A coding-agent-heavy week, with one of the year's biggest funding signals.
Cursor reportedly raising at a $50B valuation
On April 17, reporting indicated Cursor (Anysphere) was in talks to raise at least $2B at a ~$50B valuation — nearly double its ~$29.3B from six months earlier — with a16z and Thrive returning and Nvidia expected as a strategic investor. Cursor had reportedly hit ~$2B annualized revenue and reached slight gross-margin profitability via its in-house Composer model. The application layer, not the model labs, is now commanding frontier-scale valuations.
Bolt simplifies its agent
On April 13, StackBlitz's Bolt.new retired its legacy v1 Agent for new projects, replacing individual model picks with two tiers — Standard and Max — and set Anthropic's Sonnet 4.6 as the default. It's a pattern you'll see repeatedly: app-layer tools abstracting the underlying model away so you choose an agent, not a model.
Model watch: Claude Opus 4.7
On the model layer, Anthropic released Claude Opus 4.7 on April 16 — stronger coding and higher-resolution vision at unchanged pricing — the kind of model these coding agents run on.
What it means if you're picking tools
The build-an-app and code-with-AI categories are consolidating fast. Compare current picks for coding and app & site builders.
Sources: Cursor funding talks (TechCrunch) · Bolt.new release notes