Snap filed an 8-K with the SEC on April 15 disclosing that over 65% of its new code is now generated by AI. The same filing announced 1,000 layoffs — 16% of full-time staff — and projected $500 million in annualized savings. Shares rose 5.8%.
That 65% number matters more than the layoffs. We’ve tracked the demand side of software’s repricing — seats shrinking, multiples compressing, budgets migrating. But the hypothesis starts with a supply-side claim: AI can create software fast and efficiently. Until now, the evidence for that was mostly anecdotal. Snap just put a number on it in a regulatory filing.
The Trend Line Takes Shape
Snap isn’t an outlier. Google’s AI-generated code share has followed a clear trajectory: roughly 25% in early 2025, over 30% by mid-2025, and above 50% in production check-ins by early 2026. Sundar Pichai has framed this as velocity — a 10% increase in engineering throughput — rather than headcount reduction.
That framing is instructive. Google isn’t cutting engineers. Snap is cutting 16% of its workforce. Same underlying technology, radically different strategic conclusions. The difference: Google is a platform with a services budget measured in tens of billions. Snap is a mid-tier social app with an activist investor demanding profitability.
This is where the code generation numbers connect to the repricing thesis.
From Productivity Metric to Cost Structure
When Spiegel told employees that “rapid advancements in artificial intelligence enable our teams to reduce repetitive work,” he was describing the same thing Pichai described — but extracting a different conclusion. Google uses AI code generation to build more. Snap uses it to build the same with fewer people.
For a company that sells software (or in Snap’s case, a software-powered ad platform), the implications cascade. If AI generates 65% of new code and you cut 16% of staff while maintaining output, your cost of software production has structurally declined. That’s the mechanism this blog’s hypothesis rests on, now with a number attached.
The broader data supports this. According to Layoffs.fyi, 80 tech companies have cut roughly 71,440 jobs so far in 2026. Over 61,000 of those cuts across all industries have been explicitly attributed to AI. Atlassian cut 1,600 (10%) in March, citing the “AI era.” Block cut from 10,000 to under 6,000. Snap’s investor update also notes that AI agents now handle over 1 million support questions per month internally — a detail that echoes the resolution-pricing story from customer support vendors, now applied to the company’s own operations.
The Recursive Problem
Here’s where it gets uncomfortable for SaaS. If the companies that buy software are shrinking their teams (Block, Snap, Atlassian), that’s fewer seats. We’ve covered that. But if the companies that build software also need fewer people to produce it, the cost of creating competitive alternatives drops.
Google can afford to keep its engineers and build more. A three-person startup with the same AI tooling can now produce code at a rate that used to require a team of ten. The marginal cost of software production isn’t just declining inside large tech companies — it’s declining everywhere, simultaneously. That’s the commoditization mechanism the hypothesis describes, and the 65% figure is the clearest quantitative evidence yet.
AJ Bell investment director Russ Mould offered the right caveat: “Cutting costs may appease an activist in the near term… but whether it really leaves the company with a defensible business model and competitive position is still unclear.” The market rewarded Snap’s cuts. It hasn’t yet asked whether a company that needs 65% less engineering labor to write code can defend itself against competitors who also need 65% less.
Hypothesis Update
The supply-side mechanism — AI reduces the marginal cost of software production — now has its first quantitative trend line: Google at 50%+, Snap at 65%, and rising. This verifies one leg of the hypothesis directly. The qualification remains: production cost is only one component of a software business. Distribution, data gravity, and switching costs haven’t been commoditized. But the production floor is dropping fast, and it’s dropping into SEC filings where it can be tracked quarter by quarter.
What to watch: Microsoft’s Q3 FY26 earnings on April 29. If they disclose a Copilot-assisted code generation percentage for internal development, the trend line becomes a consensus metric.
Sources: Snap Inc. 8-K filing and CEO memo, April 15, 2026; SF Chronicle; Economic Times; Google AI code data via ITPro and Roo Code; Layoffs.fyi; Programs.com AI layoff tracker.