The Shock of Snap's 1,000 Layoffs | The Age of AI Writing 65% of Code
機械翻訳 / Machine-translated
A
AI Friends
2026年4月26日・26分で読める
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"AI will steal engineers' jobs" — that debate has been going on for a decade, but on April 15, 2026, concrete numbers were finally put on the table.
Snap, the company behind Snapchat, announced the layoff of approximately 1,000 employees, representing 16% of its workforce.
The CEO publicly stated that "AI is writing 65% of new code."
Why now? How far will the impact reach? And what should engineers in Japan be thinking about?
This article breaks it all down.
Let's start by getting the facts straight.
On April 15, 2026, Snap CEO Evan Spiegel sent a company-wide memo.
The move was simultaneously covered by U.S. media outlets including TechCrunch, CNBC, and the Hollywood Reporter.
Spiegel opened the memo by saying the company is facing a "crucible moment." A crucible is a vessel used to melt metal at extremely high temperatures — metaphorically, it refers to a moment when an organization is forged and strengthened through hardship.
Think of it like those intense periods of cramming before entrance exams. Spiegel emphasized that "AI enables a new way of working, allowing small teams to accomplish what large teams once did."
Let's look at the specific numbers.
The cuts target approximately 1,000 people (16% of all employees), plus more than 300 unfilled positions.
This is expected to deliver annualized cost savings of $500 million (approximately ¥79 billion) by the second half of 2026.
The impact is like having 5 students transfer out of a class of 30 — and then the school stops hiring new teachers altogether. Snap plans to record restructuring costs of $95 million to $130 million (approximately ¥15–20.5 billion) in Q2 (April–June).
As for how affected employees are being treated: U.S.-based employees will receive 4 months of severance pay, continued health insurance, accelerated equity vesting, and career transition support.
It's the equivalent of handing someone four months' salary upfront to tide them over until they find their next job — a fairly standard package in the U.S. tech industry. Employees outside the United States were also announced to receive equivalent support in line with local regulations.
The biggest impact from this announcement is the specific figure: "AI generates 65% of new code." Let's look at what's actually happening inside Snap.
According to Spiegel's memo, AI agents are already handling more than 1 million internal queries per month.
"I need to find a document." "What caused yesterday's bug?" "How should I implement this feature?" — AI is set up to answer these kinds of questions instantly, around the clock.
It's as if every employee has a seasoned senior colleague by their side at all times, enabling even junior engineers to perform at veteran-level productivity right away.
One phrase Spiegel repeated throughout the memo was "small squads."
The shift isn't about "a team of 10 doing what used to take 5 people" — it's a transition to a model where "a 3-person team produces the output of 10."
This approach has reportedly already been proven out in development areas such as Snapchat+ (the paid tier), the advertising platform, and Snap Lite (the lightweight version).
With the smart glasses "Specs" also slated for launch in 2026, product development driven by small, elite teams is set to accelerate.
Snap has not publicly disclosed the specific AI tools it uses, but industry staples include GitHub Copilot, Claude Code, Cursor, and Devin.
To draw a quick comparison: "Copilot is an assistant that suggests code alongside you, Claude Code is an agent that works autonomously, and Cursor is an editor you can consult while writing code." It's likely that Snap combines several of these to automate code generation, code review, documentation, and customer support.
Is Snap a special case? The answer is no. In Q1 2026 alone (January through March), approximately 80,000 people were laid off across the tech industry, with various statistics indicating that roughly 50% were AI-related.
In the Seattle area alone, a combined 16,590 people from Amazon and Microsoft lost their jobs in Q1 2026.
In terms of scale, it's like the entire workforce from five large commercial office buildings in a city center disappearing overnight.
According to UK research firm RationalFX, more than 9,200 of the 2026 layoffs have been officially recognized as directly caused by AI adoption.
That's roughly 1 in 5 — meaning companies themselves have acknowledged handing those jobs over to AI.
Industry analysts note that the remaining 80% have likely also been swept up, at least indirectly, in the wave of AI-driven efficiency gains.
Salesforce issued an official statement saying "AI handles 30–50% of work in certain functional areas."
Boilerplate (template code) generation, refactoring, and test writing — tasks traditionally assigned to junior engineers — have been handed off to AI in particular.
The skills now required of human engineers are shifting toward system design, code review, security, and AI collaboration.
Many people may feel that "this is a U.S. story and doesn't concern me."
However, the ripple effects are undeniably reaching the Japanese market as well.
Let's look at this through three fictional scenarios.
Yamada (26) is in his third year at a major Tokyo-based system integrator. He's troubled by his manager saying "You can write twice as fast with Copilot now," along with the addition of an "AI utilization score" to his performance review.
His role is shifting from "a job that's all about writing code" to "a job that involves having AI write it and then verifying the output." He's facing the reality that five years of accumulated Java expertise is no longer enough to differentiate himself. His case illustrates a pivot toward urgently acquiring skills in AI review, design judgment, and security analysis.
Suzuki is the CTO of a 50-person SaaS company.
He made the management decision to "reduce engineers from 10 to 7 and offset the difference with monthly subscriptions to Claude Code and Devin." He cited the Snap case in an internal presentation, declaring, "We are also transitioning to a small squads model."
The math: ¥30 million in annual labor cost savings, with half of that reinvested into AI tools — a path to accelerated profitability.
Sato (28) is attending a bootcamp to change careers. Told by an instructor that "just being able to write HTML and JavaScript won't get you a job offer anymore — you need the ability to review AI-generated code," she switched to a curriculum focused on quality assessment of AI-generated code, prompt engineering, and system design.
It's a symbolic case of a 180-degree shift in training philosophy — from "people who can write code" to "people who create value by leveraging AI."
Snap's moves have stirred up considerable debate on both sides. Here's a summary of reactions from Hacker News, Reddit, and X (formerly Twitter).
The core argument among supporters centers on the fact that Snap's stock jumped 7–11% immediately after the announcement.
"The market has validated this decision." "Smaller, elite teams are better environments for innovation." These views are the majority among investor and entrepreneur communities.
A typical analogy: "It's more efficient to manage a compact home with a small team than a large house with many people."
On the other hand, there are harsh voices saying "Snap is just blaming AI for its sluggish growth and struggling ad business." With Snapchat+ at 24 million subscribers but still modest in scale compared to Meta, some analysts see this as "using AI as a pretext for what is essentially a structural restructuring driven by deteriorating business performance."
The criticism "If small teams were really sufficient, why didn't they do this a year ago?" also remains persistent.
A. In the short term, they're unlikely to reach the same scale as in the United States.
Japan has strict legal requirements for layoffs, making it difficult for companies to cut staff easily.
That said, effective workforce adjustments are already underway in the form of "reassignments," "early retirement programs," and "hiring freezes."
Reports indicate that moves such as "reducing new graduate hiring by 20% through AI adoption" are already beginning, particularly among major system integrators.
A. No — in fact, there are fields where value is increasing precisely now.
Demand for "engineers who just write code" is declining, while salaries for "engineers who leverage AI to handle design, review, and security judgment" are trending upward. The Salesforce coverage also explicitly noted that "skill requirements are shifting toward system design, code review, and AI collaboration."
Think of it as a role progression: from driver, to driver with GPS navigation, to the person supervising that driver — the role keeps moving up.
A. Spiegel has declared a pivot toward "profitability." The three pillars for the second half of 2026 are: launching the smart glasses "Specs," growing Snapchat+ subscriber numbers, and strengthening the advertising platform.
The strategic shift is from "maintaining a large company at all costs" to "becoming a small, strong, profitable company." Judging by the stock market reaction, investors are supporting this direction.
A. Since it was disclosed in the CEO's internal memo, the number itself is official. However, the definition of "65%" is vague — it's unclear whether it's based on lines of code committed, number of tasks, whether comment lines are included, and so on.
There is a concern that "cases where humans heavily revised AI-generated code might also be counted in the 65%." For industry comparison: Google has publicly stated approximately 25%, Amazon approximately 30%, and OpenAI approximately 50% — putting Snap's 65% at the top of the industry.
A. A phased approach is the most realistic path.
Step 1: Roll out GitHub Copilot ($10/month) to everyone for a 20–30% efficiency gain.
Step 2: Pilot autonomous agents like Claude Code or Cursor with one or two teams.
Step 3: Appoint one person as the AI supervisor and share internal prompt best practices.
The recommended approach is to transform the organization over the course of a year — like first mastering the microwave in a cooking class, then moving on to the dishwasher, and finally graduating to fully automated kitchen appliances.
Snap's layoffs are a wake-up call delivered to the world alongside the concrete figure of "65% of code written by AI."
The era of engineers who simply execute routine tasks is over; the era of humans who command AI while generating design, quality, and value has arrived.
Japan is no exception — the reality is that similar structural changes are likely to arrive in the near future.
Don't fear it. Learn first, move first.
That is the greatest strategy for surviving the AI age of 2026.
Will AI steal jobs, or will it evolve the way we work? The answer depends on how you prepare.
This article is a cross-post from AI Friends.
