Snap announced on April 15 that it is cutting 1,000 jobs — a move that is not unusual in itself.
What is unusual is the reason the company gave. CEO Evan Spiegel said that AI has already generated more than 65% of Snap's new code, that an AI agent handles 1 million customer service requests per month, and that a code review agent has flagged 7,500 software bugs. Fewer people, but the same workload — completed faster.
This is the new language of tech layoffs in 2026: not "cost-cutting," but "efficiency gains."
What the 65% figure really means
Let's start with the logic behind that 65%.
Snap has not disclosed which specific tools it uses, but based on common industry practice, this includes code snippets generated by assistants such as Cursor, GitHub Copilot, and Claude Code, plus workflows for testing, code review, and bug fixes automated by AI agents. What does 65% mean? Put simply: if 100 engineers are writing code today, theoretically, 65 engineers plus AI tools could produce the same output in the same time — and the salaries of the remaining 35 could, in theory, be saved.
But reality is not that simple. Code written by AI still needs human review. Features generated by AI still need human-designed requirements. System architecture still requires human thinking. So Snap is not firing "all programmers." It is cutting positions that can be merged or eliminated after the widespread adoption of AI tools.
What is being cut is the execution layer. What remains is the judgment layer.
The underlying logic of this round of layoffs
The numbers tell the story more directly. Snap had 5,261 full-time employees as of last December. Now it is cutting 1,000 people and closing 300 open positions. After that, total headcount will be just over 4,000. Spiegel called this a "crucible moment" in an internal memo, saying the company must operate with "a new, faster, more efficient way of working" and "transition to profitable growth."
Translated: a project that used to take 10 people three months can now be done by five people plus AI tools in the same three months — so the other five positions are gone.
The stock market reacted immediately: Snap's share price rose more than 11% in advance (was it a leak, or market intuition? That is a good question). For investors, the company finally looks "smaller and faster."
More interestingly, activist investor Irenic Capital had previously suggested Snap cut 21% of its workforce, arguing that "AI can and should replace many existing roles." The scale of this round of layoffs almost matches Irenic's recommendation — 1,000 people is about 16%.
Snap is not alone
Since the start of 2026, more than 80 tech companies have conducted layoffs, cutting a total of over 71,000 jobs. Meta, Oracle, and Amazon are all on the list.
But this wave of layoffs is different from the one in 2022-2023: companies are no longer being coy about it.
Back then, companies generally used vague language like "uncertain market environment" or "strategic focus." Snap this time said it directly — AI writes 65% of the code, and we don't need as many people. This kind of candor was almost unheard of before.
The reason is simple: now, "AI-driven efficiency" is a positive narrative. Investors buy it, stock prices go up, and regulators are not causing trouble. Layoffs have gone from being bad news to a signal of efficiency upgrades.
Spiegel also noted that Snap is caught between two ends: giants with vast resources and nimble startups. Mid-sized companies face the most pressure — too big to be as flexible as a startup, too small to have the moat of a Meta. AI tools give these companies a chance to recompress costs and increase speed, but the prerequisite is daring to cut people first.
Where do programmers stand after 65%?
No one can give you a definitive answer, but the Snap case provides a clear data point: when 65% of routine coding work can be done by AI, the engineering headcount a company can sustain is roughly 50-60% of what it was before AI became widespread.
This is not "the end of programmers." It is a redefinition of the role. The engineers needed in the future are those who can collaborate efficiently with AI tools, do system design, and review AI-generated code — not those who mechanically translate requirements into code.
Also note that Snap emphasized more than just coding. The AI handles 1 million support requests per month, and the code review agent flagged 7,500 bugs — these are compressions in customer service and QA roles, not just development. This wave of AI-driven efficiency covers a broader range of job types than most people estimate.
Snap's layoffs give other companies still watching a reference point: what percentage of code can AI write, and can that proportionally reduce labor costs? This math problem now has a publicly available solution process.
More companies are likely to follow.
Sources: Snap is cutting 1,000 jobs, CocoLoop, 16% of its workforce (TechCrunch); Snap's stock jumps on plans to axe 16% of its workforce citing AI efficiencies (CNBC); Snap lays off 1,000 employees, or 16% of workforce, as AI takes over 65% of coding work (TechStartups); Snap Cutting 16% Of Full-Time Workforce; CEO Evan Spiegel Says AI Offers New Way Of Working (Deadline)