AI Wrote 65% of Their Code. Then Came the Layoffs.
TL;DR — Key Takeaways
- Claude Opus 4.7 ships with task budgets — AI agents now know when to stop, making agentic workflows cost-predictable for the first time.
- $297 billion raised in Q1 2026, with AI capturing 81% of all venture capital. OpenAI alone closed $122B — the largest private round in history.
- Snap cut 1,000 jobs after AI reached 65% of their code output, unlocking $500M in annualized savings.
Snap cut 1,000 people this week. AI now writes 65% of their code and saves them $500M a year.
That's not a warning for small businesses — it's a template. At your clients' scale, the same outcome doesn't cost $500M to achieve. It costs one microservice and about 60 days.
This week's AI brief covers that story, plus the new Anthropic model that finally lets you control what an AI agent costs to run, and the $297B quarter that proves the platforms you're building on aren't going anywhere.
At G8 Engineering, this is exactly what we build — workflows redesigned around AI, not just tools layered on top of them.
Story 01 — Claude Opus 4.7 introduces task budgets.
Anthropic shipped Claude Opus 4.7 on April 16 with one feature that changes everything for production AI agents: task budgets. You give Claude a token target for an entire agentic loop — thinking, tool calls, results, and output — and it calibrates how much effort to spend. Think of it like telling a contractor "you've got four hours" instead of handing them a blank check.
The model also delivers stronger multi-step task reliability, long-horizon reasoning, and better vision (up to 3.75 megapixels). Available on Claude, AWS Bedrock, Google Vertex AI, and Microsoft Foundry. Pricing unchanged from Opus 4.6.
The microservice angle: task budgets solve cost unpredictability — the biggest barrier to offering AI agents as a fixed-price service. Invoice processing, document review, report generation — workflows that were too risky to price just became sellable products.
Story 02 — $297 billion into AI in one quarter. 81 cents of every VC dollar.
Q1 2026 shattered every venture record ever set. OpenAI raised $122 billion — the largest private funding round in history — backed by Amazon, Nvidia, SoftBank, and Microsoft. Anthropic pulled in $30 billion. xAI raised $20 billion. Sovereign wealth funds are writing $10–30 billion checks because they've concluded AI infrastructure is as strategic as energy grids.
The capital isn't going to software subscriptions. It's funding chips, data centers, and the compute that runs the models your microservices depend on.
The microservice angle: when nations treat AI infrastructure like power grids, the platforms aren't going anywhere. Build on Claude, GPT, and Gemini with the same confidence you'd build on AWS. The infrastructure bet has been made — at $297B in a single quarter, the only question left is what you build on top of it.
Story 03 — Snap let go of 1,000 people. AI now writes 65% of their code.
Snap's CEO announced roughly 1,000 layoffs this week, directly citing rapid AI advancement. AI now generates more than 65% of Snap's new code, with the company projecting over $500 million in annualized cost savings. This isn't a startup experiment — it's a public company restructuring its engineering org around AI output.
The microservice angle: what Snap did at enterprise scale, your SMB clients need to do at their scale. A microservice that automates a process once handled by two full-time employees doesn't cost $500M to build. It might cost $15,000. That math is your sales pitch — written by Snap's earnings call.
The bottom line: Snap's story is a template. AI writing 65% of their code and saving $500M only makes headlines at Snap's scale. At your clients' scale, it looks like one microservice that eliminates a manual process, saves 20 hours a week, and pays for itself in 60 days.
At G8 Engineering, we build AI-powered microservices that help small and mid-size companies close that gap — not with another subscription, but with a workflow rebuilt from the ground up.