling its headcount during the pandemic, the company was under enormous financial pressure. Naoko Takeda, a data scientist on the Cash App team, posted that Block was “forcing AI down everyone’s throat,” but that the “productivity gains were very limited” in her experience. She turned down a retention offer that included a 75% raise and resigned. Other employees interviewed also had very different views of Block’s AI capabilities—and of whether Dorsey truly understood the issues involved.
As Aaron Levie has pointed out, CEOs are particularly prone to developing misconceptions about AI’s uses because they can rapidly build prototypes without seeing the 90% of the work required to turn those prototypes into finished products. Dorsey’s public comments about AI appear to fit this pattern exactly.
-
In April, Snap laid off approximately 1,000 employees. In the layoff memo, CEO Evan Spiegel primarily cited AI as the reason for the cuts. He also said that AI had generated 65% of new code. In reality, however, the layoffs followed an activist investor’s campaign calling for cost reductions. (Snap has reported a net loss in every full fiscal year since its 2017 IPO, and its share price fell by more than 30% in 2026.) Notably, the specific pattern of the layoffs—for example, 150 positions across multiple roles in the augmented-reality division—does not match what we would expect from AI-driven layoffs. If AI were truly driving the cuts, we would expect broad reductions in programming and other “AI-exposed” roles, rather than cuts concentrated in a particular department.
-
In May, Intuit announced that it would lay off 3,000 employees while also entering into partnerships with Anthropic and OpenAI. The media connected the two events, describing the layoffs as an AI-driven reorganization and restructuring. But this time, the CEO denied the seemingly obvious narrative, saying that “this had nothing to do with AI.” The goal, he said, was to address “roles with heavy workloads” and excessive layers of management.
We did not cherry-pick these examples. In every story about AI-driven software engineering layoffs that we investigated, we found the same mismatch between the narrative and the facts. It turns out that “AI washing” layoffs is a phenomenon spanning the entire economy, as numerous surveys have shown:
-
59% of U.S. hiring managers admitted that they emphasize AI when explaining hiring freezes or layoffs because it is easier to justify to stakeholders than citing financial constraints.
-
J. P. Gownder, a principal analyst at Forrester, said of companies preparing for so-called AI-driven layoffs: “When we ask whether they have mature, proven AI applications ready to take over these jobs, nine times out of ten the answer is no—they haven’t even started.”
-
In a Harvard Business Review (HBR) survey of more than 1,000 executives worldwide, 21% of respondents said they had substantially reduced their headcount “in anticipation of” AI, while another 39% had made small or moderate workforce reductions in advance. By contrast, only 2% said they had substantially reduced their workforce because of AI that had actually been implemented. This tenfold difference suggests that executives are just as susceptible as anyone else to the misleading narrative that AI will replace workers.
Another interesting data point comes from the WARN Act. The law requires companies to disclose plant closures and mass layoffs affecting more than 100 employees. In March 2025, New York became the first U.S. state to add an AI disclosure option to its WARN Act filings. During the first full year, more than 160 companies filed WARN notices. Not a single one checked the AI option.^1 We contacted the New York State Department of Labor, which confirmed that, as of late May, only one company—Nespresso—had checked the option.^2