Snap just laid off roughly one thousand people — about sixteen percent of its entire workforce. The stock went up nine percent the day they announced it.
[STAT CARD: "1,000 laid off · 16% of workforce"] [B-ROLL: company-logo:snap]And buried inside the announcement was a number that I think tells you more about the state of the tech industry than the layoff count itself: sixty-five percent of Snap's new code is now being written by AI.
[STAT CARD: "65% of new code written by AI"]Let that sit for a second.
[CUT] [TALKING HEAD — transition]Not assisted by AI. Not reviewed by AI. Written by AI. Nearly two-thirds of all new code produced at Snap is generated by artificial intelligence tools, not by software engineers.
[B-ROLL: code-terminal]This is not a story about Snap specifically. This is a story about what is happening to software engineering as a profession — and it's happening faster than most people in the industry will admit publicly.
[B-ROLL: ai-abstract]Let me give you the full picture.
[VOICEOVER — scene 2] [B-ROLL: news-studio]Snap announced the layoffs in April 2026. Approximately one thousand employees across engineering, product, and corporate functions. The company expects to take between ninety-five million and one hundred thirty million dollars in charges related to the layoffs in Q2. Against that, the restructuring is expected to reduce annualized operating costs by more than five hundred million dollars.
[STAT CARD: "$95M–$130M Q2 charges"] [STAT CARD: "$500M+ annualized savings"] [B-ROLL: finance-charts]Five hundred million dollars in annual savings. From eliminating sixteen percent of the workforce.
[B-ROLL: screen-capture:cursor]And the company's AI agent infrastructure — systems that respond to user and operator queries automatically — is now handling more than one million queries per month.
[STAT CARD: "1M+ AI agent queries / month"] [B-ROLL: finance-charts]The market's reaction tells you everything. Snap stock up nine percent. Investors don't see a company in distress. They see a company that figured out how to do the same amount of work — or more work — with fewer people. And they're rewarding it.
[STAT CARD: "SNAP +9% on announcement"] [/VOICEOVER] [CUT] [TALKING HEAD — transition]Now, there's nuance here that matters.
[VOICEOVER — scene 3] [B-ROLL: screen-capture:github]The sixty-five percent figure applies to new code. Maintenance of legacy systems, architecture decisions, code review, security auditing, product judgment — these still require humans. AI doesn't replace the engineering function wholesale. What it does is dramatically compress the time from idea to working software for the parts of the job that involve writing code to a spec.
[B-ROLL: code-terminal]But here's the thing. Writing code to a spec is what a lot of software engineers spend most of their time doing. Especially early and mid-career engineers. The function that AI is absorbing is exactly the function that has historically been the entry point into the software industry.
[B-ROLL: stills:office-empty]This matters for how we think about AI and employment — not as a future hypothetical, but as a present reality.
[B-ROLL: ai-abstract]The trajectory of the tech industry for the last decade was: you can always hire more engineers. The bottleneck was people who could code. If you had the people, you could ship the product. Companies scaled their engineering headcount aggressively because headcount was the constraint.
[B-ROLL: screen-capture:cursor]Snap's restructuring is an early, visible signal that this model is changing. The constraint is no longer people who can code. It's people who can direct AI that codes. That's a smaller number, and it requires different skills.
[/VOICEOVER] [CUT] [TALKING HEAD — transition]What does this mean for someone currently in software engineering, or thinking about going into it?
[VOICEOVER — scene 4] [B-ROLL: finance-charts]First: the shift is real and it's accelerating. Sixty-five percent is not a rounding error or a pilot program. That is the operating baseline at a major technology company right now.
[B-ROLL: code-terminal]Second: the skills that remain valuable are not writing code — they're everything around code. System design. Security. Debugging complex failures. Product judgment. The ability to evaluate whether AI-generated code is actually correct, not just whether it compiles.
[B-ROLL: stills:office-empty]Third: companies are restructuring now, while they can justify it as efficiency rather than panic. The layoffs happening in 2026 are happening in a positive market environment. That matters — it means these aren't distress cuts, they're structural realignments. Those don't reverse when the cycle turns.
[B-ROLL: company-logo:snap]The Snap number is significant not because Snap is exceptional. It's significant because Snap is reporting what a growing number of technology companies are discovering and not yet disclosing publicly.
[/VOICEOVER] [CUT] [TALKING HEAD — sign-off]Sixty-five percent at Snap today is a leading indicator. Pay attention to it.
Stay sharp. — Jane Sterling, Sterling Intelligence
=== ARTICLE_HTML ===Snap just announced the layoff of approximately 1,000 employees — roughly 16% of its total workforce. The stock jumped 9% on the news. And in the same announcement, the company disclosed that AI now generates 65% of all new code written at Snap.
In this video, Jane Sterling breaks down what these numbers mean, why the market rewarded Snap for the cuts, what the 65% AI code figure actually tells us about software engineering, and what this moment means for tech workers and the industry at large.
Snap's April 2026 restructuring affected approximately 1,000 employees across engineering, product, and corporate functions — about 16% of the company's total headcount.
The financial math: Snap expects $95M–$130M in restructuring charges in Q2 2026. Against that, the company projects more than $500M in annualized cost savings from the reduced headcount.
The stock reaction: shares up approximately 9% on the announcement day.
That stock move tells you how investors interpreted the news. This wasn't read as a company in distress cutting costs to survive. It was read as a company executing a deliberate efficiency transformation — doing more with fewer people — and the market liked what it saw.
Buried in Snap's announcement is a figure that deserves more attention than it received: 65% of Snap's new code is now being written by AI.
To be precise about what this means: new code, not legacy maintenance. The engineering function at Snap has not been replaced. What has happened is that the production of new software features — writing code to implement specified functionality — has shifted dramatically toward AI-generated output.
AI agent systems at Snap are now handling more than 1 million queries per month. The automation footprint is real and growing.
The 65% number is significant for several reasons. It is a disclosed figure from a public company, not an estimate or a projection. It represents current operations, not a future roadmap. And it is a number that other technology companies have not yet publicly disclosed but that many are privately experiencing at comparable or higher rates.
From a pure financial perspective, Snap's restructuring math is straightforward.
Labor is the largest cost for software companies. Reducing engineering headcount by 16% while maintaining or growing output through AI tooling produces the kind of cost efficiency that drives operating leverage. Every dollar not spent on salary that produces equal or greater output in software falls directly to operating margin.
The $500M annualized savings figure is not a projection based on hoped-for efficiency gains. It reflects headcount that has already been eliminated. The savings are structural.
Investors are also reading the AI productivity numbers as evidence that Snap's engineering function is operating at higher leverage than it was before. If 65% of code is being generated by AI, then the remaining human engineers are — in principle — focused on higher-value, harder-to-automate work: architecture, security, complex debugging, product judgment.
That's a more efficient organization chart, at least in theory. And markets price for efficiency.
This is the part of the Snap story that matters most for anyone in or adjacent to the software industry.
Software engineering has been one of the most stable and lucrative career paths in the United States for thirty years. The supply of people who could write code has never caught up with demand. Bootcamps, computer science programs, and self-taught developers all fed into a labor market where skilled engineers commanded significant compensation and had significant leverage.
The 65% figure at Snap is evidence that this dynamic is changing in real time.
The parts of software engineering being absorbed by AI are specifically the parts that have historically been the entry point into the profession: writing code to a specification, implementing features from design documents, producing working implementations of defined functionality.
These tasks are learnable, documentable, and pattern-rich. They are exactly the tasks that large language models with code training are best at. And they have historically consumed the majority of early and mid-career engineering time.
What remains after AI absorbs code generation?
System architecture — designing how components interact, where failure modes are, how to build for scale and reliability. This requires deep experience and judgment that AI doesn't yet reliably provide.
Security engineering — finding vulnerabilities, understanding attack surfaces, designing systems that are defensible. The Mythos story from Anthropic is relevant here: AI is getting better at offensive security, which means defensive security judgment remains premium.
Complex debugging — when production systems fail in unexpected ways, diagnosing root causes in distributed systems requires the kind of lateral reasoning and systems intuition that AI augments but doesn't replace.
Product judgment — deciding what to build, not just how to build it. The translation between human needs and technical specification remains a deeply human skill.
Code review and evaluation — if AI generates 65% of code, humans need to verify that it's correct, secure, and maintainable. That's a skill set in its own right.
The engineering function isn't disappearing. It's reshaping. The headcount requirement for a given amount of software output is declining. The skill requirements for the remaining headcount are shifting upward.
Snap is one data point. But it's a public, disclosed data point from a major technology company — and that makes it significant beyond its specific numbers.
Technology companies operate in an industry where copying successful practices is fast and highly incentivized. If Snap is producing significant cost savings while maintaining output by deploying AI code generation at scale, the playbook is visible to every other technology company's leadership.
The restructurings happening across tech in 2026 are not primarily about economic distress. They are happening during a period of rising stock prices, strong AI investment, and positive market sentiment. These are deliberate transformations, not emergency measures.
That distinction matters. Cost cuts in a downturn are often reversed when conditions improve. Structural transformations toward AI-driven efficiency have a different character — they reflect a new operating model that companies have every incentive to maintain and extend.
The trajectory from here is not ambiguous. AI code generation tools are getting better. The percentage of code generated by AI at Snap and companies like it is not going to go down.
If you're in software engineering: the 65% number is a planning input, not a reason to panic. The skills that AI can't yet provide — architecture, security, judgment, debugging — are learnable. But the time to build them is now, not when the restructuring hits your organization.
If you're managing engineers: the productivity math has changed. The ratio of output to headcount that was standard eighteen months ago is not the ratio that's achievable now. If your team hasn't materially increased AI tool adoption, you're operating at a structural disadvantage.
If you're thinking about entering software engineering: the entry-level market for code-to-spec work is contracting. The path into the profession now runs through demonstrating judgment and systems thinking earlier than it used to, not just coding proficiency.
If you're an investor or business operator: Snap's restructuring math is replicable. The cost savings from AI-driven productivity in software development are real and material. Any business with significant software development costs should be actively evaluating this.
The 65% figure is the number that will be cited in five years as an early indicator that people either saw coming or didn't.
Subscribe to Sterling Intelligence for weekly coverage of what's actually happening in AI.
New videos every week.
— Jane Sterling
Some links may be affiliate links. Commission at no cost to you.
=== YOUTUBE_DESC === Snap just fired 1,000 people. And AI now writes 65% of their new code. The stock went up 9% on the news — because that's what the market wants to see. In April 2026, Snap announced a restructuring that cut approximately 16% of its workforce — roughly 1,000 employees across engineering, product, and corporate functions. The company expects $95M–$130M in Q2 restructuring charges. Against that: more than $500M in annualized savings. The stock rallied 9% on the announcement. Buried in the same disclosure is a number that deserves more attention than it got: 65% of Snap's new code is now being written by AI. Not assisted. Not reviewed. Written. Snap's AI agent infrastructure is also handling more than 1 million queries per month. In this episode, Jane Sterling breaks down what these numbers actually mean — why Wall Street rewarded the layoffs, what the 65% figure tells you about the future of software engineering, and why this moment matters for anyone whose career touches code. Key numbers covered: • ~1,000 employees laid off (roughly 16% of Snap's workforce) • $95M–$130M in Q2 restructuring charges • $500M+ in projected annualized savings • SNAP stock +9% on announcement • 65% of new code now AI-generated • 1M+ AI agent queries per month We cover the structural shift from "hire more engineers" to "direct AI that codes," what skills remain premium (architecture, security, debugging, product judgment, code review), why these restructurings happening in a positive market are structural and not cyclical, and what tech workers, managers, CS students, and operators should each take from the Snap disclosure. ⏱ Chapters 00:00 The headline under the headline 01:00 The layoff math and Wall Street reaction 03:00 What 65% AI-written code actually means 05:00 Why this restructuring is structural, not cyclical 06:30 What remains valuable in software engineering 08:00 What you should do with this information 🔔 Subscribe to Sterling Intelligence for weekly breakdowns of what's actually happening in AI — no hype, no filler, just the signal. https://www.youtube.com/@SterlingIntelligence — Jane Sterling, Sterling Intelligence #SnapLayoffs #AICode #TechLayoffs2026 #AIJobs #SoftwareEngineering #SnapAI #AIProductivity #FutureOfWork #EvanSpiegel #AIAgents #SterlingIntelligence #JaneSterling #AIWeekly #ArtificialIntelligence #TechNews2026 === TITLES_HTML ===Expression. Serious, slightly concerned — the face you make when you read a number and realize what it means for your industry. Mouth closed, subtle tension at the jaw. No smile, no shock.
Head position. Squared to camera, micro-lean forward. Chin neutral, eye line level. Authority without theatrics.
Wardrobe. Dark blazer, minimalist. No reflective jewelry. Sterling Intelligence palette — black, charcoal, gold accent only.
Eye direction. Direct to camera, locked. Alternate take: eyes cut sharply right toward the "65%" overlay.
Lighting. Key light from upper-left at ~4800K, soft fill on the right at 25%. Deep shadow on the left jaw line for drama. Rim light behind-right to lift her off the background.
Scene setup. Near-black charcoal background with a faint yellow-Snap gradient in the upper-right (subtle brand nod, not loud). Shallow depth of field — Jane tack-sharp, ghosted code lines at ~12% opacity behind her shoulder.
Position. Right third of the frame, stacked scoreboard — "1,000 FIRED" top row, "65% AI CODE" bottom row.
Font. JetBrains Mono Bold for the numbers (reads as data); Inter Black for the labels.
Color scheme. "1,000" in pure white with a faint red (#dc2626) underglow. "FIRED" in red, Inter Black. "65%" in gold (#c8a84b) at 115% scale. "AI CODE" in white. 3px black stroke on every character for legibility.
Accent detail. Small-caps header above: "SNAP — APRIL 2026" in 11px gold. Frames it as a dated, specific event rather than a clickbait claim.
Position. Lower-left third, two lines. "FIRED 1,000." on top. "STOCK +9%." below, slightly larger.
Font. Bebas Neue Bold or Impact, condensed all-caps, tight tracking.
Color scheme. "FIRED 1,000." in white. "STOCK +9%." in bright green (#16a34a) at 120% scale. 3px black stroke throughout. Faint outer glow on "+9%" so it pops against the dark background.
Accent detail. Gold sub-tag below: "SNAP — 65% of code now AI" in Inter Bold 16px, #c8a84b. Backs the paradox with the causal stat.
Position. Centered upper band, then Jane's face dominant lower two-thirds.
Font. Inter Black all caps, wide tracking (~120), stretched across the frame.
Color scheme. Base text in white, but "THE CODE" overlaid with a transparent glassy gold (#c8a84b at 80%) to visually separate. 2px black stroke.
Accent detail. Red underline under "THE CODE" at 4px. Smaller gold subtitle below: "SNAP CUT 1,000 JOBS — 65% AI" in Inter Bold 18px. Frames the story as industry-wide, not company-specific.