The $6.6 Billion Question Anthropic Just Asked Lovable
Anthropic spent two years pointing to Lovable as proof that Claude was the best AI for coding. On April 17, it shipped the product that makes that proof redundant — and Lovable fired back the same day with a retention offer that reads as a defensive acknowledgment that the threat is real.
Lovable, a $6.6 billion coding assistant that runs entirely on Anthropic's API, announced double credits for its users through April 30 — an offer that went live the same day Anthropic's new prototyping tool shipped. The sequence reads as a company scrambling to keep users from noticing they can get the same thing from the subscription they already pay for. Anthropic declined to comment.
The trap is structural. Claude Design bundles into every existing Claude plan — Pro, Max, Team, and Enterprise — at no additional cost beyond overage charges. It combines a live preview feature called Artifacts, a code scaffolding system called Claude Code, and a project management layer — the same workflow Lovable sells as a standalone product to 1 million monthly active users. Lovable is now valued at $6.6 billion with $400 million in annualized recurring revenue, per Business Insider (March 2026). Anthropic had cited Lovable's reliance on Claude as the primary evidence that its models were the industry's best for coding.
Lovable had already integrated Anthropic's Opus 4.7 model into its platform in March, cementing its dependence on the same engine powering its new competitor. Cursor, another AI coding tool, recognized this dynamic earlier and reduced its Anthropic dependency by integrating Kimi, an open-source model from Chinese startup Moonshot. Lovable has no equivalent fallback — its product is Anthropic all the way down, a dependency that Anthropic is now competing directly across.
What Lovable did not build is independence from the lab that controls both its supply chain and, as of April 17, its competitive ceiling. The double credits expire April 30. Whether the users stay is the question that determines whether this was damage control or a turning point.